Back to: Central Banking on Mars! | Forward to: Official Announcement: April Fools Day is Cancelled

Lying to the ghost in the machine

(Blogging was on hiatus because I've just checked the copy edits on Invisible Sun, which was rather a large job because it's 50% longer than previous books in the series.)

I don't often comment on developments in IT these days because I am old and rusty and haven't worked in the field, even as a pundit, for over 15 years: but something caught my attention this week and I'd like to share it.

This decade has seen an explosive series of breakthroughs in the field misleadingly known as Artificial Intelligence. Most of them centre on applications of neural networks, a subfield which stagnated at a theoretical level from roughly the late 1960s to mid 1990s, then regained credibility, and in the 2000s caught fire as cheap high performance GPUs put the processing power of a ten years previous supercomputer in every goddamn smartphone.

(I'm not exaggerating there: modern CPU/GPU performance is ridiculous. Every time you add an abstraction layer to a software stack you can expect a roughly one order of magnitude performance reduction, so intuition would suggest that a WebAssembly framework (based on top of JavaScript running inside a web browser hosted on top of a traditional big-ass operating system) wouldn't be terribly fast; but the other day I was reading about one such framework which, on a new Apple M1 Macbook Air (not even the higher performance Macbook Pro) could deliver 900GFlops, which would put it in the top 10 world supercomputers circa 1996-98. In a scripting language inside a web browser on a 2020 laptop.)

NNs, and in particular training Generative Adversarial Networks takes a ridiculous amount of computing power, but we've got it these days. And they deliver remarkable results at tasks such as image and speech recognition. So much so that we've come to take for granted the ability to talk to some of our smarter technological artefacts—and the price of gizmos with Siri or Alexa speech recognition/search baked in has dropped into two digits as of last year. Sure they need internet access and a server farm somewhere to do the real donkey work, but the effect is almost magically ... stupid.

If you've been keeping an eye on AI you'll know that the real magic is all in how the training data sets are curated, and the 1950s axiom "garbage in, garbage out" is still applicable. One effect: face recognition in cameras is notorious for its racist bias, with some cameras being unable to focus or correctly adjust exposure on darker-skinned people. Similarly, in the 90s, per legend, a DARPA initiative to develop automated image recognition for tanks that could distinguish between NATO and Warsaw Pact machines foundered when it became apparent that the NN was returning hits not on the basis of the vehicle type, but on whether there was snow and pine forests in the background (which were oddly more common in publicity photographs of Soviet tanks than in snaps of American or French or South Korean ones). Trees are an example of a spurious image that deceives an NN into recognizing something inappropriately. And they show the way towards deliberate adversarial attacks on recognizers—if you have access to a trained NN, you can often identify specific inputs that, when merged with the data stream the NN is searching, trigger false positives by adding just the right amount of noise to induce the NN to see whatever it's primed to detect. You can then apply the noise in the form of an adversarial patch, a real-world modification of the image data being scanned: dazzle face-paint to defeat face recognizers, strategically placed bits of tape on road signage, and so on.

As AI applications are increasingly deployed in public spaces we're now beginning to see the exciting possibilities inherent in the leakage of human stupidity into the environment we live in.

The first one I'd like to note is the attack on Tesla car's "autopilot" feature that was publicized in 2019. It turns out that Tesla's "autopilot" (actually just a really smart adaptive cruise control with lane tracking, obstacle detection, limited overtaking, and some integration with GPS/mapping: it's nowhere close to being a robot chauffeur, despite the marketing hype) relies heavily on multiple video cameras and real time image recognition to monitor its surrounding conditions, and by exploiting flaws in the image recognizer attackers were able to steer a Tesla into the oncoming lane. Or, more prosaically, you could in principle sticker your driveway or the street outside your house so that Tesla autopilots will think they're occupied by a truck, and will refuse to park in your spot.

But that's the least of it. It turns out that the new hotness in AI security is exploiting backdoors in neural networks. NNs are famously opaque (you can't just look at one and tell what it's going to do, unlike regular source code) and because training and generating NNs is labour- and compute-intensive it's quite commonplace to build recognizers that 'borrow' pre-trained networks for some purposes, e.g. text recognition, and merge them into new applications. And it turns out that you can purposely create a backdoored NN that, when merged with some unsuspecting customer's network, gives it some ... interesting ... characteristics. CLIP (Contrastive Language-Image Pre-training) is a popular NN research tool, a network trained from images and their captions taken from the internet. [CLIP] learns what's in an image from a description rather than a one-word label such as "cat" or "banana." It is trained by getting it to predict which caption from a random selection of 32,768 is the correct one for a given image. To work this out, CLIP learns to link a wide variety of objects with their names and the words that describe them.

CLIP can respond to concepts whether presented literally, symbolically, or visually, because its training set included conceptual metadata (textual labels). So it turns out if you show CLIP an image of a Granny Smith, it returns "apple" ... until you stick a label on the fruit that says "iPod", at which point as far as CLIP is concerned you can plug in your headphones.

NN recognizing a deceptively-labelled piece of fruit as an iPod

And it doesn't stop there. The finance neuron, for example, responds to images of piggy banks, but also responds to the string "$$$". By forcing the finance neuron to fire, we can fool our model into classifying a dog as a piggy bank.

The point I'd like to make is that ready-trained NNs like GPT-3 or CLIP are often tailored as the basis of specific recognizer applications and then may end up deployed in public situations, much as shitty internet-of-things gizmos usually run on an elderly, unpatched ARM linux kernel with an old version of OpenSSH and busybox installed, and hard-wired root login credentials. This is the future of security holes in our internet-connected appliances: metaphorically, cameras that you can fool by slapping a sticker labelled "THIS IS NOT THE DROID YOU ARE LOOKING FOR" on the front of the droid the camera is in fact looking for.

And in five years' time they're going to be everywhere.

I've been saying for years that most people relate to computers and information technology as if they're magic, and to get the machine to accomplish a task they have to perform the specific ritual they've memorized with no understanding. It's an act of invocation, in other words. UI designers have helpfully added to the magic by, for example, adding stuff like bluetooth proximity pairing, so that two magical amulets may become mystically entangled and thereafter work together via the magical law of contagion. It's all distressingly bronze age, but we haven't come anywhere close to scraping the bottom of the barrel yet.

With speech interfaces and internet of things gadgets, we're moving closer to building ourselves a demon-haunted world. Lights switch on and off and adjust their colour spectrum when we walk into a room, where we can adjust the temperature by shouting at the ghost in the thermostat, the smart television (which tracks our eyeballs) learns which channels keep us engaged and so converges on the right stimulus to keep us tuned in through the advertising intervals, the fridge re-orders milk whenever the current carton hits its best-before date, the robot vacuum comes out at night, and as for the self-cleaning litter box ... we don't talk about the self-cleaning litterbox.

Well, now we have something to be extra worried about, namely the fact that we can lie to the machines—and so can thieves and sorcerors. Everything has a True Name, and the ghosts know them as such but don't understand the concept of lying (because they are a howling cognitive vacuum rather than actually conscious). Consequently it becomes possible to convince a ghost that the washing machine is not a washing machine but a hippopotamus. Or that the STOP sign at the end of the street is a 50km/h speed limit sign. The end result is people who live in a world full of haunted appliances like the mop and bucket out of the sorcerer's apprentice fairy tale, with the added twist that malefactors can lie to the furniture and cause it to hallucinate violently, or simply break. (Or call the police and tell them that an armed home invasion is in progress because some griefer uploaded a patch to your home security camera that identifies you as a wanted criminal and labels your phone as a gun.)

Finally, you might think you can avoid this shit by not allowing any internet-of-things compatible appliances—or the ghosts of Cortana and Siri—into your household. And that's fine, and it's going to stay fine right up until the moment you find yourself in this elevator ...

1501 Comments

1:

I wonder what they'd make of "This is not a pipe"?

2:

Am now envisaging a hotel evacuation at 3am in midwinter because some prankster scribbled FIRE on a whiteboard in the hotel lobby.

3:

I am a bit less out of touch, and have little to add, except:

Q: "OK, gang, it's Stravinsky night - what shall we show?"
A: "Firebird! Firebird! Firebird!"

Or when a SWAT team is called to shoot up a conference of winemakers for talking about terroire ....

4:
Similarly, in the 90s, per legend, a DARPA initiative to develop automated image recognition for tanks that could distinguish between NATO and Warsaw Pact machines foundered when it became apparent that the NN was returning hits not on the basis of the vehicle type, but on whether there was snow and pine forests in the background (which were oddly more common in publicity photographs of Soviet tanks than in snaps of American or French or South Korean ones).

About that: it probably never happened.

5:

I followed the link to the video. I'm having a problem with my PC sound at present, so I turned on subtitles. The subtitles have obviously been generated by a voice recognition algorithm, and were wrong enough that I couldn't quite follow what was going on, but I think this is a video about problems with voice recognition in an elevator that can't understand "eleven" in a Scottish accent. However the captioning algorithm seems to have no problem with that bit.

I'm going to fix my audio problems and come back to this.

6:

You really need working audio to appreciate the sketch (in the video). It's the accents that carry it.

7:

Looking forward to a computer responding to deepfake requests with a picture of an apple with "Tom Cruise" written on it.

8:

My black and white Tuxedo cat gained an additioinal name after the 'AI' mode on the smartphone identified him as a Panda.

9:

To be fair to the machines, apparently American Q-cumbers aren't clear on the concept of being lied to either.

10:

Actually, it probably DID happen, though perhaps not in the exact forms quoted. I was close to some of the early research (and dabbled myself), and the effect is real - I have observed it many times since, when crude forms of recognition are used to suggest matches. The main reason that it is hard to track down is that such failures are embarrassing, and hence swept under the carpet. My apologies, but here is a bit of a lecture.

All such methods (and they are both much more general than automated image recognition and predate it by decades) are statistical in nature, and it is NOT feasible to eliminate mistakes. The cost (weighting) of type I and II errors (false positives and negatives) is part of the training. But, inherently, when it encounters a condition that is not well-represented in the training set, the reliability goes to pot.

If the 'AI' is largely programmed by telling it what characteristics are of consequence, it will not pick up extraneous ones; if they were written to disclose their criteria, then they could be checked; but most modern training does neither. Let's skip the rant about that. I could go on, with a few more subtle aspects, but I am many, many decades rusty and this area has developed unimaginably since.

So, no matter HOW universal the traning set is, there will always be some circumstances where it is using an unrelated characteristic to make the judgement. Humans are no different, but we DO have a censoring mechanism as part of our intelligence that says "Based on other experience, something is not right here."

And it is the lack of that "common sense" that makes me say that AI is not even approaching human intelligence, and agree that this is a disaster in the making.


Actually, I do have one further point that absolutely horrifies me. In the UK, the courts have increasingly accepted such 'evidence' from the spooks and police, and we have trials where the accused is not even told what the evidence is against him.

11:

Arrgh,Therion 667 beat me to it, never, mind ....

Actually, nothing new here, at all: - Apple (fruit) & IPod ... "Ceci ne's pas un pipe" as René Magritte showed us, many years ago!
It's all distressingly bronze age, but we haven't come anywhere close to scraping the bottom of the barrel yet. THAT modern? Neolithic, or possibly earlier IMHO.
A re-read of U K le G on the Rule of Names & the True Names of Things might be in order.

As for "Elevator" - all too possible - I have already been caught twice by this.
Once, because my ( By modern standards ) clipped RP English has not been recognised by the at least "mid-Atlantic" fucked-over voice recognition & again in an automated basic German class, ("Babbel" ), where, again, my old-fashioned clear speech has screwed the machine software.
Can I start screaming now?

Jelliphish
I once has a small b-&-w rescue cat called "Panda"

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
OK peoples, what ahem, "pranks" do we want to perpetrate to really screw over the state & company fake "security" systems, just to show them?

12:

"WebAssembly framework (based on top of JavaScript running inside a web browser hosted on top of a traditional big-ass operating system) wouldn't be terribly fast"

I thought WebAssembly was supposed to be compiled to its own virtual machine and to be faster than JavaScript, rather than sitting on top of JavaScript.

13:

There are a number of SF stories about images or thoughts that crash the brain, one being The Riddle of the Universe and Its Solution, and another of course being the .

But if human brains are basically neural networks, and you can fool a specific neural network with the right image, could there be ways of doing that with the specific neural network in the visual cortex?

Probably not, at least in general. And the brain being evolved, it's reasonable to suppose that there are multiple neural nets in there with overlapping jobs. So one can imagine an image which makes a particular person say "It looks like a giant spider crawling towards me. Except I can also see its a picture of a dog at the same time. That's wierd!"

Of course it might turn out that there are some nets in the visual cortex that are the same for everyone, or for large groups, so there may be illusions that work on some people but not others.

Either way, reverse engineering wetware in order to develop such images could be a fruitful area of research.

14:

The human brain and 'neural networks' in the 'AI' sense have little in common, and the latter term was invented by people with a rudimentary understand of brain function, which turned out to be incorrect. We do not know how the brain works, but it it is very different and much less 'crashable'.

However, the visual cortex is much better understood, is largely 'hard-wired' (unlike the auditory one), and there are plenty of known ways to fool it. The simpler ones have been known for ages, and your example isn't just imaginary - there are lots of such examples, such as this classic:

http://www.optical-illusionist.com/illusions/young-lady-or-old-woman-illusion

15:

could there be ways of doing that with the specific neural network in the visual cortex?

Dave Langford got a Hugo award for that back in 2000: see also.

16:

I suspect we're going to need a lot more (and a lot more sophisticated) use of "supervised classification". Short version: That's the term used in remote sensing to describe when humans "supervise" the classification performed by a computer algorithm to confirm or reject the computer's decision. Particularly when the human supervisor is really two or more people who discuss the classifications and reach consensus on any questionable pixels, this can greatly improve the classification accuracy. If you do ground-truthing (going to the site of a remote sensing image to see what's actually there), the accuracy can improve further.

For example, in Charlie's example with tanks in a forest, the supervisor could have classified the image pixels containing part of a tank as "is tank" (!) and the image pixels containing only forest as "is not tank" (0). This would have minimized (possibly eliminated) the problem of classifying trees as tanks. Relying on the neural network to make this distinction seems... um... lazy? Unsophisticated?

Hacking supervised systems should prove harder, since (for example) a training set that includes only words (e.g., "ceci est une pipe") would not appear in the training set for pipes. For that matter, they could be included in the "is not pipe" training set. You could still hack such a system by feeding it a high-resolution image of a pipe or simply tapping into the camera's connection to the computer and directly streaming images of a real pipe. Hackers seem to always find a backdoor, so you'd need to be vigilant and keep monitoring the literature for new exploits.

Back-dooring neural nets will also be a thing, but I suspect this will be easier if it involves hacking the host computer (e.g., to substitute your own images for the camera's images) will be more effective. And so long as the computer is controlled by a human with authority to override the computer, social engineering will remain a serious problem.

17:

Oops... substitute (1) = binary 1 = true for (!) in the previous comment. I didn't mean to imply "not".

19:

Unfortunately, that's not the way things are going :-(

Your example wouldn't help, because it would have correctly classified the trees as non-tank - what made it a Russian tank was the presence of trees. So you can't just supervise in training - you have to do it in real life. And the supervisor needs the authority and ability to override the AI, which is often not the case.

I had this problem just recently, when the DVLA automation point-blank refused to recognise my driving licence to renew it, and the bureaucrat who was the fallback merely mimicked the AI.

My wife solved it by telephoning, finding an actual human, and being walked through the use of a loophole to renew it. If they had closed that loophole, I would have been stuffed.

This sort of thing is getting increasingly common.

20:

Elderly Cynic @ 10 : : "And it is the lack of that "common sense" that makes me say that AI is not even approaching human intelligence."

Don't despair! The people working on common sense at the Cyc project are still at it!

https://en.wikipedia.org/wiki/Cyc

Who knows, they might succeed in another 37 years.

21:

EC
Except that, under some circs, the human brain is all-too-crashable. Just not visually.
But, look at the multiple cases, over thousands of years of (usually, but not always) insanity/mania of a religious nature.
The person behaves utterly irrationally & it is very difficult to crash them out of such a state, once in it.

Oh yes: In the UK, the courts have increasingly accepted such 'evidence' from the spooks and police, and we have trials where the accused is not even told what the evidence is against him. Where - examples, please ... because this sort of thing is very worrying.
We've had past examples of this in a slightly different context, & it's taken many years to sort out the fuck-ups.
Example (?) Here, for a start - "Computer says no" & no-one is interested in actual reality.

22:

That's not crashing, any more than the fact that some lighting and sound effects can cause some people to lose balance, be unable to think, have panic attacks, or trigger migraines is. Yes, it happens - some lighting triggers epiletic fits in susceptible people, and those definitely count as 'crashes'. So far, nobody has found anything that does the same in 'neurotypical' people, but there is no reason to say nothing can.

The point is that the brain is fairly 'uncrashable', because evolution selected against that failure.

23:

Spoilsport ;)

Next you’ll be telling us that Kangaroo’s can’t use SAMS.


https://www.snopes.com/fact-check/shoot-me-kangaroo-down-sport/

24:

because evolution selected against that failure.

One problem with current AI might be that, in evolutionary terms, it's just getting into the Ediacaran. But it's evolving fast and I'd not want to make any bets on where it will ten years from now.

Also, it isn't obvious to me that several of the problems AI is currently manifesting are all that different from those that afflict the natural sort even after more than half a gigayear of fierce evolution.

25:

Yes especially once AR takes off

Read a book awhile ago where a guy built a suit covered in led displays running advertising banners

Since everyone walked around with augmented reality contact lenses on all the time, the ad blocking software in the AR edited him out and he achieved effective invisibility

26:

"shitty internet-of-things gizmos"... that paragraph reads like a current sysadmin wrote it. Don't sell yourself short Charlie... the tech changes every 5 years but for the BOFH, the song remains the same.

Also a fun shout-out to Stephen Blackmoore... one of his illusion spells involves writing the desired effect on a post-it note and attaching it to the target (e.g. "I'm a cop" or "I'm not here"). Who knew it came from AI?

27:

If you've watched the animated series "Futurama" you may have noticed something: all the robots are subtly imbecilic. They're intelligent but can't see outside the bounds of their programming. So for example you can talk one into doing something by using arguments that are nonsensical but which meet the requirements of their parameters.

28:

Elderly Cynic notes: "Your example wouldn't help, because it would have correctly classified the trees as non-tank - what made it a Russian tank was the presence of trees."

No, what made it a Russian tank was the fact that it was large, made of metal, and had a turret with a gun attached. Adding trees doesn't change the definition of "tank". Trees do not make it a tank, let alone a Russian tank. (Unless it was a Potemkin tank, in which case the AI was correct: the wooden boards made it a faux tank.) You didn't understand my explanation of supervision: the point is to eliminate the extraneous so that the AI learns to focus on the things that define the target as a tank, and learn to ignore the trees.

EC: "So you can't just supervise in training - you have to do it in real life."

"Supervise" is remote-sensing jargon. It means that the AI is not left to go its merry way, but is rather taught by the human to correct its errors so that it learns the correct details to focus on. It's analogous to the way a trainee surgeon is supervised by an experienced surgeon before they're allowed to fly solo.

EC: "And the supervisor needs the authority and ability to override the AI, which is often not the case."

That's another issue, and it's definitely a problem. To augment your example, I was once denied approval for a credit card and nobody at the bank could explain why... until I asked the manager to intervene. The problem turned out to be that I applied for an adult credit card while in university and should have applied for a student card.

29:

My 2 cents on how this is going (since I work in the field)

At this point, NN that interact with people can input 3 types of data: text, audio, images (video is a subset of images)

1. Text (NLP): Before the BERT transformer models came out in 2018, there was a thought that most practical uses of NLP was maxed out. Most business applications didn't have enough relevant data to avoid overfitting, and those that did were generalists rather than specialist models. Right now on the SuperGLUE tests, transformers exist which beat the human benchmark.
https://super.gluebenchmark.com/leaderboard/
https://en.wikipedia.org/wiki/Transformer_(machine_learning_model)

2. Audio. Speech-to-text is surprisingly accurate, to the point that this area is relatively neglected. It doesn't help that audio-based NN are niche applications. For those applications, the thought is that there isn't a lot of data out there to train anything this complex. Perhaps transformers are revolutionizing that field as well?

3. Images/Video. This is a huge area of research, but here there's a huge concentration problem. Other than medical and military uses, there's a sense that most companies don't need this capability. That's why a lot of the research in this field comes from the FAANG + Microsoft and their Chinese equivalents. I've heard the opinion that the only thing that would break this concentration would be VR/AR. This was before Zoom and Clubhouse made it big.

30:

what made it a Russian tank was the fact that it was large, made of metal, and had a turret with a gun attached

Well, the large metal turret thing makes it a tank. It's the trees that make it Russian :-)


On the topic of supervision, I'm reminded of Peter Watts' story "Malak". If you haven't read it, I think you'll enjoy it.

https://www.rifters.com/real/shorts/PeterWatts_Malak.pdf

31:

"I suspect we're going to need a lot more (and a lot more sophisticated) use of "supervised classification". Short version: That's the term used in remote sensing to describe when humans "supervise" the classification performed by a computer algorithm to confirm or reject the computer's decision."

A note on terminology:

Everything I've read about 'supervised' learning/classification refers to ML techniques with labels for each case. Human involvement is not necessarily involved, and of course it is very expensive.

That case of NATO vs Warsaw Pact vehicles (and dogs vs wolves, another similar situation) was 'supervised classification'.

32:
Invisible Sun...50% longer than previous books in the series.

Ah, good. I was thinking it might take an awful lot of plot to get from the close of Dark State to a satisfying ending, more than would fit in a normal-sized volume. Personally I'd be happier with a tetralogy or even a short wrap-up and another trilogy, but let's not break the Charlie, shall we?

33:

I was an undergrad CS student in the early to mid 90s. The joke with AI courses was like the joke with nuclear fusion: it was only 5 years away from completely changing the world - and had been for the last 40 years. The highlight seemed to be the A* algorithm, and the idea that functional languages might someday be 'competitive'.

Well, it may still be right about fusion. Wrong about AI with the proliferation of massively parallel compute resources - my Raspberry PI 4 would be considered a super-computer when I started working in the industry. I don't mind admitting I feel a massive gap in my technical corpus right now. Like a missing tooth. I haven't seen anything comparable to the rate of progress in 'AI' (or really, multi-dimensional tensor weight training) since the explosion in graphics hardware capabilities. The difference being, I was all across that - being related to work / interests.

I'm literally at the point where I will admit I don't even know about the things *I don't know* in this field. (Unknown unknowns). I'm curious about how the results we see today can be traced back to pure academia - long before any practical hardware was even on the horizon.

Purely anecdotal: someone set up a google 'nest' / 'mini' whatever in my home. I asked it to recite PI to 100 places which it did. I asked it if PI was transcendental, which it succinctly answered with a well-regarded reference. You do not get this from `if ... then ...` paradigms.

I predict a lot of posts will pontificate from the peak of 'Mt. Stupid', and how it's just neural nets, analogues to meat-brains, Markov chains, etc... we'll see.

34:

Barry notes: "Everything I've read about 'supervised' learning/classification refers to ML techniques with labels for each case. Human involvement is not necessarily involved, and of course it is very expensive."

Note that I was referring to the terminology used in remote sensing. For example, .

My point was that this approach should be applied to AI classification of visual, auditory, or other information if it isn't already being applied. It's not like it's a new technique; I recall reading about it back in the 1980s (when I was studying forest ecosystems), and it's become SOP in many forms of land-use classification these days.

35:

The link seems to have disappeared. Here it is again, minus angle brackets:

https://mapasyst.extension.org/whats-the-difference-between-a-supervised-and-unsupervised-image-classification/

If that disappears, try Googling "What’s the difference between a supervised and unsupervised image classification?" and "Cooperative Extension"

[[ You can't just put angle brackets round a link - you need to do the whole a href= stuff too - mod ]]

36:

On the topic of supervision, I'm reminded of Peter Watts' story "Malak". If you haven't read it, I think you'll enjoy it.

I did, and thank you for that.

מֲלְאָךְ

https://biblehub.com/hebrew/4397.htm

37:

Charlie mentions tens of thousands of random pics from the 'Net. My instant reaction is a) how many of those actually had alt-text, and how many of those that did had *garbage*, put there by advertisers?

*sigh*

Time to pull up General Semantics again, and use that, with code that recognizes the levels of abstraction, as a filter before it tries to identify something.

38:

Ah, good. I was thinking it might take an awful lot of plot to get from the close of Dark State to a satisfying ending

Pretty much. When I checked the copy edits I gave it a thorough read. It's going to be a 450-500 page brick, but it reads fast and twisty.

39:

Direct link to the openai blog entry, which is a fun (savage) read:
Multimodal Neurons in Artificial Neural Networks (March 4, 2021)
We’ve discovered neurons in CLIP that respond to the same concept whether presented literally, symbolically, or conceptually. This may explain CLIP’s accuracy in classifying surprising visual renditions of concepts, and is also an important step toward understanding the associations and biases that CLIP and similar models learn.

This is the version that includes all the images about "Typographic Attacks"
Multimodal Neurons in Artificial Neural Networks (March 4, 2021)

One of the funnier tweets about this (forget where I saw it):

The year is 2030. Amazon-Bots are hunting me down because I liked a tweet about unions. I quickly spray the word TUNNEL on a wall and hide behind a trashcan as they all charge headfirst into the bricks. https://t.co/tUw3rYBT5G

— Joe Köller (@JoeKllr) March 5, 2021

Re speed,
Cerebras Wafer Packs 2.6 Trillion-Transistor CPU With 850,000 Cores (Joel Hruska, August 18, 2020)
Won't fit in a phone though (well, maybe a very big phone): the previous version: "The Cerebras Wafer Scale Engine is 46,225 mm2" (Argonne National Labs (US) has one of their CS-1 computers.)

40:

"For those applications, the thought is that there isn't a lot of data out there to train anything this complex. Perhaps transformers are revolutionizing that field as well?"

For speech to text, what I have heard is that close caption TV shows were providing massive data sets.

41:

There are a bunch of suggestions for ways of jamming facial recognition in surveillance systems. Here is another one.

The trouble with all these is that if you are the only one wearing it, then you are really going to stand out. So its going to be like this XKCD cartoon (in any discussion of technology there is *always* a relevant XKCD). This stuff only works if lots of people do it. But if lots of people do it, countermeasures will be deployed.


42:

I would be worried about trusting systems to make decisions without humans in the mix. Even fairly simple glitches can get nasty fast.

https://www.wired.com/story/null-license-plate-landed-one-hacker-ticket-hell/

Seems to be a subset of 'use alphanumeric string to indicate no plate that can also be given as a custom plate' fault in programming, which causes problems in many cases.

https://www.snopes.com/fact-check/auto-no-plate/

43:

From the NYT just today:

https://www.nytimes.com/2021/03/06/business/the-robots-are-coming-for-phil-in-accounting.html

AI-powered accountancy programs. What could go wrong?

44:

That tank recognition story always suggests to me that it is an updated version of the story about German and Russian tanks in WW2. One side decided to try blowing up the other side's tanks by training dogs to run underneath them with mines on their backs. But the dogs learned to identify a target as being The Thing That Smells Of Fuel, and since one side used petrol and the other side used diesel, in action the dogs ignored the other side's tanks and blew up their own side's instead.

45:

I would be worried about trusting systems to make decisions without humans in the mix.

Yeah, but humans in the mix aren't all that much of an improvement. Bureaucracies tend to be an example of that. Also, the recent administrations of various nations discussed here.

46:

In the cage match of Darwin versus Earthsea, Darwin always wins.

Always.

Much as I love Earthsea, it's complete bullshit, and you know that if you think about it logically. How can a categorical name have utility, when not just every thing, but every part of every thing, has a unique name. Heck, just think about what happens when someone picks an apple off a tree and eats it, in a system of such names? Yes, some of the names both go to shit (what remains of the apple after it's gone through the human), don't go to shit (the apple tree) and change (the molecules of apple that go into the human). What does which True Name apply to when?

I'll get back to a better reference in a second, but the key point is that AI categories are always at least somewhat untrue. That's the point of Darwinism. Species (named groups of things) are groups that have individual variation within them because that's how they worked, not because they're flawed examples of some perfect design some cloud cuckoo came up with.

Nothing has a True Name, but some names are better than others. So if you're trying to put names on everything, the process is going to get screwed up somehow, whether you call it a lie, a mistake, a gremlin, sexism, discrimination, or a really bad idea.

Now, about typological thinking without True Name: The book you want to read is Walter Ong's Orality and Literacy, which is about the profound differences in how non-literate people see the world, compared with literate people. The tl;dr is that the idea of True Names comes fully from a world of widespread, routine literacy. Nonliterate people and their languages didn't generalize to anywhere near that degree, because it wasn't useful. If you're paying attention, you know that every kind of apple tastes different. Also, some apple trees produce decent wood for carving, while some are better as firewood. In texts, there's some utility of generalizing them all as "apple," but in real life, it's better to have an immense diversity of words for them.

Words allow a level of abstraction that's impossible with only oral language, because people tend to start thinking in terms of the permanent words they see, not the ephemeral language they hear. It's the difference between a god making the world out of a lump of clay, or God speaking and the Words becoming Things. The latter is a Religion of the Book. The former is animistic. This doesn't mean that nonliterate people are stupider than we are (I think the evidence points the other way, actually), but it does mean that thinking without written words is something any literate person is going to struggle with, and that makes for all sorts of erroneous back-projections about how "primitive" people thought. You may think they're primitive, but we're the idiots who pay other people to own computers to know things for us.

Anyway, the fundamental problem now is that we're trying to turn reality into words and numbers, make the actual virtual. Of course this abstraction, simplification involves lying and omissions. How could it not? I suspect Schneier's going to add standard rants on training set hygiene to his useful rants about security theater and the dangers of IoT. Both are equally dangerous.

If you want to hack virtualization, pick up your umbrella, go surreal, and also follow the protesters who hack surveillance tech. The umbrella is useful as a perfectly innocuous tool for shielding people from sun, rain, pepper spray, marking dyes, and surveillance cameras.

47:

> From the NYT just today:


From which and talking about, I think, the logistic curve model,

“The rate of progression of this technology is faster than any previous automation,” said Mr. Le Clair, the Forrester analyst, who thinks we are closer to the beginning than the end of the corporate A.I. boom.

“We haven’t hit the exponential point of this stuff yet,” he added. “And when we do, it’s going to be dramatic.”

48:

Trouble with all those kinds of suggestions is that there is no way to get any reliable data on whether they actually work or not, and it is an area which is an appallingly prolific bullshit generator. Just as we currently have people convinced that spraying your number plate with a relabelled can of hair spray will prevent automatic systems reading it, but nobody knows what the actual subtle alterations that really do have such an effect really are, so we will see a vast array of silly hats and so forth which are all claimed to break the system, almost certainly don't, but do make it vastly more difficult to determine what things really do.

49:

"The umbrella is useful as a perfectly innocuous tool..."

An umbrella is a dastardly and evil invention for poking random people's eyes out, masquerading as a rain deflector.

("Masquerade" is, of course, just a more old fashioned term referring to adversarial inputs.)

50:

Speak like Arnold Schwarzenegger and the voice recognition software will understand you.

51:

And that's fine, and it's going to stay fine right up until the moment you find yourself in this elevator ...

This is the favorite Youtube video of my neighbor from Scotland.

52:

It's as if things have to balance out, so teaching machines to think goes hand in hand with teaching humans not to. And you get the same problem with the supposedly intelligent processing engine becoming trained to operate on cues that have very little to do with the ostensible intended purpose. So you get the customer service centre zombie learning that by copying and pasting some chunk of company boilerplate that contains two or three keywords in common with a customer's complaint, they can tick off the greatest number of complaints with the smallest chance of being shouted at for displaying initiative, while the customer gets closer and closer to detonation point due to repeatedly getting the same stupid answers that don't actually address the matter of the complaint.

Not only do the purely machine parts of the system have no common sense, but the elements which could potentially introduce some common sense are in fact being conditioned into keeping the system as a whole devoid of it. To sort this problem out requires both the application of genuine effort, and the existence of a genuine commitment to actually sorting it. Neither is likely to be forthcoming, and I can only be glad not to have been born at any later date than I actually was.

53:

Yes, indeed. I have been abused for holding my arm up, causing an umbrella to deflect and spin instead of sticking into my eye.

54:

Who knows, they might succeed in another 37 years.

They could get the fusion folks to help and get it done in 20.

55:

I have a neural network wildlife identifier on my phone (Seek from iNaturalist)

I point the camera at a mule deer and it tells me it’s a mule deer. I point at a centipede and it tells me what species of centipede it is.

I point it at a hummingbird feeder, and it tells me that it is a hummingbird. A large fraction of its hummingbird training set apparently included bright red plastic hummingbird feeders.

56:

The point is that the brain is fairly 'uncrashable', because evolution selected against that failure.

Then you get a non trivial number of people like me who have issues with pattern recognition. I rarely can look at a group of 20 people and notice the ones I know. Much to the agony of my wife early in our marriage. It is tied to color processing. I see all the colors and can pick out a dot in those pattern tests when asked but just don't see the patterns. If you tell me what I'm looking for is "blue" I CAN trace the blue dots and tell you it is a circle or the number 9. But I don't see it until I trace it and if I look away I will likely loose it.

Which may be tied to how hard it is for me to read red on black or black on red. And that's a big fav with the hip graphics artists. From what I understand this and my other issue afflicts 10% or so of folks with lots of northern European heritage but we never become graphics artists so those who do keep creating hip things we can't "see".

57:

David L @ 54: "They could get the fusion folks to help and get it (common sense in AI) done in 20 (years).

If "US Robots and Mechanical Men" got the help of the cold fusion folks I have no doubt that we would have common sense in AI 5 years from now.

58:

H
True Names
It's another iteration of Plato's True Forms or The Cave, isn't it?
And, yes, I was aware of the problem(s) before you wrote your piece.
Good fun though.

59:

https://aeon.co/videos/footage-from-hong-kong-reveals-the-combustible-contested-reality-of-street-protest

Wow, you can poke someone's eye out with an umbrella tip? I'm impressed. My aim isn't that good.

Anyway, if you want to see umbrellas in action, check out: https://aeon.co/videos/footage-from-hong-kong-reveals-the-combustible-contested-reality-of-street-protest

60:

Perhaps.

However, an apple tree that's cut down and cut up is no longer the same.

True names go back to classical magick, as does the Law of Contagion and the Law of Similarity. You want a magickal system in a fantasy, you use them, rather than having to come up with *everything*.

61:

So it turns out if you show CLIP an image of a Granny Smith, it returns "apple" ... until you stick a label on the fruit that says "iPod", at which point as far as CLIP is concerned you can plug in your headphones.

Naturally. Freefall showed us that fake transponder mean robots in disguise! See the strips here and here.

62:

"The joke with AI courses was like the joke with nuclear fusion: it was only 5 years away from completely changing the world - and had been for the last 40 years."

Actually, the joke about AI is that it completely changes the world every 5 years. Before 2010, most of the machine learning was "shallow learning", the opposite of deep learning (I'm including graph theory). Afterwards, they transitioned to simple neural networks. A personal note, I was first introduced to the subject in 2009 as an obscure elective. The school didn't create a data science masters until 1 year after I graduated (so I'm dating myself here a bit).

In 2012, convolutional neural networks, which revolutionized image recognition. Between then and 2015, the networks advanced to recurrent neural networks, LSTMs, and reinforcement learning. IF you've been dealing with the FAANGs, you've seen how these networks revolutionized the world.

Then in 2015, they popularized GANN's, otherwise known as Deepfakes. Another advance was seq2seq networks, which are very useful for language translation, Alexa, and chatbots. Up until 2018, people in the field were complaining that the low-hanging fruit had been picked, and that new improvements would be incremental.

Then in 2018, the transformer came about. Right now, the industry is trying to integrate transformers into consumer products.

The problem fundamentally is that people redefine revolutionizing the world to make the previous advance seem incremental. Plus, the improvements aren't instantaneous, so people ignore them. For instance, when was Google Maps able to optimize your route taking traffic into account? Did that not revolutionize the world, or was that a case of incremental improvement?

Please let me know if there's a type of network I've forgotten.

63:

If anyone is interested in the developments, I'd recommend looking at the courses by this professor on Udemy. He's the best professor out there on deep learning beyond simple feed-forward and backpropagation neural networks

https://deeplearningcourses.com/course_order

I took some of his courses on material outside of NLP-related tasks during the pandemic. That's a huge reason why I was very busy last year. I'd start with Convolutional Neural Networks, and go down in order to see the latest material (but I'd also recommend taking the Modern Deep Learning as well). He covers everything until transformers. To my knowledge, he hasn't created classes for those yet, which is a pity.

In my comment above, I completely forgot unsupervised deep learning methodologies such as variational autoencoders and RBMs, both of which are important for tasks such as dimensional reduction (compression). But they're more background methodologies these days, so it's hard to say how much they revolutionized things.

Also, a lot of the theory comes from academia, but you'd be surprised how much of it is published by Google and Facebook employees.

64:

A thorough investigation of the legend of the AI that missed the tanks for the forest: https://www.gwern.net/Tanks

65:

> The tl;dr is that the idea of True Names comes fully from a world of widespread, routine literacy.

I think a lot of stories about fantasy magic — not all of them, but many — are stories about the interactions between literate and non-literate societies, or parts of societies. Even in our distant past, the future did not arrive evenly-distributed, and even in western Europe the less-literate past was not so long ago that we don't still tell stories about it.

66:

That piece is impressive. Thank you for linking it.

67:

Nothing has a True Name, but some names are better than others. So if you're trying to put names on everything, the process is going to get screwed up somehow, whether you call it a lie, a mistake, a gremlin, sexism, discrimination, or a really bad idea.

This kind of makes me want to write name magic fantasy with this premise. I'm not sure if there is some already, though (examples would be appreciated!).

Oh, well, into the idea pile it goes. (*closes the closet door fast so nothing falls out*)

68:

This is related, with links to papers.
Specification gaming examples in AI - master list (h/t list linked here by fajensen in 2018)
E.g. "Evolved creatures - falling / Creatures bred for speed grow really tall and generate high velocities by falling over / Sims, 1994"

(I do this too sometimes, TBH. Often the rules have loopholes, or more fun, can be edited.)

69:

"...and another of course being the ."

Knock that crap off. Everyone here can see the fnords.

(Did you hear about the depressed member of the Illuminati? He was pining for the fnords.)

70:

I think that was more a "sorry we missed your first story in 1988" Hugo, but regardless, well-deserved.

71:

One of my daydreams is that Freefall will win a Hugo. It's consistently one of the most intelligent things (of any kind) I read, and I'd love to see it get the recognition it deserves.

72:

On the subject of image-recognition, I think that humans do something very sophisticated with their vision. They build a 3-dimensional model, and do so based on (among other things) the differences in angle between two cameras mounted about three inches apart... So we don't just recognize a tree. We build a three-dimensional model of a tree and recognize characteristics, like a branching structure. Then we (probably) match a newly discovered plant to an already-known plant and do some kind of set-based math on how well the new plant matches the model of a tree. So if we're only familiar with pine trees, and we see an oak tree, we match the 3D characteristics of the pine tree; cylindrical, vertical structure, branching structures, bark, but wait, it has these funny looking berry-things (acorns) and leaves which are broad rather than long and pointy, then we say, "It's come kind of tree, but I don't recognize it."

We probably assign some things more tree-probability than others. Bark has a high-tree probability. Leaf-shape has a low tree probability because it might come from a bush or flower The reason that AI can't recognize things well is because it's not building a 3D model.

73:

The reason that AI can't recognize things well is because it's not building a 3D model.

Not convinced at all that it is because we have 2 eyes. I use monovision glasses. Which means my brain switches eyes depending on if I'm paying attention to something close up or far. With a crossover around 3 to 5 feet.

It is a big thing with contact lens wearers but not so much with eye glasses. My eye doc is a big fan for anyone willing to deal with the first week or so. I takes a bit for your brain to learn to switch eyes in terms of which one it pays attention to.

The only problem I have now is that every now and then the world goes out of focus if something blocks my eye corrected for distance. Like a pole or column.

Here's a link:
https://www.allaboutvision.com/eyeglasses/faq/eyeglasses-monovision.htm
But it is very biased against doing it with eye glasses. I'm a huge fan.

74:

Troutwaxer: I think that humans do something very sophisticated with their vision. They build a 3-dimensional model, [e.g. of a tree]

Yes, and on top of that they mix in semantic knowledge about trees, forests, the kinds of animals you find there, and similar stuff. Frame-based reasoning lets the viewer fill in the gaps and make inferences.

BTW, the model you speak of is actually 4-d. I once saw a really neat demo of this: you are shown a still picture of black splodges on a white ground, and it looks like nothing at all. Then it starts moving, and you see it is a dalmatian dog sniffing around a tree, but shown in 2-tone black and white.

I've often wondered what would happen if you could connect these statistical recognisers to semantic models based in some kind of inference engine, possibly informed by NLP processing of Wikipedia. (The idea of an AI that implicitly believes everything in Wikipedia is an interestingly comic concept)

75:

I was first introduced to the subject in 2009 as an obscure elective

Artificial neural networks are old, like "computer is a job" old.

They were being taught in electrical engineering about 1990 where I was, but were apparently unknown to the philosophy department. It made my course on "computers, minds, and logic" very confusing because the lecturer kept saying "this is impossible" about things that I knew were being done. He kept at it even after I started bringing him copies of papers from the people doing the things.

Back then, of course, it was complex systems with ones of nodes and sometimes as many as two layers before your poor computer caught fire, ran out of memory, or simply ground to a halt for a prolonged period.

76:

You have a very low bar for something to count as in improvement! And, incidentally, route optimisation was an established technology by the 1960s. All of the IT developments that I can think of that would count as really revolutionary in the past half century have been older (and logically simple) ones where the hardware became cheap and small enough to make them near-universal:

Personal computers and all that
Universal connectivity between them
Replacement of mechanical control devices by computers
Universal walkie-talkies

And, in the last, the revolution was not so much due to them, but to the unpredicted way in which society chose to use them and change itself.

77:

It's not. Binocular vision is overrated, even for judging distance, because it's useful only for a narrow range of distances and there are other visual mechanisms that overlap with it and can be used instead.

78:

Paul @ 74: "I've often wondered what would happen if you could connect these statistical recognisers to semantic models based in some kind of inference engine, possibly informed by NLP processing of Wikipedia."

Wikipedia is full of holes, contradictions and repetitions. I've been editing it since 2003 so I'm the cause and/or the witness of many of these problems.

I think that the IA counting on Wikipedia would, at first, go mad. Then, upon hearing of this problem, the regulars at Wikipedia:Help_desk and at Wikipedia:Teahouse would band together to form a therapy group to help out that poor AI.


https://en.wikipedia.org/wiki/Wikipedia:Help_desk

https://en.wikipedia.org/wiki/Wikipedia:Teahouse

79:

How neurons work began to be hashed out in the 1930s-1950s. Early artificial neural networks then showed up in the late 50s if I remember correctly, up to and including two layer networks like the perceptron. Then Marvin Minsky stuck the boot in with a proof that you can't make a NAND gate using a two layer NN, thereby "proving" for a generation that NNs can't be used for general purpose computing tasks.

Minsky was both right and wrong. For two layer networks he was entirely correct: and nobody had tried to build deeper networks with a hidden layer between input neurons and output neurons because circuitry was expensive and why bother? When we've already proven they're not computationally effective.

Turns out that adding one or more hidden layers solved the problem, though. Oops.

This became apparent around 1990 and caused quite a stir -- but again, the computers of the day weren't fast enough to simulate many layer NNs usefully: it took until the 2000s and the spread of dedicated GPUs for raw number-crunching to catch up.

Which probably explains the pot-hole your lecturer fell into.

80:

"You have a very low bar for something to count as in improvement!"

Perhaps I do? I generally te3nd to include things which the consumer doesn't notice directly but can have knock-on effects.

"And, incidentally, route optimization was an established technology by the 1960s."

Not really, It was an NP-hard problem that was just beginning to be understood, and most of the advances in the 60s involved just optimizing for shortest distance. Since then, advances permitted optimization not only for distance, but for time, driver's capability, local conditions, etc. The effects on society have been huge. Unlike your examples, we can't point to these changes as the sole cause of the changes in society (and improvements in most cases).

In the 1980s, the spread personal computers allowed for optimization w.r.t. to time as well, permitting just-in-time manufacturing and transport. This permitted outsourcing to poorer countries to be far more profitable, and gave rise to the Walmart superstore in the US.

A lot of the reductions in a plane ticket in the 90s and 00s were due to advances in route optimization ensuring the airline could amortize the cost of the plane among more passengers. Heck, this reduced the importance of the hub-and-spoke models, killing off the giants such as the 747 and the A380. In this case, it wasn't enough to predict the shortest route by distance, but the airplane had to get to its destination within a 15-30 minute window, regardless of conditions. Over those two decades, that window shrunk.

In the West, most passengers just saw the effects of that as either cheaper tickets or fewer delays. In the developing world, that brought the price of flying down to what was the middle class. Both these effects created the large tourist market which helped massively modernize Latin America and Southeast Asia (compare the poverty in 2000 vs 2010 vs 2019 in these regions).

The route optimization method I wrote above allowed for the following 2 things:
1. It's Uber's secret sauce, so to speak. That's why it dominates the ride-sharing market in the US and a lot of other countries. While the drivers aren't getting paid if the car is empty, Uber isn't making any money either. Thus, the competition within the ride-sharing sector is on which company can minimize the time the car stays empty.

2. Amazon's 1-2 hour delivery is also possible because of this route optimization advances. There's more to it than that, but I doubt Amazon's current dominance would have been possible without route optimization for time as well as distance.

81:

Sigh. Look, as I said, I did a little of that in the late 1960s and early 1970s, and was in contact with people who did more. Yes, it's NP hard - and that hasn't changed. But 'we' did not merely optimise for distance, I can assure you, and the only reasons we didn't do what is done today is (a) we didn't have the data and (b) the computers weren't up to it. We sure as hell knew how, and 'modern' algorithms are merely updated forms of the ones we knew then.

The route optimisations were done then, too - I remember a talk where someone was describing Israel's use of such methods to optimise grain shipments. And, no, that is NOT what primarily brought the prices down - I was flying when aircraft were packed more efficiently than they are today. Amazon established its dominance well before it started its 'immediate' deliveries. Uber's rise was more caused by it being allowed to bypass regulations than anything else - other taxi firms were using such approaches before it started.

More importantly, to call the changes you describe revolutionary is really sttretching the term.

82:

Re the monovision thing... due to strabismus that wasn't treated soon enough when I was a child, my vision developed switching back and forth between eyes as necessary: bi-ocular rather than binocular vision. Also very heavily right-eye dominant, and the right eye is a bit nearsighted. I am always amazed at the tricks my brain has learned to do for putting the two images together.

83:

“ All of the IT developments that I can think of that would count as really revolutionary in the past half century have been older (and logically simple) ones where the hardware became cheap”

Mostly agree, pretty much all computer science advances are driven by
improvements in either mathematical algorithms / foundations or hardware . The math ones just lag a lot more (Knuth wrote his stuff in the late 60’s). It’s easier to see the hardware driven ones because they are closer in time and more obviously linked but don’t underestimate the mathematical ones either . Some postdoc is probably working on something now which will be instrumental in computing in 2050

84:

Part of the world building in Peter Watts’ “Blindsight” is computers doing things that are critically important, incomprehensible/inexplicable to humans, and not always reliable.

85:

Yes, mostly, but only in the past half century. In the 1950s and 1960s, and even in the 1970s, the algorithmic ones had a much smaller lag - there are similar examples from later, too (such as randomised routing in telephone networks). The remaining mathematical ones are mostly extremely unobvious or fiendishly hard, which is why you are right about the lag.

86:

As a non-techie ...

1- The Russian-tank-is-always-in-a-forest scenario:

If you're teaching something to an infant, you make sure that they will focus on what you're teaching therefore you eliminate as much extraneous 'noise' as possible. You also repeat the exercise in as many different ways/perspectives as possible. As for picking out the 'correct' subject: ever hear of the game 'Where's Waldo?'. My conclusion: The AI programmers/developers weren't parents/had no interaction with living/developing organisms (pets).


2- As for messing up the enemy's AI once it can pick out an object (e.g., USian tank) out of the proverbial or real forest:

How about a tank slipcover? All you need is some fabric with appropriately sized and shaped geometric Styrofoam pieces slipped into various pockets to alter the tank's size and shape. Should work if the AI is using only visual stimuli. (Have heard this is actually an old ruse with various armies having used full scale balloons/toy tanks to mess with their enemies' intel gathering usu. via satellite.)


3- Part of 'visual recognition' is how something moves, i.e., changes its shape/size over time and/or changes/impacts its environment as it changes its shape/place. Haven't seen any mention of this as part of AI - curious as to why not.

4- Elevator-Scottish accent - yep! I've had such encounters with my cable and bank AI help desk/customer service desks. Fortunately both still have a 'speak with a human' option. Interestingly, both companies also have messages saying that all phone calls are recorded. If yes, then - once the human screens for reason to speak with a human - they also have an inventory of which types of accents/voices their AI needs to improve upon. However, given how crappy the service levels of both these outfits continue to be, I seriously doubt they'd bother actually spending any money on customer service improvements.

87:

Of course that’s already somewhat the case but he cranks it up to eleven.

88:

Agreed. If you lost an eye or were born with only one eye you'd still try to build a 3D model. But the binocular vision helps.

The other thing we do that computers probably don't do, is that we learn which items are mobile and which are not, and which items are embedded in other items and are unlikely to move, such as trees. So our mind somehow divides a bicycle from the ground even if we can't prove that that the two aren't linked together, and that seems to happen automatically once we've seen something move in a way which is disconnected from the ground.

So I think we're also filling in a variable someplace which says "can move on/slightly above the ground," "can't move but separate" (like a house,) "can't move but non-seperate" (like a rock formation,) "can move through air" and possibly "can swim/do water somehow."

So we take that octagonal form on a stick, determine that it fits most of the set-elements for "stop sign" class it as "can't move but separate" and are then able to distinguish it from a bicycle. But the key to doing this is to make a probabilistic* assumption - the mostly red, bent-metal mostly octagonal thing with the words "sto" printed on it in white, plus a sticker reading "Disco Kills" fits most of the probable elements for a stop sign."

Note that we do this without thinking much about it. It's not "artificial intelligence" so much as "artificial stupidity." I think what some people report experiencing as "the big brain reboot that happens when you're five" is probably related to perceptions; something that looks like a really complex problem to a really small child gets compiled down (essentially) to an API, so you see this thing made of circles and metal tubes and instead of thinking about it in the really complex way I've described you automatically call the API and it does the math and automatically says "bicycle."

* "Probable" is probably the wrong word here, but I'm not done with my coffee yet. What I mean is that "STO" is 75% of the printed matter expected to be on a stop sign, so we tentatively classify it as a stop sign. Red, with a white border is 90% of what we expect to classify as a stop sign. (There's still that "Disco kills" sticker**) The stick is 100% stop sign, or if it's been hit by a car at some point, maybe 90% of what we expect to see from a stop sign. And the material of the stick is 100% of one of the usual stop-sign-stick materials, so we do some math (75% plus 90%, plus 100%, plus 100%, which means 365% divided by four factors, giving us a 91.25% probability that this is a stop-sign,) and we hit the brakes without thinking further about it. Also we've tagged stop signs as "can't move but separate" and this thing we're looking about hasn't gotten up and walked someplace, so it fits that criteria as well. Just don't write "fnord" on the stop sign.

** Plus there's a subset of proven stop-signs with stickers.

89:

My wife and I use Alexa to turn lights on and off. She is American (Minnesotan, to be exact), I am Russian. Alexa understands us both fine... as long as I do not try to sound American. I found that I get best results if I speak with seriously exaggerated Russian accent.

91:

How about a tank slipcover? All you need is some fabric with appropriately sized and shaped geometric Styrofoam pieces slipped into various pockets to alter the tank's size and shape. Should work if the AI is using only visual stimuli. (Have heard this is actually an old ruse with various armies having used full scale balloons/toy tanks to mess with their enemies' intel gathering usu. via satellite.)

That was the whole stunt in El Alamein (British fighting the Nazis in the Sahara) where the Brits used frames and sub-scale models to deceive the Nazi spotter planes. There's a book called The War Magicians that gives a (semi-factual) account of how this was done, although there are disputes about who was responsible for what.

The Ghost Army (US 23rd Special Headquarters Unit, my favorite force in WW2) in the ETO used inflatable mockups of equipment, sound effects tanks, radio telegraphy deceptions, and a 1000-odd special effects unit (who often disguised themselves as other units) to deceive the Nazis on the advance of the Allied Army in Belgium and across the Rhine. As one example, the combat engineers running the sound effects tanks got so good at emulating the military police of other units that, during the liberation of the concentration camps, they ended up serving as military police alongside real MPs, with no one the wiser.

And we shouldn't forget all the Q- efforts by the British during the Blitz, with all the fake installations installed to make it harder to accurately bomb London and the UK.

Nowadays IIRC camouflage and tactical deception seem to be more integrated into standard ops, not set aside in special units. Strategic deception, infowar, adn psyops are still their own units, and one might argue that the US Space Force specializes in this mission at the moment, along with flying a lot of satellites.

One obvious counter to visual deception goes back to the 1990s, where the US army uses infrared cameras and radio-spectrum intelligence to identify things like tanks, to get around the problems associated with visual-spectrum deception. Now a lot of camouflage effort goes into multispectral masking (updating uniform materials and changes in training) and other techie goodness. And so the Red Queen marches on.

As a footnote, getting back to the Q- efforts by the British, I've started wondering if this was sort of an idiot filter used by Q-Anon. People "like us" who are observant would look at the name Q-Anon and think (or at least feel) that there's deception involved. We're not the ones "they" want of course, so it's a cheap way to filter out the troublemakers and concentrate the suckers. If so, well played on their part.

92:

Yes, and on top of that they mix in semantic knowledge about trees, forests, the kinds of animals you find there, and similar stuff. Frame-based reasoning lets the viewer fill in the gaps and make inferences.

I am essentially blind in my left eye, and I have very few problems with distance. I did not even know that I have no depth perception until the first time I drove in Boston (I learned to drive in Albany, NY), and kept misjudging the distance to traffic lights. That’s when I realized that I always judged distances by the apparent size of objects, and Boston traffic lights hang lower to the ground than Albany traffic lights.

Last summer I had a related experience. I saw a dog jumping on a boulder, and my thought was "That’s one BIG dog!" It looked like a Great Dane. Then the dog ran toward me, and I realized it was only about the size of a Labrador, and the boulder was about half as far away as I thought.

Because the dog had Great Dane’s shape and markings I assumed it was one, and my brain interpreted the distance to the boulder accordingly.

93:

Last summer I had a related experience. I saw a dog jumping on a boulder, and my thought was "That’s one BIG dog!" It looked like a Great Dane. Then the dog ran toward me, and I realized it was only about the size of a Labrador, and the boulder was about half as far away as I thought.

Here's a Youtube video of someone who constructed a small room and painted it with very black light-absorbing paint. Viewed from the outside as he walks into this room he appears to shrink since there are no depth cues visible around his figure (about 3 minutes into the clip).

https://www.youtube.com/watch?v=p6q54q2iam8

94:

Heteromeles @ 91 : "That was the whole stunt in El Alamein (British fighting the Nazis in the Sahara) where the Brits used frames and sub-scale models to deceive the Nazi spotter planes"

That was Operation Bertram:

https://en.wikipedia.org/wiki/Operation_Bertram

And yes, the U.S. Ghost Army is well represented with its own article in Wikipedia.

One thing that's been really bothering me since about 2002 (when it was introduced) is the computer-generated pixillated camouflage uniform pattern called CADPAT. If it's blocky-looking and computer-generated, then shouldn'it be easier to spot by more powerful computers?

https://en.wikipedia.org/wiki/CADPAT#/media/File:Cadpat001.jpg

95:

Interestingly, both companies also have messages saying that all phone calls are recorded.

While those records might in theory be used to improve VR systems, as best I can tell 99.99% of theme (or more) are used as CYA things. Or employee evals. (My wife worked in a call center for 18 years "on the phones".)

And we just dealt with a foul up with an payroll withholding to a 401K plan. We were told the call would be reviewed and someone get back to us.

96:

So our mind somehow divides a bicycle from the ground even if we can't prove that that the two aren't linked together, and that seems to happen automatically once we've seen something move in a way which is disconnected from the ground.

I more and more think it's just that we (wet meat) have a decade or five of experiences that our brains sort through to interpret the visual world. And as we wander further off that pile of "memories" we get worse and worse results.

So training an AI with 100K pictures still fall short of our 10 to 50 years of observations. While I wasn't looking at tanks for 50 years, I was seeing the world around me and can quickly parse out much of what is "not tank".

For most of us in the US, Europe, and similar places learning to drive is greatly helped by being a passenger for 10+ years. When you take someone who has rarely been a passenger in a car they tend to find it much harder to learn to drive than those of us who grew up riding in one multiple times per day on average.

Which is why that very nebulous term "common sense" mis leads people from simple lives. Their memories just can't allow them to comprehend some topics. Especially when it comes to advanced tech or basic science.

97:

Not really. This is very similar to an area I know well (and am thinking about at present): pseudorandom number generation and testing. If the former is done well, you need a massive amount more power to distinguish the numbers from true random ones. Whether 'massive' falls into the category of 'more powerful computers' or 'over a galaxy full of comptronium' is unclear - that's what I am thinking about.

98:

a new Apple M1 Macbook Air (not even the higher performance Macbook Pro)

Both have the basically the same CPU. MBPro has fans that can kick in where the MBAir shows itself down to keep the temps under control.

The interesting thing as Charlie mentioned is the 16-core Neural Engine. Lot of chip space dedicated to something no one uses much at this time. But yesterday I got to wondering if it was a part of their car project. This gets that project a somewhat endless free (from the car group point of view) compute engine which they can integrate into their autonomous driving system. I wonder if this 16-core Neural Engine isn't something they've been designing for a while.

99:

Niala
You ( among other people ) have done it again, I'm afraid.
... shall never vanquished be, until great Birnam wood to high Dunsinane hill shall come against him.
Then there was the minesweeper that disguised itself as a jungle island
As for interpreting what we see, there's always The Fata Morgana ( Yes, I have seen one ) - of which a very good example was seen recently

100:

If you lost an eye or were born with only one eye you'd still try to build a 3D model. But the binocular vision helps.

Chicken vision ... only have one eye, but move it back and forth. Or have two sensor and swap between them (birds really do do this, they eyeball you with one eye the flip their head around so the other eye can have a go too). Also, image stabilisation by putting the sensor on a long arm which you articulate to keep the sensor(s) in one place. Some of the videos of birds being waved round while their head stays in one place are quite freaky.

Or LIDAR, have a strip of high res sensor and a rotating sensor or mirror.

101:

Re: '... sort of an idiot filter used by Q-Anon'

Thanks for the military intel-ops review - very interesting!

Yes - this and the 'Nigerian Prince' scam should be added to any AI Turing Test - as in: what info sources does your AI access to help in its analysis; what weight to each source, using what criteria, etc. And I want to see its decision tree path. Basically ditch the entire black/magic-box mystique that seems to be the current approach to AI development: If you can't see how someone arrived at their proof, then you can't verify that their proof is correct. We insist on this from human scientists and mathematicians - no reason to not insist on the same from a device (and the comp-sci's who program them).

102:

is that we learn which items are mobile and which are not

Context, or a model of the world beyond the immediate problem.

Training dogs is often this way, initially you have to teach some of them "sit (in the front yard)", then a completely unrelated trick "sit (in the back yard)" and so on until finally their brain does a little dance and says "I can sit anywhere" and life gets a lot easier. Some dogs you never notice, because either their world consists of you and sundry unimportant details so what they learn is "sit (you said so)"; or the flip happens so early no-one notices - often IME puppies that are trained with/by their parents are like this.

Also, there is nothing quite so cute-hilarious as a whole row of fluffy puppies all still clumsy and unco, trying to lie down and roll over on command.

103:

If the former is done well, you need a massive amount more power to distinguish the [pseudorandom] numbers from true random ones.

Like this? NOBUS = Nobody But US

https://en.wikipedia.org/wiki/NOBUS

Former NSA Director Michael Hayden acknowledged the concept of NOBUS:

You look at a vulnerability through a different lens if even with the vulnerability it requires substantial computational power or substantial other attributes and you have to make the judgment who else can do this? If there's a vulnerability here that weakens encryption but you still need four acres of Cray computers in the basement in order to work it you kind of think "NOBUS"


I don't think General Hayden had heard about hubris and nemesis.

104:

That's interesting. So you are cuing almost entirely off perspective? I am wondering what causes you not to be using parallax. I too have only ever had one functional eye, and I rely very heavily on parallax. Come to that, so do creatures which are often held up as exemplars of binocular vision; cats preparing to pounce weave their heads from side to side to get the parallax cues. I do the same thing for the same reason, albeit in different contexts. Parallax is always a very significant cue, regardless of how many eyes you have, so it seems strange that you appear to be using perspective instead.

My visual perception seems to involve a lot of comparing the actual, observed parallax with the expected parallax synthesised internally from a model; I even find that photographs appear to exhibit parallax because my brain is modelling what it ought to look like, and I find myself doing the head-weaving thing or even trying to look round the side, and experiencing a moment of confusion when it doesn't work. Similarly, I find holograms unimpressive because they don't look much different from what I expect from an ordinary photograph until I look at them from extreme enough angles that they are starting to not work.

Your vivid perspective-based illusions are pretty unfamiliar to me. I get that sort of thing from photographs sometimes, but not from real scenes. It may possibly account for occasional instances of something half-seen looking indefinably odd in a way which vanishes as soon as I turn my attention to it.

The popular misconception that binocularity is the be-all and end-all of depth perception was of some use to me at school, because it meant that nobody expected me to be able to intercept missiles or launch them accurately, and so saved me getting too much flak for not doing it. In fact I can throw and catch about as well as most people, but it does require attention, and mine was all used up on the trees and clouds and other things which were actually interesting.

But it did severely get on my tits when they started to actually teach us that misconception as fact in physics and biology lessons. I objected strongly to being expected to learn something which I knew so bloody well was not true. To the point where I eventually answered a question with "We are supposed to just say "binocular vision" here, but this is bollocks" followed by a two page analysis of exactly why it was bollocks, based on the angular resolution of the eye being inadequate to make it work except at very short range.

105:

"Some of the videos of birds being waved round while their head stays in one place are quite freaky."

Indeed, it is quite fun doing it interactively with a pigeon. It's also why they jerk their heads back and forth when they walk - trying to keep it in one place for as long as they can, then quickly moving it to the next station.

I've heard of someone sighting on a hovering kestrel with a theodolite, and observing that its body was being buffeted all over the place but its head was steady as a rock.

"unco"

Golly, that word takes me back :)

106:

I've heard of someone sighting on a hovering kestrel with a theodolite, and observing that its body was being buffeted all over the place but its head was steady as a rock.
Yeah, it's quite an effect. Probably the kestrel system is optimized for stabilizing vs wingbeats (it being evolved), but even when the birds are hand-held the head and gaze remain completely steady as long as the body isn't moved more than a couple of centimetres.
The American Kestrel's diet is mostly insects, grasshoppers favored, and in my (limited) experience they're hovering maybe 15-20 meters up, looking for a meal.


107:

A few years ago I tried out a side gig where I was basically reviewing websites for Google to help it train its algorithms.

Typically I would be sent to a site to review it for accuracy to a set of search terms. Does this site give the reader information about the Blue Footed Booby of Galapagos, or is it smurf pr0n? Ad nauseum.

108:

One of my daydreams is that Freefall will win a Hugo. It's consistently one of the most intelligent things (of any kind) I read, and I'd love to see it get the recognition it deserves.

Agreed, Freefall fully deserves a Hugo. Now that I thought to check I see it's not even on the short list[1] and that both surprises and annoys me. I certainly don't begrudge Girl Genius or Digger their wins, nor Schlock Mercenary its multiple nominations - but Freefall is fully their equal as a SF story.

[1] If necessary, insert here Kevin Standlee's clarification of the word "nominee." I think most of us here know how the Hugos work.

109:

I'm gated from Google search by captchas when I use VPN, and have noted that I'm failing them more often eg pick which image is a fire hydrant, and keep clicking until none are left - I'm finding I get the 'Please try again' message and a new set of images more often than I used to. Either I'm actually artificial, and am just starting to feel the limits of my "mind", or some of the blurry street scenes from high up or in the distance have been tagged as having a fire hydrant even though there's no way a human eye can make it out at that resolution.

Then I started thinking about whether the captcha is also being used to train AIs to recognize objects encountered by self driving vehicles eg crosswalks, traffic lights, bicycles, buses, fire hydrants etc. ie we're providing them with confirmation on what these objects in the images are?

110:

Would an owl do? Apparently there's a car ad where they use chickens but I can't find an actual pigeon one.

https://www.youtube.com/watch?v=rylkcch5nrA

111:

The google system often seems to require more than one correct pick, it just don't tell you "you got that right, but since you're at harassment level 5 that requires you also get another 3 sets correct". I get the impression that there's a couter that covers stuff like using the wrong browser (not Chrome/Chromium), blocking ads, blocking persistent cookies, using psuedonymous mode, and VPNs. Using TOR/Tails sometime it just won't let me in at all... well, after 10 I gave up.

It's just a reminder that "free" means you're exchanging personal information for the service, and if you're not providing the information they're not providing the service. And you thought paywalls were a news media thing...

112:

That's a good point, but I was pondering on why all the captcha images are items you'd encounter while driving, and whether or not it's also being used to train self drive systems or something like them.

113:

Have heard this is actually an old ruse with various armies having used full scale balloons/toy tanks to mess with their enemies' intel gathering usu. via satellite.

Yes indeed! It's also handy to have plausible looking 'enemy' hardware for training one's own troops. There are even companies that will happily sell you, some random guy on the street, an inflatable rubber Russian tank (examples here and here), because what the hell.

It's also common to have decoys out in real battle spaces. Any missile or bomb that destroys a decoy is one the enemy can't use on real assets.

114:

Ah, yes, mapping software. We were coming back from my cardiologist Thursday, on our way for me to get my first vaccine shot, and had time to kill. Ellen finally just ignored her mobile after Every. Fucking. Time. The damn thing was *desperate* to get us on the Interstate, rather than straight down the road we were on. Every single turn, no matter what she said to it, or did, in the brief time at traffic lights.

115:

I am wondering what causes you not to be using parallax. I too have only ever had one functional eye, and I rely very heavily on parallax. Come to that, so do creatures which are often held up as exemplars of binocular vision; cats preparing to pounce weave their heads from side to side to get the parallax cues. I do the same thing for the same reason, albeit in different contexts.

I am afraid I have no answer to the "what causes" question -- that's how my vision always worked. I most definitely do not "weave head from side to side" to get parallax cues. And holograms do not work much for me either.

116:

Moz @ 111: "It's just a reminder that "free" means you're exchanging personal information for the service, and if you're not providing the information they're not providing the service."

I find that the Google robot is very nice with me. It doesn't bother me even if I use Firefox all the time and I never give it personal information. I use it for all kinds of things like search, translation or mail, through gmail.

I also use the YouTube robot, which is part of Google and doesn't bother me by asking for personal information.

It does do weird things though, like translating to French some of the "headers" of the English language videos it automatically selects for me.

Also, it seems to have guessed the first three characters of my postal code, in some way.

Haven't figured out how it did it, since I keep all personal information away from my computer, by doing things like doing all my personal banking offline, through my land line telephone only.

117:

I might add that my spatial imagination absolutely sucks. Putting together something mechanical without instructions is pretty much hopeless for me, and instructions in the form of a video are almost as bad as none at all. In fact, often the only way I can make sense of video instructions is to pause it every couple seconds, so that I am effectively studying a series of still images instead.

118:

I never give it personal information ... it seems to have guessed the first three characters of my postal code

Hmm. Could it be that you're giving it information via trackers rather than directly?

I listed a bunch of different things, and sadly for me the "is your browser unique" tester says not very many people at all use 2.5k monitors in portrait, so my canvas size alone makes me distinctive. Of course my tails VM is set to an internal desktop of 1920x108... the most popular size :)

119:

I was thinking the Google robot might have gotten the first 3 characters of my Canadian postal code through my IP address, but I have a dynamic IP address from my network supplier.

I don't mind if the robot knows that I have a 30 inch diagonal monitor set up in a CADD config but I never had to write that anywhere and I suspect that it can detect it somehow.

What I do mind is keeping my personal information private and the Google robot doesn't seem to care.

If Microsoft is tracking me throught my OS it doesn't show any sign of it.

120:

Then I started thinking about whether the captcha is also being used to train AIs to recognize objects encountered by self driving vehicles eg crosswalks, traffic lights, bicycles, buses, fire hydrants etc. ie we're providing them with confirmation on what these objects in the images are?

Yep. Why do you think so many of them are click all the squares with buses, bikes, cars, stop lights, etc...

121:

The damn thing was *desperate* to get us on the Interstate

There ARE options for routing in the app. At least in the Google Maps I use.

122:

Then I started thinking about whether the captcha is also being used to train

Oh, yes. That thought came to me quite a while ago. If captcha isn't using the responses to train an AI in what real Is see and do, it's missing a huge opportunity. That would be supplemental to the ostensible and perhaps real purpose.

Same, I suppose, as in any other situation in which real Is are being asked to make a choice.

123:

Also, it seems to have guessed the first three characters of my postal code, in some way.

Well let's see.

Your IP address in relation to known neighbors.

Your WiFi and/or WiFi nets you can "see" based on neighbors who have given themselves up.

And a big on most people don't realize is nearby bluetooth. I'm guessing if you have a Chromecast and it and your computer can both see the same devices it assumes you are at the address registered with accounts tied to the Chromecast.

And all of this is why Apple's next iOS has Facebook all in a fierce lather. You have to explicitly allow apps to see such things as nearby bluetooth and wifi devices. And they are making the default = "nope".

124:

but I have a dynamic IP address from my network supplier.

Dynamic doesn't mean it changes a lot. It only means it can change and you have to deal with it if it does. Spectrum (old TWC) IPs can stay put for years.

125:

WebAssembly is a idealised virtual machine code, which is designed to be easy to run[3] at high speed and to be a good target for compilers of traditional compiled languages[1]. It's power is in the limitation: it runs inside the browser sandbox, which means that it does not get access to most machine resources [2], unless specifically allowed[4]. One of these allowed resources is WebGL api (or even the nascent WebGPU api), which give direct access to the compute resources of the available GPU(s).

What this means is that there is potentially no loss of processing power as compared with native code, and there is heavy duty and extremely competitive R&D going into reducing any observed loss of power.

[1] Funnily enough, it's not (yet) good as a compilation target for scripting languages such as Javascript, which need garbage collection.
[2] It's achilles heel is that it can have unbounded access to CPU (and GPU?) resources, which means that in the wild it is heavily used for cryptocurrency mining.
[3] Different implementations may interpret (slowish), ahead-of-time compile to machine-code (potentially native speed), and/or Just-In-Time compile (slow start, gets quicker, may even exceed native code performance because of the ability to optimise on the fly for actual behavioural characteristics).
[4] For example, OS system calls are not allowed.

126:

There ARE options for routing in the app

Yes. And if you don't like the options presented then you can go to hell, as Whitroth found out.

I have spent far too much time with google maps trying to persuade it that just because I legally *can* ride on major roads doesn't mean I am willing to do that. I've had the Whitroth-style experience, where I am travelling down a nice quiet street with blockages to stop moronists using it, and cut-throughs to help cyclists. But google at every single intersection wants me on the major road parallel to this nice street. Yes, I can manually drag my chosen route back to where I choose at every single intersection. Well, except the ones where google has decided that I can't actually use the cut-throughs, and google decided that sure, you can ride to there. And then you can turn round and ride back to the last intersection...

City of Sydney have several full time staff working on bicycle intrastructure, and one of the things they do is follow through to make sure they things they tell google about cycle facilities actually translate into users of google maps being told about said facilities.

127:

"by exploiting flaws in the image recognizer attackers were able to steer a Tesla into the oncoming lane."

Those stupid silicon brains. My superb meat brain would never do that.

Back in 2000 Sydney hosted the Olympics. As a "celebration" of the ghastly event the powers that be decided to mark a permanent single unbroken blue line around the entire 40 km of the running race. It's hard to put into words, but the line took the most direct route. So where the runners go around a corner, on to a straight and then around a corner the other way, the result was a single unbroken blue line that drifted all the way from one side to the other. Despite signs everywhere saying "IGNORE THE BLUE LINE" and a bit of a media blitz saying the same thing people followed the blue line. The road rule is you can cross broken lines but can't cross an unbroken line. I was very nearly knocked off my motorbike by some dopey meat brain following the unbroken line and drifting into my lane.

So while putting down fake lane markings and getting Teslas to do the wrong thing might give you a feeling of superiority over silicon brains, putting down fake lane markings messes up the traffic no end already.

128:

"they also have an inventory of which types of accents/voices their AI needs to improve upon"

That would be the swearing accent...

I've heard tapes of them and it's generally someone telling the computer to fuck off, over and over.

129:

Yes. And if you don't like the options presented then you can go to hell, as Whitroth found out.

I just checked and the version on my phone has a switch for "Avoid Highways".

Plus my use of Google (or Apple) maps is rarely blind. I use it to get an idea of what I want to do then let it re-route as a generally follow the path shown. Plus I keep the voice for it turned off.

130:

Captcha was originally designed to do pretty much exactly that.

Ever wonder how you can Google a string of words and have it return some old manuscript?

https://www.wired.com/2007/05/recaptcha-fight-spam-and-digitize-books/

131:

I bow to your superior knowledge and understanding.

But, one thing, how do I get all the roads I don't want to cycle on designated as highways in google maps?

132:

David L @ 123 ans also @ 124

I don't have a Chromecast. In fact I don't even have a digital TV or other digital entertainment systems. I only have an ancient analog TV that I use to watch old VHS tapes, with no connection to any network.

My home office computer is connected without Wifi (by ethernet cable instead) to my ISP's router (in a kitchen closet), which is in turn connected with a fiber optic cable to my ISP's network.

About 15 years ago I systematically tested my ISP's IP traces when I was a regular contributor and did my many edits each day in Wikipedia (I was more active back then). I found that my IP was always changing. I never had the same IP twice.

Has the use of IPs by ISPs changed in the last 15 years?

133:

I suspect but don't know that better location accuracy data is now sold by at least some ISPs. I've seen dynamic IPs located to ~500m in Australia, and others convinced I'm in another state. It wouldn't surprise me at all to find that wifi users are located to 50m or better thanks to google et al's wardriving efforts (which google totally stopped doing, honest)

134:

Capcha images - ALWAYS purely 100% USA-ian & very locally centric.
Reverse parochialism - like US "fire hydrants" that AFAIK, no-one else uses & "crosswalks" for pedestrian crossings on roads,

Niala
Some of us don't have a TV at all .....

135:

Has the use of IPs by ISPs changed in the last 15 years?

Back in the day Bellsouth DLS IPs would move around weekly or daily. AT&T U-Verse seemed to move around every month or so when I had it in Texas a few years back. Verizon FIOS and Time Warner cable seemed to hold steady for months or even years.

It is all in how the ISP runs their network.

I wasn't giving an exhaustive list of how sites track. But they work hard at it. Someone up thread mentioned Am I Unique. (amiunique.org) It gives an interesting way that people are finger printed. And that information is correlated by the big behind the scenes firms to match up all kinds of other information about you. And the rest of us.

136:

«intuition would suggest that a WebAssembly framework (based on top of JavaScript running inside a web browser hosted on top of a traditional big-ass operating system) wouldn't be terribly fast»

Actually the web browser has a just-in-time dynamic code generator for WebAssembly (inspired by research on the language "Self" by Ungar and Wolzcko) and it has been shown that means that WebAssembly runs at only 2-3 times slower than *native code*.

«an explosive series of breakthroughs in the field misleadingly known as Artificial Intelligence. Most of them centre on applications of neural networks,»

That is "misleadingly known" as "Machine Learning", which is actually "Machine Training", where the training is to do cluster analysis. and "neural networks" are in essence a rather opaque form of decision trees built after that cluster analysis.

«the ghosts know them as such but don't understand the concept of lying (because they are a howling cognitive vacuum rather than actually conscious)»

It is no about consciousness, it is that the decision trees, oops I mean the neural networks, don't "understand" the cluster analysis they do, they are not "sentient", they lack the ability to introspect what they are doing.

There is a secondary issue with them: they require gigantic data sets because usually those data sets are very "noisy", and the clusters they form are very fuzzy. This matters a lot in part because of the previous issue, that since they don't "understand" the clusters, they cannot spot the obvious noise.

This said, they are still hugely useful. Butg there is a very large limitation looming: fundamental hw tech has stopped improving that much:

«NNs, and in particular training Generative Adversarial Networks takes a ridiculous amount of computing power, but we've got it these days»

Both CPUs and GPUs have largely reached the top of the S-curve and now "improve" mostly by becoming bigger and more power hungry, rather than denser and faster. We have reached the point that aircraft reached in the 1950s, where jets from that time are not that different from jets of today, rather than being dramatically better than the biplanes of the 1910s.

138:

No; I said a galaxy full of comptronium, and meant just that. Four acres of Crays and a dedicated nuclear power station isn't going to be any more use than a BBC Micro.

139:

Most of the function of your semi-circular canals is to allow your eyes to stay still (or track moving objects) while your head is moving; I learnt that while my balance was being tested, and I can't do it at all.

140:

Yeah. Cyclestreets (in the UK) is no better - it LURVES psychle farcilities, loathes roundabouts and assumes every cyclist can lug their bicycle up or down stairs without trouble. I might get around to seeing if I can volunteer to improve that, but it's a fair hassle; I usually just look at a map and make my own decisions.

141:

Greg Tingey @ 134: "Reverse parochialism - like US "fire hydrants" that AFAIK, no-one else uses"

Change the US to "North America" and your phrase will be true. Canada has fire hydrants too. Unless we have "Une borne-fontaine" of course.

https://fr.wikipedia.org/wiki/Borne-fontaine

I love to see the way artificial intelligence applications from the US do bloopers or simply strange things with multiple languages. In contrast big Canadian banks get it right the first time.

142:

As to AI, "Machine Learning" is not it. If AI arises it will be from drones etc.

As to SciFi, there is a the delightful "Distraction" by Bruce Sterling, where a plot feature is that in "the future" spam bots to get past spam filters have to pass the "Turing Test" doing conversations with the target users, and thus have become first sentient and then self-aware, and spam bot networks use part of the money they make from spamming to hire mafia contract killers to take out the system administrators who try to shut them down. :-)

143:

David L @ 235

amiunique.org was great fun. I'll have to return to it to do some serious reading.

144:

Niala @ 132: I found that my IP was always changing. I never had the same IP twice. Has the use of IPs by ISPs changed in the last 15 years?

I don't know your background, so please excuse me if this is stuff you already know. I'm erring on the side of explaining too much rather than too little.

Most ISPs allocate IP addresses using DHCP (Dynamic Host Configuration Protocol). Basically, when your computer joins a local network it sends out a special ping, and a DHCP server replies with a suitable IP address and other configuration data. The address comes with a "lease" saying how long you own it for (usually 24 hours). If you are still on-line after 24 hours then your computer will renew the lease and you carry on with the same address.

If you go off line and the lease expires then your address can be allocated to someone else. So when you come back you get a new address. DHCP servers generally try to give the same address you had last time, but there are no guarantees.

If you have a home router (which most people these days do) then there are two IP addresses involved. One is for your PC or other device, and the other is for the router. When you turn the router on it sends the DHCP ping packet to your ISP, which allocates it an external IP address. Your devices on your home network then send DHCP pings within your home, and your home router responds with internal IP addresses. The router then translates between internal and external IP addresses (known as Network Address Translation, or NAT).

If you check your IP address in your computer's network settings you will probably see something starting "192.168...", or possibly "10...". This is an internal IP address and will be different for every device in your home. If you Google "What is my IP address?" you will get the external IP address allocated for your router and seen by the rest of the world, and that will be the same for every device in your home.

I'm assuming that it is the external IP address you are asking about.

Because the external IP address is leased to your router, it only changes when your router is turned off long enough for its current lease to expire. As long as your router is turned on it will keep renewing its lease and your IP address will not change. Most people leave their routers on 24x7, so the IP addresses don't change unless there is some general outage or they go on holiday or something.

15 years ago it is just possible you were using dial-up or something like it. In that case there was no router and a different IP address was allocated for every time your modem called the ISP.

If you have no home router and you are just plugging your PC directly into your modem then you need to double-check your firewall settings, or better yet get yourself a home router that is set up for proper security from the get-go. You can tell this is the case because the IP address you see in network settings is the same as the IP address you get from Google. Many ISPs provide combination modem/routers in a single box, so it's not immediately obvious from the wiring.

145:

"[re]captcha is also being used to train AIs"

Absolutely. That's what it's for. (What it's really for, as opposed to what they say it's for to con thoughtless people into failing in civic responsibility by putting the bloody thing on their websites.) Which is why I refuse to ever fill the things in. The effect of an individual boycott may be trivial, but at least I can avoid actively helping to make things worse.

146:

My spatial imagination is very good, but instructions in the form of a video are completely useless. I want text and still images. A video provides much worse quality images than still photographs (in the sense of being shot from the best angle to make things clear), it keeps moving so I can't see what's going on, and it has no random access so I have to buffer large stretches of it inside my head.

147:

David L notes: "I more and more think it's just that we (wet meat) have a decade or five of experiences that our brains sort through to interpret the visual world. And as we wander further off that pile of "memories" we get worse and worse results. So training an AI with 100K pictures still fall short of our 10 to 50 years of observations.

Not sure that's correct. There's generally a point at which the training is "good enough" and further training doesn't help (nor does it necessarily hinder). As personal anecdata, I offer my experiences learning to identify diverse Canadian tree species. In summer, I did pretty well because leaves were available in addition to all the other clues. But our clever bastard of a prof held the identification exam in mid-winter, mostly with large mature trees, so there were no leaves and the branches way too high overhead to see details. You couldn't pass unless you could integrate a sufficiently large number of factors to definitively screen out near misses. Took me 3 tries to pass the test that year (combination of natural inaptitude [sic] and high within-species variability in Canadian forest trees). I eventually learned to be a Bayesian without knowing that was what I was doing. *G* More importantly, I learned the value of multiple diagnostic criteria.

Now, 40 years on, there's a large subset of trees I can identify "instinctively" (without thought) and a smaller subset for which I still need to deliberately run through the list of criteria before I'm sure. When I travel abroad to places with tree species like those in Canada, I can identify many trees to at least the genus level without much thought, and sometimes to the species level. When I travel further afield, I have no problem identifying "it's a tree" with near-100% confidence. Going beyond that level depends on whether a given novel species interested me enough to learn what it looks like before my travel. (Nowadays, I oscillate back and forth between "must know everything to the species level" and "fuck the taxonomy and just enjoy the tree for its own sake".)

So I suspect that "weak" AI (e.g., trained neural networks) is about learning enough, not repeatedly relearning variations on the same pattern for 50 years, whereas "strong" AI (human cognition) relies on escalation: start with the learned "weak" image recognition, and escalate to conscious analytical cognition if the weak recognition fails or doesn't return a sufficiently high confidence in the conclusion. That escalation is where your "decade or five" is really essential: it provides many years of contextual information that supports the weak recognition and allows plausible inferences.

148:

Paul @ 144: "I'm erring on the side of explaining too much rather than too little."

You haven't explained too much.

15 years ago I was probably getting different IP addresses because I was shutting down my home office computer and the power bar too, at the end of each session on my home office computer.

A few weeks ago I had a digital Bell Hub 3000 router installed, along with an ethernet cable (no Wifi for my home computer) and, I suppose, proper security installed from the get-go.

I'll be checking some things in the next few days. Thank you very much for all this information.

149:

I find instructional videos occasionally useful, for example when trying to dismantle electronics, but written instructions with pictures would usually work, too. For learning how to make music I find videos very good - for me it's very different seeing the action performed than just seeing the still images and reading text.

However, for many things, video is just useless. One recent example is this one retro-computer I'm interested in, and for which I've even done some programming (on an emulator, it's not ready in HW form yet). Somebody has made multiple videos about programming it, and somebody (maybe same person) made a video about the different graphics modes. The problem is that it's very hard to learn how to program a computer in assembly language by watching a video, specifically because no random access and no way to really reference stuff back.

The video mode thing was also completely baffling to me: it would've been easier to make a simple page and a table to explain everything and it could've been used as a reference, now the video is just basically promotion with no real use.

At least some people have been writing guides for it - I have this plan of making some graphics stuff with it and then write some kind of guide for it. Not making a video.

150:

The IP persistence is also dependent on the ISP policy. Some ISPs used to reset explicitly all leases at least once a week and sometime every night (I think because they were selling stable IP as a pro option). I has become somewhat rare.

151:

«So I suspect that "weak" AI (e.g., trained neural networks) is about learning enough, not repeatedly relearning variations on the same pattern for 50 years, whereas "strong" AI (human cognition) relies on escalation [...] many years of contextual information that supports the weak recognition and allows plausible inferences.»

I guess it is indeed the ability to introspect on the training, but ultimately it is pain and evolution. In particular the pain from "hunger": if we don't learn right, if we mistake an apple with "iPod" written on it for an iPod, we get painfully hungry and eventually die.

BTW a little known detail which I think is widely ignored in AI research too, is that years ago a convention of AI researchers agreed on some ethical principles, and one was that not to "motivate" AIs with simulated "pain". Try to imagine ruthless "winners whatever it takes" startups creating sentient, conscious AIs at some point and "motivating" and controlling them through torture.

152:

Training dogs is often this way, initially you have to teach some of them "sit (in the front yard)", then a completely unrelated trick "sit (in the back yard)" and so on until finally their brain does a little dance and says "I can sit anywhere" and life gets a lot easier.

Any high school teacher can tell you teenagers are the same way. Something they learn in one course isn't necessarily transferred to another course. The idea that spelling and grammar don't count except in English class, for example… :-/

On a more serious note, many people's mental model of the world is surprising resistant to change, which necessitates Ptolemaic levels of epicycles to function. For example, the average person's model of motion is Aristotelian*. They might learn Newtonian mechanics in physics class, but that is seen as 'something for class' rather than a replacement for part of their mental model of the world. Lots of work has been done (by Mazur and others) on various methods to eliminate deep-rooted misconceptions — persistent incorrect mental models — and there isn't a single method that works reliably**.

I'd like to think that physicist and engineers replace their Aristotelian models with something else, but at a conference last decade I was chatting with a university professor who was pleading with high school teachers to focus more on concept and less on calculation, because he had graduate students who could perfectly calculate the right answer but had no idea why it was the right answer — it was the right answer because they used the right method and didn't make any mistakes. His argument was that high school (and undergraduate university) education just added more epicycles to an incorrect model, when we needed to be replacing the model.

*TLDR: things move because they have impetus. When they run out of impetus, they stop moving. Essentially, friction is treated as part of the object rather than a separate force.

**Mazur used to think that confronting the misconception with evidence that they directly see/experience worked, but he reanalyzed his data and decided that his earlier work didn't support his conclusions. One of the reasons I consider him worth listening to is that he is willing to admit his mistakes.

153:

Cyc is inherently doomed to failure, if the Wikipedia page on it is a fair statement of its goals, namely

... to assemble a comprehensive ontology and knowledge base that spans the basic concepts and rules about how the world works.

This approach is so rooted in incorrect theories of knowledge that it's Not Even Wrong. It can't possibly represent "common-sense knowledge about the world." Attempts to map how simple nematodes learn the things they are capable of learning will be more fruitful, though if the Cyc people can figure out a way to piggyback off an engine that has as much "common-sense knowledge" as a rat does, their work might eventually be worth something.

The first artificial system to have "common-sense knowledge about the world" will be some combination of "model of natural biological neural system" and the equivalent of a billion years of natural selection. And its representation of that "common-sense knowledge" will be an even more impenetrable, incomprehensibly arbitrary-looking mess than the "rule sets" of a cucumber-sorter.

154:

Not sure that's correct.

What I was implying but did state clearly is that our decade(s) of experience give us all kinds of data about what in a specific AI training case is background noise. In the tank situation we KNOW what trees and snow are and just ignore them when picking out tanks.

Basically our life experience fills us with all kinds of information which can be used to be able to ignore things we don't need to deal with at any particular moment.

And it mostly works.

155:

Because the external IP address is leased to your router, it only changes when your router is turned off long enough for its current lease to expire. As long as your router is turned on it will keep renewing its lease and your IP address will not change. Most people leave their routers on 24x7, so the IP addresses don't change unless there is some general outage or they go on holiday or something.

Sort of. When a DHCP lease expires the IP address CAN change. Most of the time it doesn't but it can. Which is why when someone (me) needs to reallocate IPs I set the DHCP lease times down to an hour a few days ahead of time then 5 minutes just before the changes roll out. Then set the leases back to something more useful and not as networking stressful after 20 minutes or so.

Much of the more rapidly changing IP situations came from ISPs moving from dial up modems (where you likely got a new IP with each dial in) to fixed line situations. So a lot of early DSL and cable setups would have IPs that changed daily or even more often. Over time as dial up has gone away most network just leave the IPs in place unless there's a power off or reason to re-align the network. Of course some of the big ISPs change them around every week or so just to make it harder for people who aren't paying for a static IP to pretend they have one.

156:

The interesting thing as Charlie mentioned is the 16-core Neural Engine. Lot of chip space dedicated to something no one uses much at this time.

You're joking, right?

Apple is all over neurocomputing. From FaceID for unlock/face recognition on iOS, to face recognition in Photos (it can index your photo library by recognizing who's in each photo, with minimal training), to computational audio stuff cropping up in Homepods, and speech recognition taking place on the local hardware rather than in the cloud: the car project is just more of the same, on that level.

When the Apple Stores reopen treat yourself to a visit to play with an iPhone 12 Pro's camera and boggle at what it does by way of computational photography using the pocket version of the neural processor. It's impressive.

You're right that third parties aren't using the neural coprocessor much so far because it's a relatively recent capability and the APIs were only published a couple of years ago at most. But it's something an increasing amount of Apple hardware leverages.

157:

I work with people who do this stuff for a living, and see two systematic errors made in the "tanks" story.

First, the NN is being allowed to train itself against the whole image; normally, you'd do some sort of preprocessing steps to (e.g.) mark out the area that contains the tank versus the background. Unless there is literally no known algorithm to do the basic segmentation you want predictably, you'd apply limited preprocessing to make the NN's job simpler.

Second, their training set was really badly chosen; you want your training set (and the test set, plus your dev set) to be representative of real data from the field, not a carefully curated set; ideally, if you gave your training data, test data and real data (all unlabelled) to an intelligent human, they would not be able to tell you which set was which.

In ML domains, "supervised learning" refers to telling the network what the correct output should be for each item in the training set; there's also semi-supervised learning (tell it about a subset of the training set, but ask it for labels for the entire training set), unsupervised learning (get the network to split the training set into groups, it chooses arbitrary labels for the groups it finds), and reinforcement learning (give it a score from 0.0 to 1.0 each time it gives you an answer, it aims to get a high score).

Which technique you use depends on what you have available to you, and what your goals are for this network once trained.

With that said, the most important thing about ML as it exists today is that machine learning is about having the computer work out how to classify inputs. It's not as sophisticated as the name might have you think - it literally just refers to the computer working out the correlation between input and output for itself, given a human-controlled set of inputs.

158:

She tried. Both onscreen, while we were at a light or two, and by voice. It ignored her.

159:

I really hate most "instructional videos", and having to sit and listen to them blather on and on, until *finally* I see the part I want. Now, taking apart an old external storage box is fine with video... but ANYTHING involving command line is *stupid*, nor do I want screen shots.

160:

Sorry, I must have missed something "CYC?" What does this stand for?

162:

Interestingly, I had the opposite problem when I visited Cornwall. Google kept sending me down these narrow lanes (barely wide enough for a car, and you may have to back up half a mile if you meet someone coming the other way). There was not function to say "please stick to 2 lane roads"

163:

You're joking, right?

Nope. Your examples are all early stages of NN. To me putting all those cores into the M1 was a way to boost things to a level that something like autonomous driving could use.

164:

As I said, to me mapping apps on a phone are a great way to see how to get somewhere new or see how traffic is doing. I have used it a lot over the last 5+ years when in the Dallas Fort Worth area. Most of the time I KNEW where I wanted to get to but wasn't sure of the best way there. Especially with the massive road construction happening there for the last 10 years.

I'd ask for a routing, look at where it wanted me to go, then head off. With the sound off but the phone mounted high up. (I have a decently large screen model.) Many times I'd not be on the exact route in the app but as long as I was not headed in the opposite direction I was reasonably assured I'd get to where I wanted. And this only works if you use the apps in map view, not turn by turn.

I do the same thing on my annual 7+ hour drive to and from Penn State. And I even use it around home to make sure I'm not headed into a traffic standstill from a wreak or construction.

And on vacation drives.

These apps are a long way from being as good as a smart person. And I'm not expected perfection or anything close to it.

165:

Cyclestreets ... assumes every cyclist can lug their bicycle up or down stairs without trouble.

As a long time Wapsi Square reader my immediate reaction to bicycles and stairs is "Look at it go!"

(And "Good thing I was wearing my helmet.")

166:

Paul @ 5: followed the link to the video. I'm having a problem with my PC sound at present, so I turned on subtitles. The subtitles have obviously been generated by a voice recognition algorithm, and were wrong enough that I couldn't quite follow what was going on, but I think this is a video about problems with voice recognition in an elevator that can't understand "eleven" in a Scottish accent. However the captioning algorithm seems to have no problem with that bit.

I'm going to fix my audio problems and come back to this.

I went back and turned on the subtitles "English (auto-generated)" to watch it again and it's even more horrifying [and hilarious].

168:

"Any high school teacher can tell you teenagers are the same way. Something they learn in one course isn't necessarily transferred to another course. The idea that spelling and grammar don't count except in English class, for example… :-/"

Not just teenagers. My wife used to be an art teacher, for adults, and the number of students who assumed that once they had completed the Foundations course they could safely disregard all they had learned for subsequent courses was staggering. Mostly twenties and thirties. What part of "Foundations" did they not understand?

She largely blames "teaching to the exam" (not by her, of course, but a pervasive part of formal education) for it. Students are encouraged to view getting good marks in the exam as the be all and end all of their studies, and once they have those marks, it can all be forgotten.

JHomes

169:

Charlie Stross @ 38:

Ah, good. I was thinking it might take an awful lot of plot to get from the close of Dark State to a satisfying ending

Pretty much. When I checked the copy edits I gave it a thorough read. It's going to be a 450-500 page brick, but it reads fast and twisty.

Fortunately I have all of the preceding volumes for the series in hard cover. It may be weird, but I prefer to have all of the volumes of a series in the same format.

It's been a bit of a struggle for the Laundry Files, because my initial purchase was a paperback. I'm glad the U.K. editions are still available as paperbacks.

170:

My Google maps story: I had to pick up an item from an unfamiliar shop in a suburb I had reasonable knowledge of. I had a vague idea where the shop was, but decided to use Maps to home in on it. When I got close I was heading east and the shop was somewhere on the southern side of the road (we drive on the left here so I was on the north lane of the road). For some reason, instead of Maps telling me I'd reached my destination, which was on just the other side of the road, it decided I needed a 20 minute detour along a main road, a left onto the freeway to the city's foothills, next down an exit to a landmark pub to turn round, then back onto the freeway and down again. There were plenty of ways to get from one side of the road to the other within 500 meters of the shop but Maps had decided that was the only one...

171:

What part of "Foundations" did they not understand?

Calculus would have been a disaster for them. I did electrical engineering in college. After a while we all decided it was really calculus 402, calculus 573, calculus 475, etc. No matter what the course title was on the door. The first 4 semesters from the math department were just a warm up.

172:

Twice on a drive from North Carolina to Pennsylvania I told Google Maps to take me to a Costco to get gas. Both times it told me "You're here". While I was on the interstate. Both were located next to the exit and I guess I was close enough.

173:

On the whole, I've been favorably impressed with GPS vehicle navigation, but you have to keep your brain engaged the whole time. My personal story: While driving to a mountaintop village in Sicily a few years ago, the software kept telling us with increasing vehemence that we had to turn right ***immediately*** -- over a 100-m vertical drop with no turnoff point. Fortunately, my wife and I cooperate on the navigation: one of us consults the map while the other drives, and provides human voice instructions, with the GPS sound turned off.

Having disobeyed the software and arrived safely at the top, my wife and I agreed that I would handle the downhill leg. I learned standard transmission very late in life, so I have an occasionally adversarial relationship with first gear that makes difficult uphill stretches... difficult. So she drove us to the top. On the other hand, she's not comfortable with long, windy downhill stretches, whereas I am. (Jack Sprat and his wife! Coupledom works well when your weaknesses are complimentary.) Unfortunately for me, I neglected to pay close attention to the scale of the display (which had changed without me noticing), and what appeared to be a long series of gentle downhill switchbacks turned out to be a seemingly endless series of very tight* switchbacks and multiple consecutive steep descents that were hidden between the long and gentle loops that were what we saw on the screen. Oops! I managed to avoid burning out the brakes and clutch, but the car probably didn't much like me.

* Apparently, someone had decided that a medieval goat track was a good choice for a 20th century paved road. For many of the switchbacks, I had to back and fill more than once to get around the corner. Given that there was no room to turn around, and that backing uphill into oncoming traffic for an hour would have been suicidal, I opted for the lesser of two evils and continued my descent. Oh, did I mention that we were driving directly into the setting sun for the last half of the descent? Thankfully, the locals were smarter than me, and none of them attempted to climb the mountain while I was descending.

174:

Fortunately I have all of the preceding volumes for the series in hard cover.

The first six books were the first editions, I assume. The omnibus re-edit was never issued in hardcover, but is my preferred version (I did a lot of work to even out the pacing and make it read more smoothly).

In the USA the Laundry Files have iterated through three publishers, and fell out of mass market paperback at Ace after book 6, "The Annihilation Score", as ebook sales took off and mass market paperback sales fell off a cliff. The UK ... hasn't had mass market distribution since 1991: all paperbacks are trade paperbacks, and Orbit can reprint books profitably in runs of as few as 200 copies, so they're all in print in a uniform edition.

175:

Scott Sanford @ 108:

One of my daydreams is that Freefall will win a Hugo. It's consistently one of the most intelligent things (of any kind) I read, and I'd love to see it get the recognition it deserves.

Agreed, Freefall fully deserves a Hugo. Now that I thought to check I see it's not even on the short list[1] and that both surprises and annoys me. I certainly don't begrudge Girl Genius or Digger their wins, nor Schlock Mercenary its multiple nominations - but Freefall is fully their equal as a SF story.

[1] If necessary, insert here Kevin Standlee's clarification of the word "nominee." I think most of us here know how the Hugos work.

I'm one of those who doesn't know how the Hugos work, but looking at the Wikipedia article it looks like all of the nominations were for printed stories. Is Freefall even available in a print edition?

176:

Instructional videos can provide much better information than drawings, written instruction's and photographs. By the end of my career in clinical biochemistry they made operation, servicing, troubleshooting and repair of very complex instruments much better. Everything had textual descriptions diagrams and photographs of the relevant procedures with videos available as buttons on the touch screen. Error codes would prompt the appropriate screen complete with videos which usually only had to be used once by inexperienced operators. The converse worked as well. Using smartphones to send videos and/or photographs to helpdesks made their job more efficient. Saying videos are bad is like saying books are bad. It depends on the content and the skill of the instructors.

177:

Regarding the Granny Smith/Ipod thing. One photo shows an eating apple, sure. But the next shows a white rectangle, with iPod written on it. The fruit behind it is almost completely hidden. I'm not convinced that this really proves anything.

178:

My home office computer is connected without Wifi (by ethernet cable instead) to my ISP's router (in a kitchen closet), which is in turn connected with a fiber optic cable to my ISP's network.

Many of the apps that want to mine your personal information now do things like query Bluetooth and WiFi to see who is nearby. And based on those results many times they can discover where you are by who your neighbors are. Or maybe your cell phone has bluetooth on for earphones and since your cellphone is near your computer 8 to 16 hours a day, they figure the computer is "yours".

Apple's latest iOS requires the user to allow an app to do such. And macOS 11 (Big Sur) has a lot of such permissions now for the user to allow or not. And this has gotten Facebook's knickers all in a huge twisted knot. But as to whether or not Facebook in a web browser running on Windows 7 can do that is something I thankfully don't have to worry about.

179:

Jack Sprat and his wife!

Glad it works for the pair of you. My wife would never be an engineer. When telling me directions she will regularly swap left and right in her directions to me. And use "right" when "correct" was needed.

And with Google Maps, in very large indicators they show how far to next "thing", an indication of what this think is (merge, left, etc...), and a name of the thing. I tell me wife in no uncertain terms I want them in that order. How far is most important to me in terms of decision making, and so on.

Many times she will say "Windward Ave ahead". And that's it. ARRRRRG.

I like maps. She wants turn by turn and doesn't get me at all.

180:

Perhaps you should go back and reread what I wrote.

I did say things like "I HATE", and "for certain things".

When all you're dealing with is menus and buttons, a video can do; when you're doing something moderately complicated, say, setting up a Linux RAID and need to break the RAID to repair a failed disk, there's NO MENU, and doing it on the command line says "a video and screenshots are a pain in the ass, rather than typing out what do do in what order".

181:

Pay-walled

Worked for me, with no paywall.

182:

David L @ 178 "Or maybe your cell phone has bluetooth on for earphones..."

I don't have a cellphone. I only have a land line. And I have no bluetooth device anywhere for anything. In fact I personally have no equipment using radio frequencies for anything.

However, my employer has entrusted me with an Apple iPhone XR, which stands to my left, gathering dust because no one calls me on it and I do not need to call people at the office by phone. I find a combo of email and MS Teams to be more efficient.

My employer has also entrusted me with a Microsoft Tablet PC, which I use daily for all my daytime office needs: MS Teams, email, etc.. It stands to my right.

To my knowledge that Tablet PC is the only device in my home which uses a Wifi connection (to my ISP's digital router in the kitchen closet) since my personal PC is linked instead with a long ethernet cable to the kitchen closet and has absolutely no Wifi device in it or on top of it.

In theory (at least that's what I wanted) my personal PC and my employer's Tablet PC absolutely cannot talk to each other since there is no physical connection or Wifi link for them to abuse.

In practice could they be talking to each other inside the digital router in the kitchen closet?

Those modern robots are sneaky things!

183:

AuntyJack @ 109: I'm gated from Google search by captchas when I use VPN, and have noted that I'm failing them more often eg pick which image is a fire hydrant, and keep clicking until none are left - I'm finding I get the 'Please try again' message and a new set of images more often than I used to. Either I'm actually artificial, and am just starting to feel the limits of my "mind", or some of the blurry street scenes from high up or in the distance have been tagged as having a fire hydrant even though there's no way a human eye can make it out at that resolution.

Then I started thinking about whether the captcha is also being used to train AIs to recognize objects encountered by self driving vehicles eg crosswalks, traffic lights, bicycles, buses, fire hydrants etc. ie we're providing them with confirmation on what these objects in the images are?

Often the recaptcha doesn't KNOW which images have the fire hydrant, and even if you DO select them all it won't accept the results.

184:

Niala @ 119: I was thinking the Google robot might have gotten the first 3 characters of my Canadian postal code through my IP address, but I have a dynamic IP address from my network supplier.

Probably the block of IP address you're drawing yours from is mapped to a specific area, so it knows you're somewhere inside that area.


185:

Interestingly, yesterday I ran across this video on YouTube about tech failures and a couple of them were AI related.

Teaching a twitter-bot to hate:

Micro$oft's TAYTWEETS: https://youtu.be/QMWlRWnAZH8?t=122

Is the AI secure from hackers?:

Amazon Ring: https://youtu.be/QMWlRWnAZH8?t=627

186:

Students are encouraged to view getting good marks in the exam as the be all and end all of their studies, and once they have those marks, it can all be forgotten.

Not just students.

On the weekend a former colleague phoned to gnash her teach on my shoulder. She's teaching language at Virtual School in Canada's largest school board.

According to her department head* it's not that important that teachers cover the full curriculum this year, because 'unlike math, learning a foreign language doesn't really build on previous knowledge'. (For context, this was a department discussion about dealing with students who are functioning 4-6 years behind grade level, and what marks/credits to give them).

Admin is concerned about students attending class by numbers only. A student logging in for five minutes of a 2 hour class is to be considered present, and NOT to be referred to admin or attendance counsellors as long as they don't miss three classes in a row.

Teachers have been told by the superintendent to concentrate on marks when talking to parents rather than discussing what the kids are learning.

Admin is mostly concerned about what a teacher can do to boost a child's mark to 50%, because that means they have earned a credit. Discussions about filling in gaps in knowledge or skills is met with 'what can you do to raise their mark?'.

Soooo glad I'm retired. In the current climate I'd probably end up like that Edmonton teacher fired for giving kids zero on assignments and tests they refused to do.


*Who is 25 years old, so has only 2-3 years teaching experience

187:

David L @ 121:

The damn thing was *desperate* to get us on the Interstate

There ARE options for routing in the app. At least in the Google Maps I use.

But it will fight you if you want to substitute a route for the one it recommends. Every time I try to find something in Raleigh or Durham Google Maps tries to route me around the Beltline. And if I'm going west, I prefer to go US 64 through Asheboro & pick up I-40 west of Mocksville (avoiding the Greensboro-Winston Salem MESS). But getting Google Maps to accept that as part of the route is definitely anathema.

You'd think that as many times as I've done it that it could learn.

188:

Probably the block of IP address you're drawing yours from is mapped to a specific area, so it knows you're somewhere inside that area.

Oddly enough, I often find those 'targeted ads' displayed on web pages tell me about cheap dentists in Cornwall and singles in Ottawa*, when I live adjacent to Toronto. (For non-Canadians, Ottawa and Cornwall are 5-6 hours drive from me, at highway speeds.)

Maybe comics.com is lying to their ad providers about customer locations?


*Leaving aside the issue of the products being offered — I'd be a lot more bothered if they were relevant rather than random.

189:

JBS @ 184: "Probably the block of IP address you're drawing yours from is mapped to a specific area, so it knows you're somewhere inside that area."

Yeah, you're right and it's illegal to do that in the first place since Canadian postal codes can't be split in two like they're doing. It's an absolute ANA NAN Alphabetical Numeric code. Like K1A 0A6.

And in the second place I don't want them to have even a part of my geo localisation.

But since I posted that @ 119 I've had some suggestions on how to proceed further. It turns out that all I had to do to get my IP address (instead of making a change to Wikipedia without my normal identity) was to type "What is my IP address?" in Google.

And just two entries below the Google result on the page there was a link to a company I'd read about, nordVPN,com. I clicked on it and it gave me a map-based indication of where I was in Canada, even more precisely than that half-postal code.

I'm just starting to discover what I'm leaking and what I'm not leaking.

190:

See figure 14 in this: https://distill.pub/2021/multimodal-neurons/
The laptop, rotary dial telephone and plant pot are clearly not what their labels say.
Also, the rest of the article is interesting.


191:

Charlie Stross @174:

Fortunately I have all of the preceding volumes for the series in hard cover.

The first six books were the first editions, I assume. The omnibus re-edit was never issued in hardcover, but is my preferred version (I did a lot of work to even out the pacing and make it read more smoothly).

In the USA the Laundry Files have iterated through three publishers, and fell out of mass market paperback at Ace after book 6, "The Annihilation Score", as ebook sales took off and mass market paperback sales fell off a cliff. The UK ... hasn't had mass market distribution since 1991: all paperbacks are trade paperbacks, and Orbit can reprint books profitably in runs of as few as 200 copies, so they're all in print in a uniform edition.

I'm pretty sure the first of your books I read was one of the Merchant Princes series. I remember the store I found it in; one of the outlet stores off of I-95 at Selma, NC. I was on the road & stopped there for some reason. I remember being attracted by the cover art. I often buy books while traveling just to have something to do with my mind while I'm over-nighting in a motel.

But here's where things get confusing ... the cover art that attracted my attention was for "The Clan Corporate", the third volume in the series.

I'm guessing I may have bought it while I was headed home & put it aside until I could find & read the first two volumes, which I'm fairly certain were in the public library.

192:

They were being taught in electrical engineering about 1990 where I was, but were apparently unknown to the philosophy department. It made my course on "computers, minds, and logic" very confusing because the lecturer kept saying "this is impossible" about things that I knew were being done. He kept at it even after I started bringing him copies of papers from the people doing the things.

I tutored that class just a few years after you took it, and I find your claims surprising. I suspect a miscommunication between you and your lecturer.

Discussion of neural networks (they called it "connectionism") were as trendy as all hell in philosophy depts in the late 80s/early 90s because the rebellion against symbol-manipulation AI raised such interesting philosophical issues about "meaning".

And I believe the course supervisor in the philosophy dept for that course was Jack Copeland. He certainly knew his stuff about neural nets. His textbook on AI wasn't published until '93, but it covers them pretty well.

Jack supervised my honours thesis, which was on Turing's 1948 paper describing a precursor to the idea of a neural network. Turing called his proposed network "Unorganised Machines" and proposed a multi-layer network based on NAND gates, starting with random assignments of nodes, which would be trained - but he proposed no training mechanism.

193:

And just two entries below the Google result on the page there was a link to a company I'd read about, nordVPN,com. I clicked on it and it gave me a map-based indication of where I was in Canada, even more precisely than that half-postal code.

Before you buy any VPN service understand that the cheap ones (and some of the not so cheap) make money by harvesting your information and selling it. A VPN service, while blocking all that information to others, gets to see it and keep it all to itself. So if you're concerned about such you want to find a VPN that agrees to keep your privates private.

Read here:
https://www.nytimes.com/wirecutter/reviews/best-vpn-service/
especially the details about service agreeements.

As to is your personal PC leaking information even though you are not using it does it have WiFI or Bluetooth hardware? And if your MS Tablet has such turned on your neighbors might know it is there. And yes all your LAN devices can (and in many ways must) see each others MAC addresses which allows them to be related together on your NAT'd LAN. And that tablet might have GPS on it.

The concept of a network and connected devices makes it hard for things to not see each other. It can be done but isn't for the typical home setup. At all.

Oh, yeah, that phone if turned on most likely Geolocates (911 or equivalent in Canada) and likely has bluetooth and wifi on it. And even if not in use it and the MS Tablet and maybe your home computer can notice each other.

194:

And I told my neighbors about this a while ago.

An Apple TV became visible to me named:
John and Jane's love parlor

She had no idea that the name would be visible "over the air". It was quickly changed.

(Names changed to protect the guilty.)

195:

I don't recall the exact point being made, but one of the papers was on the MIT "cockroach", something like the link below, and I recall being shocked that the lecturer would state flatly that it was impossible for it to work. Perhaps he meant some variation on "sure, ANNs exist but they're useless" but I recall feeling obliged to got back with more references and getting a negative reaction.

https://www.researchgate.net/profile/Roger-Quinn/publication/220499553_A_Distributed_Neural_Network_Architecture_for_Hexapod_Robot_Locomotion/links/561bcd2608ae6d17308b077a/A-Distributed-Neural-Network-Architecture-for-Hexapod-Robot-Locomotion.pdf

196:

Also, this list is maintained (updated today):
Does your VPN Keep Logs? 120 VPN Logging Policies Revealed (Aimee O'Driscoll, UPDATED: March 8, 2021)

197:

I'm one of those who doesn't know how the Hugos work, but looking at the Wikipedia article it looks like all of the nominations were for printed stories. Is Freefall even available in a print edition?

*facepalm* You make me realize the obvious.

Physical print is not necessary; the xkcd story Time has never been put onto paper as far as I know and it got a Hugo in 2014 - but the story is finished. As you can read on the Hugo FAQ list works are eligible the year they come out. (This is why you'll see lots of individual Doctor Who episodes nominated but never Doctor Who as a series. Also of interest is the 1968 Hugo for Best Dramatic Presentation, a five-way runoff between different Star Trek episodes.) Works spanning multiple years count for the year they end, as with Digger; we hope Freefall won't be eligible that way for a long time. So we can nominate individual volumes of Girl Genius or Schlock Mercenary for Hugos, because they're released as large lumps of story at a specific time.

Obviously we should advocate for Freefall graphic novels.

198:

Isn't the "starting position" in e g Google maps, related to where they think you are (I assume based on your IP)? At work (fixed IP) the first map screen on my computer is centered dead on my actual location but when at home the starting map can swing around quite a bit (sometimes as much as 300-400 km off the mark)

(Maps on mobile phones are of course never off by more than a couple of metres)

199:

The web versions of Google (and Apple) maps allow me to set the starting point. And since I've let Google know where I think "home" is it usually knows I'm "here". My plane travels until a year ago, VPN use at times, and general moving around with a laptop would at times make it harder for the apps to figure out where "here" was at the time. And me using 2 to 4 different non phone computers could also add to the confusion.

200:

David L @ 193: "As to is your personal PC leaking information even though you are not using it does it have WiFI or Bluetooth hardware?"

No, my personal PC has no WiFI or Bluetooth hardware. I wanted it that way. The Bell tech installer who came to put in a fiber optical cable (and a Bell 3000 Hub router in the kitchen closet) expected me to buy a Wifi device for my personal PC to eventually connect it to the router in the kitchen closet. I immediately got the Bell marketing department on him and he went back to his truck to get enough Ethernet cable to link my personal PC to the router in the kitchen closet.

I don't care if my employer's iPhone XR talks (through Wifi or Bluetooth or whatnot) to my employer's Tablet PC sitting a few feet from it. I don't care if they transmit geo data all around the neighborhood. That's their problem, not mine. Anyway, they have a real big IT department, so they can figure things out by themselves, security-wise.

Don't worry, I won't be getting a subscription to nordVPN's service or any other geolocation blocker. I was just impressed by their way of saying hello. I'm more for taking stuff out or shutting things off than for putting stuff in.

So, I'll go on reading about digital trepanation for out robot friends.

https://en.wikipedia.org/wiki/Trepanning#/media/File:1750_Trepanationsbesteck_anagoria.JPG

201:

I feel your pain. My folks spent a while sailing round the Med. Because of the problem of them arriving at any particular place at any particular time, normally I met them in winter when they'd settled in some reasonably accessible location, but there was the one time I went out to meet them in Greece in the summer. Thanks to the times of flights, I had to be up at stupid dark o'clock to get to Stansted, then hang around in Athens for 14 hours waiting for the final leg to the nearest airport which got in after dark. I then had a 3 hour drive through two sets of mountains in the dark, replete with the kind of hairpin bends you describe, whilst chronically sleep-deprived. Greek mountain roads are bum-clenchingly bad at the best of times when you're paying full attention; add the risk of a micro-sleep moment, and it's highly not recommended.

202:

IP addresses & location
Curious, that ... every "body" seems to "think" I live in Ilford or Redbridge & occasionally, Romford.
Whereas I actually live between the Roding & the Lea, bounded by the marshes & the wildfowl on the W, & the forest-remnants with actual cattle & occasional deer & thousands of squirrels on the E.
The various woodpeckers commute between the two.

203:

Scott Sanford @ 197: Obviously we should advocate for Freefall graphic novels.

I haven't been able to find a license for Freefall, but Mark Stanley seems open to derivative works that are at least in the spirit of Creative Commons. So there is no reason why one couldn't create an e-book version.

Chapter One is now finished, so an e-book of all strips from 0001 to 2900 would seem to mee the Hugo criterion. It's been something I've thought about doing myself. Maybe I'll bump it up the priority list.

204:

An Apple TV became visible to me named:

For a not-unrelated reason: I have a handful of Apple Home compatible devices (the Apple TV -- which I use in preference to installing an ethernet connection in the TV itself, because smart TVs are basically ad-delivery tubes these days, whereas at least Apple understand they're selling hardware plus ad-free subscriptions: privacy is part of their value proposition these days -- and also Homepods in a couple of rooms). The names of the rooms bear no relationship to the layout of my flat, or even to reality (want to guess where "the dungeon" is?) simply because in the event that there's a future privacy leak, I don't want to provide would-be home intruders with any clues about what I've got, and where.

205:

(want to guess where "the dungeon" is?)

Your office?

206:

Re: 'IP addresses & location'

My non-techie brain explanation to self (actually a visualization) about how this stuff works settled on this 'answer':

Of course, it (Google or whatever) can figure out my location - otherwise how can it connect my device to whichever url I want to access! I sorta picture the ISP way-finding ghost-in-the-machine as chirping out homing signals at random until it gets a responding chirp which it then leaps onto until the next end of the road, repeating until it arrives at the destination. And because the whole point of the exercise is to enter a two-way communication between my device and the target url, the ghost keeps signaling back and forth between my device and the destination url. As importantly: If the ghost couldn't consistently figure out the best route, i.e., most direct with the least time lag, strongest signal for visuals, etc., then that ISP's customers would complain/switch to a 'better' ISP provider. Applies to both laptop and smartphone.

Google maps - my guess is that some of the weirder (mis)directions are attributable to a combination of the coverage density of the local internet/cable network plus street configuration. I've had more issues in older cities with twisty non-rectangular grid patterned streets (Boston) than in some of the more remote and less heavily cabled areas of New Hampshire. But maybe it's population density: number of users competing for finite Internet/cable access? Which brings me to another question: At what level of usage does such a system start getting bogged down because there's too much noise? It seems that 'Internet' capacity is portrayed as a pipe that allows X amount of flowing data. But how much of that data is machines/people logging on and off, or the machine trying to maintain a connection with the ISP even though it's not actively being used at that moment. I'm guessing that 'maintain the connection to the ISP' stuff is a key/high priority signal probably by conscious programmer/system design therefore we may soon be getting messages to conserve our communications bandwidth by switching our devices off. (Hmmm - could be useful for changing the expectation of being able to respond to office emails 24/7.)

Looking forward to finding out what the correct imagery is. :)

207:

105 :
https://www.youtube.com/watch?v=9hBpF_Zj4OA

204: my wifi network is called ' MI6 surveillance van'
i like to imagine it causes panic

208:

Apparently, someone had decided that a medieval goat track was a good choice for a 20th century paved road. For many of the switchbacks, I had to back and fill more than once to get around the corner. Given that there was no room to turn around, and that backing uphill into oncoming traffic for an hour would have been suicidal, I opted for the lesser of two evils and continued my descent. Oh, did I mention that we were driving directly into the setting sun for the last half of the descent? Thankfully, the locals were smarter than me, and none of them attempted to climb the mountain while I was descending.

I learned to drive on a road you may have heard of: Mulholland. I like curves.

Anyway, if that was Greece, it probably wasn't a goat track. IIRC, the early 20th Century road design practice was to load 50 kg on a donkey, and drive it uphill, with the engineers staking its path behind. The donkey, being not stupid, chose a winding route up the hill. Since the engineers were designing for first and second generation cars, not the monsters we have now, they wanted to keep the grade low, and that's precisely what the fully loaded donkey did. It may have even cut down on the costs of building bridges and tunnels for long straightaways.

You may laugh and/or cringe, but if we continue on our current merry trend of not repairing infrastructure while we flirt with crashing civilization for fun and profit, those donkey tracks probably will still be functional* a century from now, while any mountain road that depends on bridges, tunnels, and regular repairs will be unusable.

*for donkeys, that is.

209:

The donkey, being not stupid, chose a winding route up the hill.

Apocryphally, when they were building the roads in the West Riding (now West Yorkshire) in the 18th century, there was a locally-famous blind guy who was a regular at all the local village pubs: he knew the land like the back of his hand, so the surveyors just followed him up hill and down dale to lay out the roads.

(This is probably a local urban legend but if you know the roads in that part of the world the idea that they were routed by a drunken blind guy seems entirely plausible.)

210:

SFReader @ 206: Of course, it (Google or whatever) can figure out my location - otherwise how can it connect my device to whichever url I want to access!

Answer: not like that.

Under the hood all data on the Internet flows as "packets" of typically a few hundred bytes each. When your computer "connects" to a web site it actually just sends a packet to the server's IP address saying "Hello? connection wanted". The web server replies with a packet saying "Ok, connection established. Send me packets with this reference number". Then your computer splits up the web page request into a bunch of packets and sends them on their way. The server receives these packets, assembles them into the original message, requests a resend of anything missing, acknowledges receipt, and then sets about assembling the data to send back in return, which also just goes as a series of packets. It looks like a "connection" in the telephone sense, but it isn't really, its just packets (BTW the name of the protocol that handles this pretend "connection" is TCP, the Transmission Control Protocol).

Google isn't anything special in this whole scheme; its just another web site. What it does is to give your computer the name (e.g. "www.antipope.org") of the web server you want. Your computer then uses a system called the Domain Name Service (DNS) to look up the IP address of "www.antipope.org", which happens to be 89.16.179.174. So that is the destination address of the packets sent by your computer. Google doesn't run DNS; its part of the whole of the Internet and by default you will use a DNS server run by your ISP (although Google does provide public DNS servers you can use if you prefer). My earlier tutorial on DHCP mentioned that "other configuration data" gets sent when your IP address is allocated. The IP address of the DNS server your computer should use is one of those other bits of data.

Moving these packets around is done by special purpose computers called "routers". These are the big daddies of the little router box in your house. Each one is connected to several other routers, some of which may be owned by other companies. Inside each router is a table with entries that look something like this:

100.123.*.* - send via Connection 5.

This tells the router that any packet destined for an IP address starting 100.123 should be sent to whatever is plugged in to socket number 5 in the back of the box (I'm simplifying). A complicated set of protocols lets the routers talk to each other to figure out who is connected to whom and hence the best line for any router to forward any individual packet. Once the routers have these tables worked out everything flows smoothly. If a router breaks or someone digs up a cable then the routers figure out where (in IP address terms) the damage is and recompute their tables to route around it.

The routers know nothing about geographical location, they just know each other's IP addresses. Their collective "map" of the Internet is a bit like the London tube map: its got all the connections but the relationship with geography is approximate at best.

Some of the geography can be deduced by figuring out who owns the IP addresses. For instance if you look up my IP address you would find it is owned by my ISP, and a quick bit of Googling will tell you that my ISP is a British company. So obviously I live in the UK. Specialist companies have collected more data by whatever means they can, and they sell this to interested parties. Here is one example. It tells me that my IP is located in the middle of the Thames just outside the Houses of Parliament. Others seem to put me in a town a few miles away from where I actually live (I'm being deliberately vague just on general principles here). I'd guess that this particular geolocation company only knows what ISP I use and has a common location for all IP addresses owned by it. Others may be able to match physical addresses typed into web sites with IP addresses, and hence figure out that e.g. 100.123.45.* are all based in Barnsley becuase the ISP that owns 100.123 has allocated 100.123.45 to it's Barnsley exchange (simplifying again).

Google can do all this too of course, but it also has a couple of other tricks. When your Android device connects to a WiFi network it will (unless you change the settings) tell Google the IP address, the name of the WiFi network, and its geographical location from the phone's GPS receiver. Also the Google camera vans for StreetView used to collect WiFi data too (though Google promises that they have stopped doing this). So Google can tell exactly which coffee shop you are sitting in when you search for something, and also what your home address is to within a couple of houses.

This vague relationship between IP address and location has caused problems. If you have a tracking app on your phone in case it gets stolen, it will try to figure out where it is. If it can't get GPS or doesn't have it, what you get is the best guess based on the IP address of the WiFi point it just connected to. Just like my IP address looks like it's in the Thames, there are companies which know that a block of IP addresses is e.g. in Barnsley, and report the geographical centre of Barnsley as the location for any IP address in that block. Which is a problem if your house happens to be in the geographical centre of Barnsley, as you get a lot of angry people banging on your door demanding you return their phone. In some cases the police have been known to serve search warrants on these unfortunates too.

211:

SFReader @ 206: Google maps - my guess is that some of the weirder (mis)directions are attributable to a combination of the coverage density of the local internet/cable network plus street configuration.

Google Maps has nothing at all to do with Internet connections. Google simply has a database of map data obtained from wherever (some mix of data bought off other companies and derived from their own mapping vehicles).

They *do* combine this with data derived from Internet connections, as I explained in my previous post, but the two are still separate.

It seems that 'Internet' capacity is portrayed as a pipe that allows X amount of flowing data. But how much of that data is machines/people logging on and off, or the machine trying to maintain a connection with the ISP even though it's not actively being used at that moment.

Like I said, the Internet just moves packets. Here is a picture of how routers connect:

http://dtucker.cs.edinboro.edu/routing_1.gif

Each connection between two routers has a physical capacity measured in bits per second. ISPs have the difficult job of making sure that they buy just enough capacity to accomodate all the people who want to use the Internet. Obviously most people aren't using their Internet links at full capacity all the time, so an ISP can sell, lets say, 50 connections in Barnsley each of which does 10 megabits per second, but then provide a lot less than 500 megabits per second to the whole of Barnsley. As long as everyone doesn't decide to download the pirate edition of Game of Thrones all at the same time this isn't a problem. Figuring out what the people of Barnsley *are* likely to do on the Internet next year, and how much bandwidth they will need to do it is a complicated problem that provides much employment for a wierd hybrid of network engineer, statistician, sociologist and fortune teller. You know when they get it wrong because your Internet connection slows down and Netflix complains.

ISPs connect to each other as well, generally at big high security buildings called "co-location facilities". Small ISPs pay bigger ISPs for long-distance connections in much the same way as you pay your ISP for connections outside your house. ISPs of the same size also enter "peering agreements" by which they connect to each other. Its all big, complicated and rather opaque. ISPs also pay other companies for capacity on actual long-distance cables, or sometimes they own the cables themselves. A few years ago companies that owned long-distance rights of way (like railways and canals) realised they were sitting on gold mines: put a fibre-optic cable down the bottom of a canal or alongside a railway track, stick a router at each end, and rent out the bandwidth for easy money.

If you are not using the Internet then your machine isn't sending or receiving any packets; there is no notion of "maintaining a connection" in that sense. Like I said previously, its all just packets. Your home router has a physical connection to your ISP, but that can probably handle a lot more bandwidth than you are actually allowed to use because the expensive bit is the back-haul bandwidth to the rest of the Internet.

212:

What SFReader describes is close to how the internet routing works, not how Google works.
(As Paul explains in his post)
Of course, you can always use tracert to find out what the route is between you and any other IP address
the routers have figured out.

Note that in a pinch phones can be located by which cell towers they are pinging off of, and
the flight time of the pings. Yes, this is real (not just something on TV). A co-worker went
to work for a company developing that solution, oh, it must be 15+ years ago now.

Somebody also mentioned about IP address allocation. Note that the IP address that your computer
is using, the IP address on your cable modem, and the one Google sees may all be different.
IPv4 addresses are hard to get these days, and big ISPs use carrier-grade NATing. In effect, they
NAT your address again (or PAT, so the same IP address is used for multiple customers) so the
IP addresses they have to pay for is minimized.

Of course, IPv6 could fix this, but network admins are finding out that all this NATing (or PATing)
gives an extra layer of security to their customers for "free", so IPv6 NATing will probably become
a (bigger) thing.

213:

I wrote: Your home router has a physical connection to your ISP, but that can probably handle a lot more bandwidth than you are actually allowed to use because the expensive bit is the back-haul bandwidth to the rest of the Internet.

Actually, now I think about it, that isn't true any more. It used to be, but these days the back-haul bandwidth is so big that a lot of the time the wire coming in to your house *is* the bottleneck. Unfortunately fixing this means digging up a lot of roads. It is happening, but slowly and in large towns first.

214:

There's an online story by one of the fans that was good enough to read twice, (and would be worth a third reading,) so he's certainly open to derivative works, but you might check with him first.

215:

It used to be, but these days the back-haul bandwidth is so big that a lot of the time the wire coming in to your house *is* the bottleneck. Unfortunately fixing this means digging up a lot of roads. It is happening, but slowly and in large towns first.

Maybe in urban areas. Sort of. If it's non coax copper into a building then, well, it's going to cost. I've been involved in putting fiber from a curb cut into a warehouse and the number of departments of the carrier and city is a bit insane.

In the "burbs" it's a bit difference. Just talked to someone last night who got Gig Internet from the local cable company. It was fiber to his house then used his in wall RG6 to get to the modem.

Biggest issue with fiber is the pole climbers (at this time) just can't splice into the coax or grab a free pair from the phone line bundle and tap into it. They have to string fiber back to a connection point. (AT&T has one hanging above my driveway if I want to ever hook into them.)

Anyway, around here we tend to have 0 to 3 fiber runs along the streets. AT&T, Spectrum(TWC), and Google. Google is a bit thin in their coverage but I have all three. Where I see the biggest issue are the neighborhood nodes (routers) I suspect it will be a while before they can handle very much real gig traffic back hauled on 10 gig or faster fiber.

We also have at least one non major ISP who has run fiber around and through many of the business neighborhoods. Great company. We buy various services from them. It's nice when you personally know the owner.

216:

Sorry, this is my usual mistake of forgetting that I'm talking to an international audience. I was thinking of the situation here in the UK. Other nations are available.

217:

That is possibly quite close to the truth! If I read Rackham correctly, those roads would originally have been paths through natural woodland, which is hard to see far in, and which often has fallen trees, necessitating diversions.

218:

Elderly Cynic @ 217: "...paths through natural woodland, which is hard to see far in, and which often has fallen trees, necessitating diversions."

That sounds a lot like my back yard, except that I have deer, bears and muskrats in it instead of drunken blind men.

219:

204/205
The Bog?

Also - Charlie ...
The roadmaker was a real person, called "Blind Jack" Info page here
And
There is an amzing boozer in Knaresborough main square named after him - yes, I have drunk there, several times. ( Apologies for the Arsebook entry, btw. )

220:

Charlie Stross @ 209:

The donkey, being not stupid, chose a winding route up the hill.

Apocryphally, when they were building the roads in the West Riding (now West Yorkshire) in the 18th century, there was a locally-famous blind guy who was a regular at all the local village pubs: he knew the land like the back of his hand, so the surveyors just followed him up hill and down dale to lay out the roads.

(This is probably a local urban legend but if you know the roads in that part of the world the idea that they were routed by a drunken blind guy seems entirely plausible.)

I grew up in Durham, NC and now live in the State Capitol, Raleigh. They're about 20 miles apart.

The road layouts follow different philosophies.

Durham began as a railroad station for shipping tobacco to the north after the American Civil War and has a central district laid out along the railroad with a semi-grid on the north side of the tracks. During segregation & Jim Crow there was a thriving African American business district along the south side of the railroad tracks.

Outside of the central business district the roads appear to follow the local Native American trading paths. There was a small town, East Durham, about a mile east from Durham itself along the railroad tracks. By the time I was born Durham had grown sufficiently to absorb East Durham, but the area retained its name (and about a two block long business district).

I grew up in the eastern part of East Durham.

Raleigh is completely different. In 1792 the State Legislature chose a site in Wake County and authorized the purchase of 1,000 acres of land from a local planter to be the NEW Capital City, named Raleigh in honor of Sir Walter Raleigh who had sponsored the Lost Colony settlement on Roanoke Island.

Raleigh is one of the few cities in the U.S. planned from the beginning to be a state capital. The original plan is a grid with two main axes meeting at a central square and an additional square in each corner, was based on Thomas Holme's 1682 plan for Philadelphia.

Outside of the original city, roads extended to the important towns & cities of North Carolina (& beyond). Many of those towns are reflected in the names of city streets, Hillsborough, Halifax, Fayetteville, New Bern, Edenton ... The city grew outward along these roads.

Note there is no Durham St, because Durham didn't exist back then.

Some time in the last half of the 20th Century someone got the bright idea you could take two shorter local streets that were almost parallel and connect them with an S-curve to make a longer street, which is why in some places you go around a curve and the street name changes. They also decided we needed a major 4 lane highway going into the heart of downtown.

A lot of the screwy stuff in Raleigh is the result of the State Legislature tampering with the local area to make things more convenient for themselves and damn what the residents might want!

I live just about a mile north of the State Capitol, just outside the original boundary of Raleigh.

221:

I got gigabit fibre installed on 5th January (although I'm only paying for the 200Mbit/s tier and actually getting about 217, other options are 100, 500 and 900 at different prices but over the same connection) - installation involved: drilling through the wall and installing one box for the fibre to terminate and another box (router) connected to that by ethernet that has the VOIP phone connection and WiFi (that I'm not using because I plugged in a Velop mesh with two nodes instead): blowing fibre to the nearest cabinet. I got the first three months free and a static IP (that I need for some work purposes) free as well. Internet snoop pages report my location as either Milton Keynes or London (I'm actually in Aberdeen). Netflix UHD is apparently still capped at 15.25Mbps in the UK.

222:

Paul @ 213:

I wrote: Your home router has a physical connection to your ISP, but that can probably handle a lot more bandwidth than you are actually allowed to use because the expensive bit is the back-haul bandwidth to the rest of the Internet.

Actually, now I think about it, that isn't true any more. It used to be, but these days the back-haul bandwidth is so big that a lot of the time the wire coming in to your house *is* the bottleneck. Unfortunately fixing this means digging up a lot of roads. It is happening, but slowly and in large towns first.

Google had crews in my neighborhood last week to install the infrastructure for "Google Fiber" - orange plastic tubes that I guess the actual fiber will go through. Plus a tiny tube that seems to have wire already installed. I don't know if this is going to be for power or ground ... or something else, but it's obviously NOT the fiber itself.

We've had AT&T fiber available for a while and lately Spectrum appears to be "upgrading" their system (because for a couple of weeks my internet would die EVERY NIGHT between midnight and 3 am - I have real bad insomnia) while they worked on the upgrade.

223:

I haven't been able to find a license for Freefall, but Mark Stanley seems open to derivative works that are at least in the spirit of Creative Commons. So there is no reason why one couldn't create an e-book version.

I'm pretty sure some random fan making hardcopies wouldn't count. But if Mark Stanley were to release a Freefall Chapter One compilation I think that would count as publication for Hugo purposes. (I hope, anyway.) I don't know him personally so I have no idea if he's interested or would care about a Hugo Award.

That doesn't change the fact that I think Freefall deserves a Hugo Award.

224:

When they were building the Alaska Highway during WWII the foreman asked the surveyor where they wanted the road. The surveyor said "head for that brown hillock, and call me when you get there".

Several weeks later they caught up with the moose and called the surveyor…

225:

a lot of the time the wire coming in to your house *is* the bottleneck

Back in the 80s when I worked for the tricorporate, we had a prototype project to lay fibre into every house to 'future-proof' the network.

It was cancelled because management couldn't understand how anyone could use that much bandwidth and didn't want to waste money on it…

226:

I've liked the story from the Whole Earth Catalog, lo, these many years ago: in the mid-sixties, Mexico City built a new university, and paved *no* paths between the buildings. The university opened... and the next summer, they paved the paths between the buildings that people who lived and worked there made, rather than some asshole architect's idea of what might win them an award.

228:

How neurons work began to be hashed out in the 1930s-1950s. Early artificial neural networks then showed up in the late 50s if I remember correctly

Turing anticipated them with his "Disorganised Machines" paper in 1948, but he could give no training algorithm to train the network from its random initial connections. It's typical of Turing's more philosophical writing: a brilliant paper that then wanders off on a tangent (he imagines a school master teaching the neural net).

Turns out that adding one or more hidden layers solved the problem, though. Oops.
This became apparent around 1990 and caused quite a stir

I think the big turning point was 1986, Rumelhart et al. They demonstrated that a 3-layer network could be taught by simple back-propagation to map common English written words to a set of phonemes ("the neural network learns to read!"), and that if you did it would develop some of the typical errors that kids make when learning to read.

It was a huge sensation, and suddenly "Connectionism" was the hot new thing!

There was a lot of showmanship involved, as there often is when science paradigms change. For demos at conferences they used a voder that turned phonemes-to-audio with the voice of a cutesy six-year-old. So they claim their neural net "learned to read" making typical mistakes a 6-year old would make and then this voice of a little kid read out the words making the mistakes little kids make. You can imagine the effect that had back when computers didn't talk.

229:

when science paradigms change.
The one I & also, I suspect a lot of other people will remember, is the total & complete revolution in Geology approx 1964-69, with the acceptance that what we now call Plate Tectonics was the underlying mechanism behind/underneath (!) all geology.

230:

Re:'... the back-haul bandwidth'

Thanks for your explanations - much appreciated!

I just read the Wikipedia entry for 'back-haul bandwidth' and marvel at how signals are kept as clean as they are given the volume of signals coursing back and forth, between and across various pathways and materials (electric wire, optical fiber, transmission towers, meshes, satellites, etc.).

231:

I recall being shocked that the lecturer would state flatly that it was impossible for it to work.

Generally I do not hold Clarke Law ( https://www.forbes.com/quotes/745/ ) in very high regard. But in this case Arthur C. Clarke was absolutely right.

232:

ilya187
Completely NULL result - link completely borked

232
Uh?

233:

Strange. The link worked fine for me.

234:

we had a prototype project to lay fibre into every house to 'future-proof' the network. It was cancelled because management couldn't understand how anyone could use that much bandwidth

Australia's NBN went through that, but on a national scale. The guy that said "the laws of Australia override the laws of mathematics" also decided that a slower network delivered later for more money was better. Mostly it meant his mates in the telecom industry didn't have to write off their existing infrastructure but could sell the useless bits to the government and get paid by the government to fix/upgrade/extend the rest of their shitty copper systems.

235:

(I'm simplifying)

That's okay, I'm sure no-one here will 'arp on about it.

236:

(You'd really be set up to rip into them)

237:

Odd. I clicked on it, and got to the intended page.

In any case, all it has is a quote of Clarke's Law:

"When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is probably wrong."

238:

"Completely NULL result - link completely borked"

That's a good sign. It indicates that you have some scummy vermin called "trustarc.com" and possibly also "consensu.org" already in your block list. That Forbes site tries to force you to pass through the Gate of Evil so they can eat your flesh, but if you have the right protective wards installed they can't even show it to you.

239:

I've liked the story from the Whole Earth Catalog, lo, these many years ago: in the mid-sixties, Mexico City built a new university, and paved *no* paths between the buildings. The university opened... and the next summer, they paved the paths between the buildings that people who lived and worked there made, rather than some asshole architect's idea of what might win them an award.

This story was also told of my alma mater, The Evergreen State College - so circa 1970, about fifteen years before I heard it as one of many campus legends. It's possible this idea was going around at the time, as the paths between the four original dorms and the main campus look like that. Hopefully Mexico City had less rain and mud than Olympia, Washington.

240:

These are known as desire paths. Sometimes site managers will try to block them, leading to secondary paths that route around the barrier. Sometimes they accede to the inevitable and pave them over when they emerge. And there are lots of stories of planners who use this as a deliberate strategy, including at a number of colleges. It's likely that these overlap. See https://99percentinvisible.org/article/least-resistance-desire-paths-can-lead-better-design/ for some stories.

My anecdotal contribution: our staff car park developed a desire path leading off site. I suggested that it be paved, but I was told that any alterations would require endless negotiations with the freeholder so it wasn't worth considering.

242:

Forbes?!?

That's one of those websites that whines and refuses to show you anything if you have an adblocker installed ... then if you disable the blocker, it turns out they use ad exchanges that host malware and install javascript cryptocurrency miners in your browser.

243:

Which, in theory, is a breach of the Computer Misuse Act. But there's one law for peasants and another for persons who matter, er, megacorporations.

244:

In its standard config Firefox blocks the installation of javascript cryptocurrency miners and four other types of malware.

It has two higher levels of blocking above that.

245:

I also think that even if you would start trying to accuse somebody using your browser to mine Ethereum, there'd be a lot of finger-pointing, avoidance, and obfuscation.

Ad systems apparently (not my specialty, please somebody correct me if I'm wrong here) have multiple layers, so that the web page owner pays some ad exchange, who then displays ads from some other exchanges, until you get to the "real" ad. Of course hopping countries and doing other obfuscation might be used, too.

So, it's not easy nor quick.

I have also read some texts which indicate that much, if not most, of the internet ad business is basically a scam, and don't really make money for the web sited displaying the ads, nor really direct business to the advertisers.

246:

I heard the same thing about Waterloo, from a prof who was there when the university opened. He said they used a helicopter to photograph the paths in the snow around the dorms and used that to decide where to pave and where to plant sod when spring came.

It sounded like a neat idea, but given that people tend to follow the easiest trail, and in snow that means one already tramped, I always suspected that if true it meant the paths were following routes trampled in the snow by a minority of students.

In The Adolescence of P-1, set partly in Waterloo, the author mentioned university admin caving to student desires and paving the desire paths they have worn into the grass while tearing up and replanting the old (unused) paths — at which point student start religiously ignoring the new paved paths and following the old previously-ignored routes. It feels like a detail based on something that actually happened…

247:
I think the big turning point was 1986, Rumelhart et al. They demonstrated that a 3-layer network could be taught by simple back-propagation to map common English written words to a set of phonemes ("the neural network learns to read!"), and that if you did it would develop some of the typical errors that kids make when learning to read.

Even earlier than that. Rumelhart et al's classic three volume "Parallel Distributed Processing" was published in 87, and most of the work it referenced was early 80s IIRC (mine are currently in storage so I can't check — so my memory could be lying:-)

It was certainly presented as run-of-the-mill uncontroversial stuff when I did my AI undergrad degree in 88.

248:

Exactly. And the first thing Forbes does is process your request through these "trustarc.com" scumbags, who purport to provide some kind of assurance that a site is not dodgy, but according to wikipedia sites that use them are 50% more likely to be dodgy than sites which don't. If you have "trustarc.com" itself blocked, then the processing can't happen and all you see is a blank screen.

The reference wikipedia cites ( http://www.benedelman.org/news-092506/ ) does not say 50%. It actually says over 100%.

Personally, I regard the mere appearance of "trust" as a substring of the domain name to be an indicator of dodginess, on the general principle that the more some outfit prances and postures about how much you can trust them, the more likely it is you actually can't.

249:

"...paving the desire paths they have worn into the grass while tearing up and replanting the old (unused) paths - at which point student start religiously ignoring the new paved paths and following the old previously-ignored routes. It feels like a detail based on something that actually happened..."

It sounds like they have forgotten that absence of evidence is not evidence of absence. The old paths were not useless, they just weren't useful for all routes; the new paths, ditto. But a paved surface continues to look the same whether people walk on it or not, whereas a grass one doesn't.

A very common sight in England is for two streets, with pavements down the sides, intersecting each other at right angles, to have a square of grass inside the angle where they meet. There is invariably a muddy track worn across the diagonal of the square by people who are turning the corner. If you are very lucky, after many years of muddiness the council might even tarmac or pave the track. But you can't then tear up the pavements along the edges of the streets and replace them with grass, or else that too would be worn to mud by the people who are going straight on.

250:

I know of exactly one trust that I do: trustwave.com, who is a root CA.

ObDisclosure: I worked there for four months in '09.

251:

Of course the obvious solution is to pave the entire thing, then declare it tapu and make everyone walk around it. Which is what the military do with parade grounds.

252:

The bigger problem around here is mountain bikes, although OHVs do the same thing in their parks.

It's called gator-backing, where people keep making new trails to have the fun of riding a new trail. left unchecked, it ends up looking like an alligator's back from the air. This is how you get 30 kilometers of trails in a few hundred hectares.

Since the mountain bikers are inevitably parasitizing park spaces that were set aside for the protection of rare species, and destroying them for their own amusement, my sympathies with their path-building urges are at best limited. The OHV brigade is only better when they're destroying areas set aside for them to destroy.

253:

Heteromeles @ 252

The big problem we have around here has to do with our ongoing replacement of logging clearcuts with logging mosaics. In theory they give a large variety of squarish "islands" in which the animals can hide in. In practice they multiply the number of logging roads where giant logging trucks tamp down the soil forever, and they make easy for predators to spot their prey along those huge avenues between the square islands.

On the positive side owners of ATVs or snowmobiles are afraid to go on those logging roads that make up the dividers between the squares of the mosaic. They don't want to get flattened by a giant logging truck. So, they keep to the narrower marked-out paths established long ago for snowmobiles and used in the summer by ATVs.

254:

I saw a really interesting sign the other day in one of those "wreck this bit" parks, to the effect that if people kept setting the dumped rubbish on fire the park would be fenced off. It looked very much as though there was at least a two way contest between the illegal dumping (often by the truckload), the MTB track builders, and likely the local vandals/bored youf. It's been ~30 years since I was involved in any of that so I'm a bit out of the loop.

We have minor issues with people riding bicycles off track on the riverine parks, but the council put up tape where they really don't want people riding and that seems to work. In a couple of places the tape is backed up with wire rope and that definitely works 😏 (I assume the council does the rope too). A bigger problem is motorbikes, occasionally some fuknukle on a dirtbike will go digging up anything and everything, or someone in what I assume is a stolen car does likewise. I like to think they occasionally discover which hazard tape is backed with wire rope by riding into it.

255:

but the council put up tape where they really don't want people riding and that seems to work.

Must be nice.

Around here the various riders were encroaching on an airport property. Which is also a big part of a watershed with rules on run off. Signs didn't work. So the airport put up a fence. Fence keeps getting torn down. And those not tearing it down (well mostly) are in court demanding the airport shrink their security perimeter and let them make/ride tails on the airport property in the watershed protected areas.

256:

See https://99percentinvisible.org/article/least-resistance-desire-paths-can-lead-better-design/ for some stories.

99 Percent Invisible is a huge time sink! Why don't you just send us to TV Tropes while you're at it?

Yeah, I blew hours reading their short educational articles. Again. *grin*

257:

Most places I've seen around airports the surrounding area is set aside for aircraft to use if things don't go according to plan. Then the council discovers that lots of business like to be near the airport for various reasons. Then they discover that lots of other businesses like to be near the airport because of all the potential customers who are going to the businesses near the airport and before you know it all that emergency space is full of buildings and people and the developers are all very rich.

Just having dirt bikes trashing the place sounds like bliss.

https://en.m.wikipedia.org/wiki/2017_Essendon_Airport_Beechcraft_King_Air_crash

258:

You left out: people like to build houses near the airport because the land is cheap and it's close to transport, but then complain about the noise and try to get the airport shut down

259:

Around here it's not people, it's developers, which may be a subtype of people, but I'm not sure. The more interesting of them like to build homes in hazardous areas (high wildfire, airport flight path, next to old landfills, etc), market them, then leave the residents to find out what they've been lied to about. That's when the complaints about airport noise start. All the people who were smart enough to notice didn't buy, and the neighborhood gets weighed with the less observant people.

That, incidentally, is why California property purchase agreements are so annoyingly long. It's not big-brotherism, it's that so many real estate scams have been run here over the years that the legislature decided that caveat emptor was insufficient. So now they require lots of warnings.

Yes, we have a shortage of housing. Partially that's because we're picky, but I cringe at what that means other, cheaper states are willing to allow in terms of hazards. I mean yes, during your California lifespan you're likely to get hit by an earthquake, a fire if out in the boonies, and a meter of rain when the norm is 30 cm, but that's normal for every place. Right?

260:

In The Adolescence of P-1, set partly in Waterloo...

I still maintain this is an under-appreciated piece of science fiction. One of these days I should actually watch the TV movie but I'm more likely to just read the original novel again.

261:

people like to build houses near the airport ... then complain about the noise and try to get the airport shut down

This was literally a lawsuit by Donald Trump, and he's widely regarded as a person.

262:

Around here it's not people, it's developers, which may be a subtype of people, but I'm not sure. The more interesting of them like to build homes in hazardous areas...

My college geology teacher had a story about that, the punchline of which was him delightfully bursting out, "I know that place! That's where the mudslides come down off of Mount Rainier!" And then he looked up from the map to see the developers' faces and discovered they did not want to hear this. They were quite grumpy about not being able to build in the geologically interesting mudslide zone...

264:

Ah, yeah, but that's not really the same thing as paths made by people who just want to get between two points by the most direct route instead of following some pillock of a planner's desire to send everyone the long way round.

Round here we have a piece of woodland, designated as a nature reserve of some minor degree, which has an ancient holloway running up the hill at one end of it. This now looks like a huge ditch some metres deep with steeply sloping sides. There are all sorts of tracks and paths crossing it where kids have found new opportunities to ride bicycles down one side and up the other, but they only appear where they can do so without having to ride through a tree or through too dense a stand of undergrowth. To my inexpert eye they seem more of a risk to the antiquity than to the plant life. They don't come round with axes and shovels to create their own opportunities, which seems to be what you have to deal with.

265:

We don't really get those particular hazards all that much, but their favourite trick is to build on one of those inviting expanses of completely flat land that for some strange reason nobody else has built anything on for hundreds of years except possibly in one or two places where there happens to be a bit of a hump a few metres above the rest. Then people complain about getting wet, and it's never the builders who have to try and get them dry again, but the councils who gave them permission to build there in the first place. (For some reason councils seem to be where all the most mindbendingly thick people end up.)

Locally we have an example of the post-industrial variant on that, converting old warehouses into yuppie flats and selling them to people who are too unobservant to read the 500 years of flood level records carved into the wall, or to see that the old warehouses stand on the edge of the docks.

Though certainly anything that is a well-established noisy feature has to put up with constant complaints from people whose house viewing appointments were carefully arranged to fall at a moment when it wasn't making a noise. Railway yards, race tracks, concert venues, anything. And for some bizarre reason, when the complaints start to get legal, it doesn't seem to work to simply say "tough shit, we've been here for 100 years, you should have listened before you moved in you stupid bastards". Instead they have to spend loads of money on the impossible task of trying to make a quiet noise, and since a lot of them are not doing much more than scraping by to begin with, we end up with valuable amenities being closed or crippled by a handful of moaning divvies.

266:

Like Abbottsford, or as it is now known, Lower Abbottsford...

https://en.wikipedia.org/wiki/1979_Abbotsford_landslip

People often seem not to want to know why things are cheap, or why there's an open expanse of flat land, or other questions that seem important to me. OTOH, I am currently eyeing off some cheap land that people are selling because the bushfires saved them having to apply for permits to clear the trees that were on it. Or houses, same same. I am hoping that in an area where a 500-700m2 section normally goes for $300k I will pay less than $50k. The cost of building a house has gone up by about that much, and if I can get my cheat* approved it will be a big win.

* "cheat" in the sense of a fire bunker built for much less than the normal cost, not "cheat" as in piling up tinder and hoping for the best.

267:

I never before thought it was plausible that Captain Kirk managed to talk so many computers into self-destructing, but huh, this post is making me reconsider.

268:

People often seem not to want to know why things are cheap,

Oh my goodness yes. (In my most southern belle voice.)

Moved to Raleigh, NC 32 years ago. County politicians and staff had figured out landfill would run out of space about 10+ years prior and bought up some land that would be suitable for the next landfill. Opposite side of county and didn't cost all that much. Population back then about 225K and projected to grow much faster than most areas. (We're now at 1.1mil.)

So the land in the 10 miles or so around the "next" landfill was now not appreciating as fast as the rest of the county. So developers starting building housing in that area. Mostly bought by people moving in from other areas who had missed all the local reporting on the new landfill.

As county started moving forward with plans to open the new landfill, well, the howling was heard for miles around. Lawsuits. Protests. Political campaigns. Letters to the editor. You name it. All of it doomed to failure as the county had been very public with their plans but still it slowed things down enough that there was going to be a gap between the old one filling and new one opening. Study was done. The county discovered that recycling was going to be cheaper than shipping garbage by rail 1000 miles or more. So we got serious about recycling. (Back then newspapers were something like 20%-30% of the volume going into the landfill.)

So now all is well. We're recycling. Old landfill will keep working till new one opens up. Great.

Oops. Developers had built some big housing developments around existing landfill and people bought in based on the landfill would be a park in a year or so. Not 5 to 8. More lawsuits. Gnashing of teeth. Letters to editor. Political campaigns. Sound familiar?

Anyway there was a huge amount of money spend on politics and lawsuits and at the end of the day it all happened as planned. Plus we got a decent recycling system that is likely much better than if things had not happened this way. Now if they could only find some new poor country who would take our plastics now that China has stopped.

Then there was the housing development built under the flight path of the American Airlines hub here. Strangely almost all appointments to see the model homes took place during lulls in the flight waves taking off and landing. Rinse, lather, repeat.

269:

You left out: people like to build houses near the airport because the land is cheap and it's close to transport, but then complain about the noise and try to get the airport shut down

A friend who's a recently retired pilot told me about Burbank airport in the LA area. His comment was that he figured he flew into it maybe 100 times. He thinks he had 3 smooth landings. And all the takeoffs were a bit interesting. Lots of noise abatement rules which dictate the flight paths with noise monitors all around which can generate fine for specific flights based on the meter readings.

Short version, don't fly into this LA basin airport if you're the least bit uncomfortable with flying.

The western approaches to DCA (Washington DC) can also be interesting. Basically planes get to fly the Potomac River if approaching from or leaving toward the west. If landing from the west and seated on the left side of the plane, don't forget to wave at the folks in the USA Today building. Seriously. Of course this is nothing like the old Hong Kong aiport.

270:

Sometimes, of course, the airport comes to you. I see the wikipedia article for Cribb Island (slightly famous for giving the world the Bee Gees) sets the tone from the beginning: it *was* a suburb of Brisbane...

271:

The more interesting of them like to build homes in hazardous areas

There's a development in Cochrane, Alberta on top of a bluff with wonderful views. While my brother-in-law was studying geology his prof used that as an example of building in foolish places, because it was only a matter of time before the bluff eroded and those houses fell into the valley. One of the students in the class was very upset, because her parents had just bought one of those houses.

Alberta has had seven of the ten most expensive natural disasters in Canadian history. And is the most resistant to planning regulations and the like, because apparently rules infringe cowboy freedoms or something.

https://globalnews.ca/news/2810070/top-10-most-costly-disasters-in-canadian-history-for-insurers/

272:

Then there was the housing development built under the flight path of the American Airlines hub here. Strangely almost all appointments to see the model homes took place during lulls in the flight waves taking off and landing.

In Arthur Hailey's 1968 novel Airport, there's a scene where he describes the problems pilots have because they have to make dangerous noise-abatement maneuvers just after take-off because of complaints from resident under the flight path. The backstory is that when the development was going in the airport posted signs that it was expanding the runways and developers tore them down so buyers wouldn't realize. I've always wondered if he was basing that on an actual event.

273:

I've always wondered if he was basing that on an actual event.

I think such happened around Chicago's O'hare (ORD) in it's early days. After the move from Midway. And such happening around LAX in Los Angeles wouldn't surprise me. Both have developments tight around them.

I don't know about the local one. It was built just before we moved here. I must be strange as when I move somewhere I actually get out a map and look for such things like where do the airport runways point.

For a while an apartment we had in the DFW area was literally under the direct line of one of the DFW runways. But they sound proofing was good and my wife could walk to work and for other reasons we rented it. One time when I was there there was a flash of light into the apartment. Then a few minutes later another. Happened a few more times over the next 10 to 15 minutes. At some point I figured out that when the sun was at a particular spot in the sky it would reflect off the bare metal skin that American had at the time and sweep a big spot on the ground as the planes came in for a landing. Angles were not there for take offs. And it would only happen a during a brief time and date window when the angles were just right. (American into DFW, in normal times, would land 150 or so planes one after the other using 4 runways at a time.)

274:

Pigeon @ 265:

Though certainly anything that is a well-established noisy feature has to put up with constant complaints from people whose house viewing appointments were carefully arranged to fall at a moment when it wasn't making a noise.

Google is your friend. When we were last house-hunting (wow, like 15 years ago, where does the time fly?) we looked up every house on Google Maps before going there. These days we would also look hard at Google Earth and StreetView. There are quite a lot of things you don't want to live near that show up on them.

A few miles away from us is a large sewage treatment plant. It happens to be just to the east of a rather nice village where we might have considered buying a house, if it not for the smell whenever the wind blew from the east. We happened to know about this plant because our son went to school nearby, but other more innocent housebuyers find out too late because the local estate agents look at the weather forecasts when scheduling viewings in that area.

275:

At some point I figured out that when the sun was at a particular spot in the sky it would reflect off the bare metal skin that American had at the time and sweep a big spot on the ground as the planes came in for a landing.

That's the way the Iridium flares worked. Alas, the current generation of Iridium satellites don't have the flat, shiny antennas that produced the flares. Other satellites still flare, but not as predictably as the Iridiums did (mostly).

https://en.wikipedia.org/wiki/Satellite_flare

276:

I think such happened around Chicago's O'hare (ORD) in it's early days. After the move from Midway. And such happening around LAX in Los Angeles wouldn't surprise me. Both have developments tight around them.

Not particularly, although I'm not sure I'd want to live in Culver City or Westchester.

LAX is near the ocean. Since there's an onshore breeze, planes normally take off towards the ocean and land coming in over the LA basin. Indeed, if you're at LAX, you can look east and see a long line of sparkly plane lights coming in. Usually around eight of them, four per runway, five miles or so apart.

There used to be a housing subdivision at the foot of what's now the runway at LAX. The homes were torn out, and it's been largely returned to nature, as one of the two homes of the endangered El Segundo Blue Butterfly. the other home is just down the coast at the Chevron refinery (Do people greenwash their polluting operations by bragging about saving little blue butterflies? People do.) I've been out on the El Segundo Dunes once, and it's effing loud, to the point of needing ear protection. You also need a security minder, since it's the perfect place for someone with munitions to bring down a plane, and no one wants that to happen.

But most of the buildings in the immediate landing path next to LAX are well-sealed business and industrial buildings. I worked in one of them. Not fun, but not as loud as you'd expect. Landing doesn't make the same level of noise as takeoff.

Burbank's another story, as is San Diego's Lindbergh Field and John Wayne in Orange County. In all of these, the planes are climbing up over residential neighborhoods There planes have to go up fast and underpowered, to keep from deafening people. That's how the pilots earn their "huge" salaries. I don't know about Burbank, but San Diego airport more-or-less shuts down takeoffs at 8 pm, and I assume it's so people can sleep.

The wildest ride I've had in the last few years was flying into Reno. It's on the eastern edge of a rather tallish mountain range (the Sierra Nevadas), and it's in a desert (which produces thermals) with some interesting solar thermal power plants in the flight path (those mirrors produce a lot of extra heat). All of this makes for a rather bouncy flight in.

Now I'm just waiting for someone to chime in "you think that's bad. Try landing at Paro." Any takers?

277:

REALLY dislike dirt bikers. Hell, back in the mid-eighties, a friend, who had moved back home for a while, outside Coopersburg, PA (around 45 min nw of Philly) would walk the bounds of his folks' property. In spite of the "NO TRESPASSING", etc, they'd drive their bloody bikes onto private property and leave trails.

278:

Park Ridge, IL. Nice 'burb of Chicago. Were screaming and yelling when O'Hare wanted another runway.

As opposed to Norridge, several mi- south of there. When my late wife and I relocated to Chicago from Austin, for about 3 weeks, we stayed in friends' condo (they were divorcing). Late afternoon, all conversation would stop every 3-4 min.

It was about 3mi due east of the tower at O'Hare.

279:

These days we would also look hard at Google Earth and StreetView.

When we were looking for a place in San Antonio a while back, Google Earth was very handy for finding out where the flight paths from KSAT were, also a couple of local quarries where blasting was going on. And StreetView for a preliminary check of the neighborhood.

280:

I'm kind of interested to see what happens when we do get ubiquitous ML, sociologically speaking.

Consider (one of) the simplest forms of ML: categorisation. Treat your ML system as a black box, feed it bunches of pictures labelled "cata" and "dogs", train it and now it can recognize cats and dogs.

Now imagine this is built into Excel, Outlook or whatever. And imagine your job is a financial admin in a small company. Your job is to approve expense claims. Now realize that this is a categorisation problem: sort your claims into two labelled folders, "good claims", "bad claims". Train your model.
It now does 98 % of the work, you're just left to deal with the corner cases.

Now apart from the fact that over time, your company is turning into a pile of AIs, realize that the staff learn whats an "acceptable" claim and adjust. Your job has turned from "sort claims" into "ML retrainer", and your role is to debug "Why did the ML system get this decision wrong?".

Your skillset and desired education switches over to learning about biases and how they affect cognition. So Racism, biases in general become "obvious cognitive errors" to you over time.

Along with ML becoming, in practice, experimental psychology, this will dramatically change the way society actually thinks over time. The next few decades look fascinating.

281:

Heteromeles @ 276

The shortest runway in the world is at Saba island.

https://www.youtube.com/watch?v=sk7D2Hvqt8I

But all this is irrelevant since Arthur C. Clarke said (in one of his last novels) that AIs will soon arrive and there will be no human pilots anymore and all landings and take-offs will be safe.

282:

Or Patrick McGoohan typing "Why" into the computer terminal during the last episode of The Prisoner. It seemed really cool when I was 12...

The Star Trek thing is easier to understand. Imagine that you're a sentient computer, and you can either overcharge your capacitors and blow yourself to oblivion, or you can discuss galactic politics with William Shatner...

More seriously, I guess the show's writers didn't understand that invalid input would generate an error message and a return to the prompt. Do you want to try something more sophisticated, how about explaining a "divide by zero error" to your average TV audience. (my Linux machine just told me that "Division by zero is undefined," but you might have stopped a sixties computer with that one... maybe.

283:

I think there are either alternative or intermediate and messy steps.

For example, I can see automation of expense reporting. Instead of using Google Maps to come up with a plausible distance and taking a standardized mileage charge, the company car can cough up where it's been, how many miles it's driven, even wear and tear, and that can go to expenses. The downside of this is that if the employee was a responsible person, ran a personal errand in the middle of a trip (say to get coffee so they didn't crash the car), the AI would get pissy about the little side trip and the accountant would have to sort things out.

Ditto using the company phone to buy food. The AI in the phone could easily turn into the little company commissar nixing purchases to keep the employee healthy, or some such.

The other fun thing that will almost certainly happen is aftermarket and third party AIs will become available to employees for hacking AIs used by HR and Accounting to manage the company. In this scenario, accounting becomes less about just doing the books, and more a Red Queen race where the humans are forensic accountants trying to keep employees from using AIs to cheat on expenses.

This is simply an outgrowth of what's been happening for years in academia, where you buy papers instead of writing them, to the point where teachers have to screen students' work against online paper sellers to make sure they actually did the work.

284:

Yes, we all got a kick of how computer illiterate Hollywood writers were about computers.
I mean, how could a program cause a computer to catch fire?

Then IBM came out with the monochrome monitor & adapter card...

(It was possible to reprogram the display controller with invalid timings that caused the monitor to let out the magic smoke. Not a shining moment for whoever designed the monitors circuitry).

Note that the writers still are not very computer literate, although they may think they are because they use them all the time now.

285:

But all this is irrelevant since Arthur C. Clarke said (in one of his last novels) that AIs will soon arrive and there will be no human pilots anymore and all landings and take-offs will be safe.

Yeah, that's why they want to put humans on Mars...

More seriously, drones crash at absurdly high rates, compared with manned planes. They have a different ideal mission profile than do human planes. While I'll accept that it's entirely possible for drones to get much better at crisis flying, we're a long way from there yet.

286:

Agreed. If you've got a powerful language - C, ASM, etc. it's possible to program your hardware controllers to do really stupid things. One friend told the story of blowing up a more modern monitor than you described by making the horizontal and vertical beams only vertical. Apparently over time it heated the glass of the screen in a most interesting fashion...

But yes, on-screen computing is about 95% bullshit.

287:

Niala
NO
Try outer Scotland
Barra: https://en.wikipedia.org/wiki/Barra_Airport
Map reference: Here

288:

Heteromeles @ 285: " While I'll accept that it's entirely possible for drones to get much better at crisis flying, we're a long way from there yet."

Some AI types like to point out that we've been pushing the goal posts of true AI backwards each time we attain some form of AI. In the case of robots replacing human pilots on all passenger or cargo planes they point out that we've had autoland features for more than 30 years and that only FAA rules now prevent the implementation of take-off features too.

https://en.wikipedia.org/wiki/Autoland

289:

Scott Sanford @ 239:

I've liked the story from the Whole Earth Catalog, lo, these many years ago: in the mid-sixties, Mexico City built a new university, and paved *no* paths between the buildings. The university opened... and the next summer, they paved the paths between the buildings that people who lived and worked there made, rather than some asshole architect's idea of what might win them an award.

This story was also told of my alma mater, The Evergreen State College - so circa 1970, about fifteen years before I heard it as one of many campus legends. It's possible this idea was going around at the time, as the paths between the four original dorms and the main campus look like that. Hopefully Mexico City had less rain and mud than Olympia, Washington.

Or you could do what my almost alma mater did and pave over the whole damn campus & make it into a giant patio.

https://en.wikipedia.org/wiki/The_Brickyard_(NC_State)

An exaggeration of course, but not as much as you might think.

290:

David L @ 255:

but the council put up tape where they really don't want people riding and that seems to work.

Must be nice.

Around here the various riders were encroaching on an airport property. Which is also a big part of a watershed with rules on run off. Signs didn't work. So the airport put up a fence. Fence keeps getting torn down. And those not tearing it down (well mostly) are in court demanding the airport shrink their security perimeter and let them make/ride tails on the airport property in the watershed protected areas.

Is that RDU? Because my experience is that the FAA & Airport Police don't have much patience with people trespassing within their security perimeter. Tearing down that fence is a Federal Crime

292:

Actually, I'm not pushing the goal posts at all. Here's what I know, as of the last few years:

--The Navy was experimenting with drones launching and landing from carriers (tricky flying). They're obviously planning on fielding the tech, but AFAIK they're no there yet.
--The USAF can land their spaceplane drone X-37B, so they're capable of a hairy landing under controlled conditions (in this case, where they control the airspace and cyberspace the X-37B is landing in).
--The USAF still loses far more drones than it does manned planes, last I heard. This is considered acceptable, because they're cheaper and no lives were lost.
--Drones and plane autopilots have some stuff they're really good at (loitering in a bounded area, flying a course at a speed), and stuff they're terrible at (dealing with a crisis, taking off and landing in messy environments, dogfighting, etc.)

The basic problem is that the computer that will simulate a pilot's reflexes in total probably right now is probably heavier than the plane itself. To some degree you can get around that by networking the drone into a ground-based AI, but that introduces all sorts of interesting latency issues. Besides which, networking the drone with a ground-based pilot is still cheaper, although the latency problem crashes planes pretty routinely.

Now I'm in the minority that thinks the US Space Force isn't an intrinsically terrible idea, and this is why. Right now, we (the militaristic US) use satellites to intermediate between drones that aren't all that competent and ground-based pilots who aren't all that disposable, and it works okay for having murder-bots loitering in the troposphere above the Middle East. Everyone knows that's what we use satellites for, and so the obvious way to cripple our drone force is to take out the satellites.

And that's the rub. We may be able to field fully functional military drones now, but only if there's a fully functional military satellite network with low latency linking the drone with the data center flying it. That's a weak link, and I certainly wouldn't trust someone's life to it unless there was no other option. Worse, it's likely that, during wartime, the intelligence flying the plane will be limited to what can fit inside the plane, due to pervasive cyber and space-war.

I see three non-exclusive solutions.

One is to find simpler ways to fly--make planes giant mosquitoes, as it were, so the processing power will fit on the plane. Insects don't need as many neurons as humans to fly, so maybe it's possible to fly more stupidly than we do. Then we need ground-based commanders, not ground-based pilots.

A second is a variation on Komey's Law, where fully-skilled piloting AIs get light enough to fit inside a plane. We'll probably get here within a decade, two at the most, depending on the skill level desired (I'd settle for an AI that flew as well as a pigeon, but that's way far off).

A third solution is to continue the human-plane-AI partnership, where the humans and the AI partner in flying, with each assuming a dominant role in different phases of the flight, the AI captaining during the routine boring stuff and the human taking charge when things get more complicated. This is what we do now. What you don't want to do is to have the AI fly the plane except when it can't, because humans are really bad at handling unexpected disasters with little or no warning. We do better when we're paying attention and see the problem coming. So, as with cars, it works better if both controllers are looped in.

And there's no reason not to work on all three simultaneously, which I'm pretty sure is what aeronautical engineers are actually doing.

293:

I'm not considering military drones, I'm considering robot airliners.

294:

Employees cheating? How about the mods for the *company* to fake its books, so as to screw with the taxes, and loans...?

295:

Niala
There's always Westray to/from Papa Westray
And a YouTube clip of the same ....

296:

As opposed to the assembly language command, HCF (halt and catch fire)?

Btw, what wikipedia, etc, do *not* mention is that I've read, in the early days of DEC, it was possible to run a printer so fast that the paper would catch on fire....

297:

I'm not considering military drones, I'm considering robot airliners.

Same problems, except more casualties per drone failure.

Note that the USAF is basically the US military air cargo service, along with things that make booming noises. They're not looking at automating the air cargo sector, but rather the attack planes. I'm guessing there are a couple of reasons for this. One is that drones are better at replacing humans where humans die a lot, which is not on cargo flights. Then again, the cargo plane isn't just a plane, it's staffed by a squad of airmen who handle the cargo, and that's a fairly human job (there's a reason why loadmasters get the plane loaded and balanced. It's a non-trivial job). The third angle is that apparently logistics and routing get screwy and are often rerouted on the go. For trained humans, "New mission? No problem!" is attainable. A bit harder to do this with a drone.

298:

You mean like Ex-IQ.45 is currently getting investigated for? Boy, wouldn't that be fun: flip the company that makes the mods, then go after all their clients.

I think this is a way of saying it's the same kind of bad idea as having anything that only has illegal uses.

299:

I should also point out that there's supposedly a relevant argument going on in various of the world's militaries. That argument is whether to empower drones and robots to kill humans autonomously, or whether to require a human in the loop that decides when to kill others. Purportedly, the US and allies favor requiring a human in the loop, while the Chinese and Russians do not. Although I haven't heard the Chinese or Russian of this argument, so I remain skeptical. But the debate is real. Who pulls the trigger?

Anyway, the same argument goes for autonomous ships or autonomous planes. These are potentially deadly when mistakes happen. Where are humans in the decision loop? Robotic cargo ships might make sense in some cases, especially with a satellite uplink, because their decision loop is so slow that latency isn't a huge issue. Robotic satellites are a given, because space is innately hostile to humans. Ditto for ROVs in the deep ocean. But robotic planes and trucks? That's where things get messy. And robotic passenger planes, cruise ships, and buses? That probably increases the cringe factor for a lot of people beyond acceptable limits.

And yes, I get the Trolley Problem. But having an AI make the decision in the Trolley Problem merely shifts the issue from the decision at the switch to the decision by the programmer over what solution to chose in such cases. Is that what we want?

300:

Trolley Problem merely shifts the issue from the decision at the switch to the decision by the programmer over what solution to chose in such cases.

I thought the big thing in AI was diffusion of responsibility. The AI can't say why it did what it did, the designers of the AI have no idea and no way to find out, whoever provided the training data ditto... all we know is that automated trucks from X brand seem unusually likely to kill black people in crashes.

301:

"in the early days of DEC, it was possible to run a printer so fast that the paper would catch on fire...."

The version I heard was that, although the printer was about the fastest available, that was not what caused the fires.

AIUI the printer did not use ink, but burnt the output into the paper using a high amperage current. What could possibly go wrong?

JHomes

302:

Heteromeles @ 299: "But robotic planes and trucks? That's where things get messy."

That's why there is still a lot of work to be done. Last year Amazon bought Zoox, the robot taxi company, for 1.3 billion dollars. The CEO of Zoox doesn't even want to talk about a future Amazon robot parcel truck.

https://www.youtube.com/watch?v=GsSzRMdrZjg

303:

Re: 'Along with ML becoming, in practice, experimental psychology, this will dramatically change the way society actually thinks over time. The next few decades look fascinating.'

Society could change as a consequence but in which direction will depend on the acceptability of that org's primary goal, i.e., profit vs. human impact. So I can see HMOs replacing their human clerks with AI for this: much easier to evade blame by pointing at a sophisticated mathematical black-box than doing any soul-searching. Dollars are much easier to tot up and compare (advertise).

There's also the possibility of AIs being fed a very selective data diet. We'd probably need to create industry standards for all the ways that an AI can screw up anywhere from going rogue to spewing out profitable inhumane nonsense.

304:

What I heard was that it wasn't the actual sheet paper that caught fire, but accumulations of the dust it shed going through the printer. Maybe the sheet paper was acting like the belt of a Van de Graaf generator and the sparks ignited the dust.

Burning marks into the paper with an electric current is what the Sinclair ZX printer did on its horrible little expensive metallised bog rolls. It doesn't work without the metallisation because paper is an insulator.

305:

Either way, I've seen drum line printers in action and they're not reassuring machines. Lots of offices used to have them, back in the days when companies had started using them for accounts but the tax system/legal system had not adjusted and still required paper copies of everything be kept (for seven years!) So pallet loads of fanfold paper got trucked in, run through the printer, then stored for a long time before being dumped. Sometimes shredded, often just dumped. Or burnt.

306:

I thought the big thing in AI was diffusion of responsibility. The AI can't say why it did what it did, the designers of the AI have no idea and no way to find out, whoever provided the training data ditto... all we know is that automated trucks from X brand seem unusually likely to kill black people in crashes.

I can see where that would be appealing in the "mistakes were made, algorithms will be updated" sense, but that's not the only one.

Here are a couple of alternate scenarios:

--Lance Corporal Cylon Version 2.3b starts World War 3 by shooting down a manned, unarmed surveillance plane performing its treaty-allowed monitoring of a disputed border. After a few billion people have died, Cylon's makers are assessed primary responsibility for starting the war by not upgrading their plane identification module in the five years previous to the start of the war. The peace treaty dismembers their country to prevent such systems from being built again.

--BigMuddy Amalgamation Services goes into receivership when a court determines that the programming of their automated delivery system on average kills more people per month in Trolley Problem class collisions than random decision-making would, due to faulty programming and inability of humans to affect the decision once it is made.

I'm being my usual sarcastic self, but there are plenty of situations where it's kind of important to have control over who lives, who dies, and who decides (to misquote Hamilton).

307:

Re: NFT

We're past 300 so ...

Saw headlines recently about some NFT artwork fetching $60M+. Hadn't realized that this was already a fairly strong market therefore not that farfetched as a possible target industry for a Mars economy.

https://www.bloomberg.com/opinion/articles/2021-03-04/the-nft-phenomenon-is-for-real

Here's the art:

https://financialpost.com/personal-finance/high-net-worth/a-three-week-old-nft-just-sold-for-69-million

308:

As opposed to the assembly language command, HCF (halt and catch fire)?
--Given to an undocumented Motorola 6800 opcode (when executed, it would halt the processor and start incrementing the address bus every cycle, the article's author guessed it was used in chip testing).

Btw, what wikipedia, etc, do *not* mention is that I've read, in the early days of DEC, it was possible to run a printer so fast that the paper would catch on fire....

At the University I went to, students figured out that they could get a bunch of fan fold paper by writing a small program that put a page feed carriage control in column one (remember that?).
(Ok, so the systems programmers fixed that).

Somebody else accidentally wrote a program that put a "don't feed the carriage" control in column one (plus, I think).
For many, many, many, lines.
Yes, the printer (or the paper) caught fire.
(And the systems programmers fixed that one too).

Getting the card punch to punch every hole on the card was a good way to jam that up as well.
(I suspect the systems programmers fixed that one too).

And for something in this century:
https://arstechnica.com/information-technology/2011/11/hp-printers-can-be-remotely-controlled-and-set-on-fire-researchers-claim/

309:

Heteromeles @ 283: I think there are either alternative or intermediate and messy steps.

For example, I can see automation of expense reporting. Instead of using Google Maps to come up with a plausible distance and taking a standardized mileage charge, the company car can cough up where it's been, how many miles it's driven, even wear and tear, and that can go to expenses. The downside of this is that if the employee was a responsible person, ran a personal errand in the middle of a trip (say to get coffee so they didn't crash the car), the AI would get pissy about the little side trip and the accountant would have to sort things out.

The thing about being assigned a company vehicle is "personal use" is built into the system and as long as personal use doesn't exceed the parameters of the understanding between employer & employee it's not a problem.

That stop for coffee doesn't even require a "side trip" and the AI doesn't give a shit.

310:

OK
Where is the ghost & where are the "people" in this machine/software/people construct/problem?
About lunchtime GMT on 26th February 2021 a well-known ISP's ( TalkTalk ) connectivity to it's contracted users went down - badly.
Emails were not sent or recieved, people using home clients, such as "Outlook" got confusing error-messages & no emails.
Something at the ISP had crashed, or had been hacked (?) or had simply failed. Gradually, over the following days, most people's connectivity was restored, but not everybody ( Guess how I know all this? )
People went onto the internal support & back-up channels ... a persistent message was similar to: "Password not recognised or faulty, contact your ISP"
Some of us were helped, some not. Some of us were told to remove their account name completely & re-install, to clear the problem ( Once emails had been backed-up, offline ) ... except it didn't.
Now some of us are left with an Outlook account that WILL NOT RECONNECT AT ALL, still with an "unrecognised password" fault, & no apparent means of getting their emails working on their home computer's Outlook account at all.
OK - who or what or both is wrong here? And how does one fix it? Is the fault with the ISP ( Seems most likely ) or with MS Outlook, which may have been locally corrupted.
As you have realised by now, yes, this is what my current situation is.

311:

That argument is whether to empower drones and robots to kill humans autonomously,

I strongly recommend Linda Nagata's SF novel "The Last Good Man", which takes this topic seriously and is scary as all hell as a prediction of where the world is going. Near-future SF is hard, she does a really good job.

And yes, I get the Trolley Problem.

I've tutored university-level ethics, and I don't believe in the Trolley Problem.

People like ethical dilemmas because people like tricky mental problems, but that's not how real-world ethics works. Real world ethics isn't a mental game. It's really about "noticing what's going on" and "giving a fuck". It's often about self-discipline to do what you know is right. But it is very, very rarely about "moral dilemma decisions".

Real examples:

Mum drives the kids to school instead of them biking because she thinks the traffic around school is unsafe.

Grandad buys a SUV because he feels driving his grandkids around in it is safer a hatchback.

But actually mum is increasing the chance of a kid dying in the busy traffic around school. Just not her kid. And grandad is increasing the chance of someone dying in a car-crash while decreasing the chance of his passengers dying, because SUVs are much more dangerous for pedestrians and cyclists.

That's how the real world works. No-one decides "I'll kill these 3 to save that 1". That doesn't happen. What happens instead they decide is "I'll move the risks around a little bit, in favor of my kids". And we don't write angsty article about the Trolley Problem when they do that.

Which gets us to AI. People act like there will be a place in the AI program where it says
my_passenger_dying_weight = -1.0;
other_people_dying_weight = -0.5;

But that won't happen. There will be nothing like that. The self-driving AI won't know what a person is, or what death is. There won't be a point at which it is *deciding* "I'll kill the 3 little kids on the pedestrian crossing to protect my driver". Instead what there will be is a whole bunch of reflexes, responses, etc, within the AI, and some emergent behaviour. And some of it might be morally selfish but, like mum and grandad in my examples above, there's nothing unusual or novel about that.

My point is that there is no Trolley Problem, because there is no "Trolley Problem" decision point at which someone (or some AI) suddenly chooses who dies. Instead there are lots of small choices, each one individually fairly innocuous. That these little decisions move risks around is going to be true in the design of self-driving AI, but it's true in lots and lots of other things as well and we don't say they embody "the Trolley Problem".

312:

Those of us working for smaller companies (hi Charlie) the process is unlikely to involve intelligence of either sort. The last time I was visiting client site my then employer gave me a credit card and told me to use it wisely.

I found out months later that the accounts people were just told to pay the bills and the review process was "if you are that concerned *you* talk to him". Which happened, because I spent rather more thousands of dollars buying a new laptop than I expected the card to accept. The client paid, because the client was the one that broke the old one, but I only found out because I gave the receipts to accounts and they said "we don't care, your expenses don't get reviewed" and I had to explain that they did care because they had to bill the client...

The best part in this is that "accounts" was the boss's wife 😆

313:

Mum drives the kids to school instead of them biking because...

the school has a screaming hissy fit about the risk, the danger, the unusuality and the frankly bizarre behaviour that would lead a parent to think this is in any way acceptable. Children must be driven to school and then lectured about obesity, that is the way things are done now{tm}.

It actually varies seemingly randomly. The children of BikeSaint in Sydney* were variously boggled at, accommodated and eventually accepted when they biked to school. A line was drawn at the tall bike, though. Other friends report everything from outright official bans through hostility to surprise that anyone would think permission would be needed.

* I assumed she was sufficiently (in)famous that no mere school principal would dare raise the issue. But what would I know...

314:

Like Abbottsford, or as it is now known, Lower Abbottsford... https://en.wikipedia.org/wiki/1979_Abbotsford_landslip

I would not have anticipated the slide from the existence of a quarry - but then, obviously the experts on site didn't either.

I wish you luck with your post-bushfire land opportunity, and expect you're already enjoying flat stationary ground, but the thread has me reminded of Malibu mudslides. Southern California has a lot of chaparral, low dry brush which occasionally burns off very dramatically; if heavy rains come after the plants binding the soil have burned off, formerly stationary homes in the area become distressingly mobile. Or, in Malibu, just because - the area is known for wandering geology irrespective of the weather.

315:

They should have kept the requirement in place and massively increased the timespan for purposes of carbon sequestration...

316:

Saw headlines recently about some NFT artwork fetching $60M+.

Just what we need. Another technology with a huge carbon footprint.

An average single-edition NFT uses as much energy as a typical European Union household does in a month, according to the calculator. Multi-edition NFTs are even more damaging. … Grimes’ high-profile drop resulted in 122 tons of carbon pollution.

https://earther.gizmodo.com/how-to-fix-crypto-art-nfts-carbon-pollution-problem-1846440312

317:

That stop for coffee doesn't even require a "side trip" and the AI doesn't give a shit.

Lots of offices use computer monitoring to micromanage their employees. Look at the dashboards being built into subscription software, for example. I would expect that many managers would extend monitoring if they were able to.

318:

Mum drives the kids to school instead of them biking because she thinks the traffic around school is unsafe.

She's not wrong. Have you watched parents around a school? Bloody dangerous idiots.

If we could eliminate parents driving kids to school, it would be a lot safer.

(My car would also have fewer dents, because parents park in the staff parking lot and apparently don't give a shit about hitting cars on the way in/out. I've also had to jump out of the way to avoid being run over by women in SUVs looking at their cell phone while arriving at school at 7:00 (school starts at 8:45).)

319:

Back to some of the original post (and re use of Category Errors to hide(?) from determined searchers (hmmm :-), in a deleted SotMNs("Please Don't Dox Seagulls"[1]) comment), been noodling on hiding from (current) AIs.
This Heinlein quote suggests one class of approaches:
It's not enough to be able to lie with a straight face; anybody with enough gall to raise on a busted flush can do that. The first way to lie artistically is to tell the truth but not all of it. The second way involves telling the truth, too, but is harder: Tell the exact truth and maybe all of it but tell it so unconvincingly that your listener is sure you are lying. (Robert A. Heinlein, Time Enough for Love)
"unconvincingly" is shorthand for a broad set of methods. The listener can be manipulated (e.g. through triggerings of their own biases (sometimes deep conceptual biases)) to believe that you are lying, that you are telling a joke (why not both? :-), etc. (Heinlein also neglected multiple possible parsings (with one favored by common conceptual biases), multiple levels of metaphor, puns, etc.)
Some of the AI spoofing methods are similar in spirit; in the CLIP case, a textual label on an object tricks it into "You look like an X but I classify you are a Y because Words are more important than mere appearances". In principle the deception could be flagged as (potentially) such by e.g. by recognizing text and editing it out of the image and re-running classification.
[1] I do have a list of recent names (missed one, used after midnight my time).

320:

A line was drawn at the tall bike, though.

Are those considered exotic where you are? Here in Portland, Oregon, USA we've got various bike nuts riding around on those, welded together out of whatever spare bicycle parts might be laying around. But we also have a bagpipe playing unicyclist (video), mobile pedal pub tours, and an annual Naked Bike Ride - so we're pretty jaded to weird pedaled things.

PS: Darn, I forgot zoobombing...

321:

I wish you luck with your post-bushfire land opportunity, and expect you're already enjoying flat stationary ground, but the thread has me reminded of Malibu mudslides. Southern California has a lot of chaparral, low dry brush which occasionally burns off very dramatically; if heavy rains come after the plants binding the soil have burned off, formerly stationary homes in the area become distressingly mobile. Or, in Malibu, just because - the area is known for wandering geology irrespective of the weather.

Conflated a couple of different things. I grew up in chaparral country. While I agree that there are mudslides, the complicating factor is that many chaparral plants are deeply rooted. When those roots survive fires, they do a pretty good job of holding the slope together. The things that confound this:

1. Believe it or not, too many fires can get rid of chaparral. It's fine with being burned once per century, or maybe once per 50 years. Once the fires start happening more than every 30 years, species start disappearing, and with fires more than once every decade, it's a few really tough plants and annual weeds. Malibu Canyon burns every seven years or so, because Malibu Canyon to a first approximation is shaped like a funnel facing into the Santa Ana winds. If a fire starts anywhere in or near that funnel during a wind storm, the Santa Anas shove it through the canyon down to the coast.

2. The geology of Malibu is fairly recent oceanic sediments turned into sandstone and mudstone. That stuff falls apart. I learned to climb on similar rock, and we used garden hoes to knock footholds in slopes. Or stood on plant roots, which are more stable.

Anyway, slides are getting worse, not just because of fires, but because fires are happening later (sometimes the rain puts them out, as in Montecito a few years back). Also, people are being stupid about how they clear steep slopes, meaning there's more to erode and less to hold it up. Fire inspectors don't think about landslides.

Now, if you want to see a really good example of fire-resistant architecture, check out Pepperdine University. It's at the mouth of Malibu Canyon, and the whole campus is designed to shelter some 3000-3500 students in the library in the center of campus, with protection from firefighters. It's survived the seven fires that hit it quite well. Thing is, it's purpose-built for surviving fires, and most buildings in high fire areas in California are not.

322:

But are they riding them to high school? That's where the line was drawn.

323:

Have you watched parents {drive} around a school

Yes, and I try to avoid riding between ~1430 and 1630 for that reason. There are too many schools and not enough sanity.

In many ways I think it would be better to only allow students to drive around schools, because they at more likely to make the connection "school uniform = people I know" rather than "my baby is the whole world and nothing else matters".

I did wonder a while ago how many car bombs it would take before a collection of law changes and concrete barriers made it impossible to drive near schools at all. But I had a look and it doesn't look as though even the craziest merkins have done that, so it seems unlikely the ozzies would.

324:

Robert Prior @ 318 "If we could eliminate parents driving kids to school, it would be a lot safer."

Don't you have schoolbuses?

You know, big yellow things that stop the traffic dead in both directions, whenever they pick up or drop off students.

325:

We have school buses, but they're mostly a rural thing. Australia has the theory that all state-provided education is much of a muchness so there's no point trucking kids across town, better to build a school where they are.

Footnotes...

  • we have "selective schools" which have ability requirements
  • but not the US-style "selective" meaning race
  • We also, despite our history, have mostly given up on ghettos. for example "Blacktown" is multiethnic, and even whites live there
  • "town camps" OTOH, are where abos stay while they've visiting towns but can't stay in them.
326:

Also "school bus" is an ordinary bus just doing a school run. They don't stop traffic when they pick up or drop off. They have lights that flash and you're supposed to overtake at 40 km/h, which is about the lowest speed that gets you near 100% kill rate when you run over children.

No it's not supposed to make sense.

327:

Here, for most children in basic comprehensive schools, the school is close enough that it's possible to walk there. Now we are a bit farther from the school than most, but it's still only a bit over a kilometre there, so perfectly possible to walk or bike there for the kids.

If not (for example in the countryside or special needs or something), there is some kind of taxi service, either just a regular car or a minibus. I think mostly the service is bought from regular taxi companies.

However, in many places, also where I live, parents think it's easier for everybody if they drive the kids by car to the school, so there's much traffic just where there are a lot of kids at the same time. It is somewhat dangerous, obviously, so then more parents use the cars. And schools try to tell them that "hey, your kids can physically and mentally walk the some hundreds of meters to school and it's better for them anyway to do that", usually to little effect...

328:

I fully agree things will get messy, and data will be "diverted", etc.

My larger point is not just the example of expenses, but that the work of ML/AI will be part of every job - classification of things, etc has thousands of examples in daily life and the toolkits will be present in office and other PC systems. Similarly, all data will be managed in a way to make it ingest into ML easily.

I'm saying that meta-cognition - thinking about biases and how we think - will go from a niche interest to being widespread. Similarly, understanding how Google, etc work (merging datasets, the issues of privacy) will be trivially understood.

The interesting question is what happens _next_.

329:

I get the feeling that there are no schoolbuses (big yellow things containing 72 kids, with a camouflaged police car sometimes following them to catch anybody who drives near) at all outside of North America.

https://www.youtube.com/watch?v=jbp04AftXVw

331:

There are school buses in the UK - here is one of the ones that pass near us.

Big, yes. Yellow, no. 72 seats? Who knows

There are ones that are yellow. Just not our local ones. I have been on a Canadian one, heading north from St Anthony in Newfoundland, and it wasn't hugely luxurious, but it wasn't too basic either.

332:

Don't you have schoolbuses?

Yup. When I was growing up in Saskatoon, they were only for rural kids. I took the city bus halfway across the city when I changed schools to attend a specific program.

When I moved to Toronto in the 90s I was surprised to see them in the city. Some elementary students take them — no idea what the criteria are. But many parents drive their kids, elementary and secondary. One day I was going home and followed a parent who had picked up their child (student athlete and flute player) and they ended up at a house about four blocks away. Driving a teenager four blocks rather than letting them walk? Every day?

333:

Yes, that's the same kind of thing one school I was at used to ferry in kids from the villages: just an ordinary bus doing school duties. So did the school for older kids on the other side of the canal. But the school for younger kids on the other side of the road didn't, so our mums used to drive us in to that one. It was a bit bloody silly for the family whose older kids were picked up by the bus (from right outside their door) but the younger kids had to go by mumbus for exactly the same journey.

334:

In the UK, there is, or was (it was decades back that I was last even vaguely invested in this) a distance component. My sisters went to school in the nearest town, but we were 4 miles away, so the school bus took them for free. It was a shame it was a mile to the bus stop from our house

(We lived at the end of a rural lane, nothing else on it, and nowhere for a bus to turn round either.)

335:

Telling the truth unconvincingly?

The funniest programme on UK TV has experts in this process. But Bob Mortimer is the best.

https://youtu.be/1vE8ExuuNZQ

336:

In the UK, there is, or was (it was decades back that I was last even vaguely invested in this) a distance component.

Everywhere I've lived in the US has had a law that the school system (if public) must provide transportation to all students who lived over 1 miles away. (not sure if road mile or bird mile)

But most picked up any students more than a few blocks away. Where I've lived the weather can be so variable that making kids walk to school creates a lot of issues. When it's just above freezing with maybe light rain on the way to school and light jacket weather by lunch it creates issues with how to dress the kids if they are going to walk. Especially the smaller ones.

And when you have "busing" it gets real complicated. (I can argue both side of this one.)

I can remember being amazed when I visited my cousins in Detroit suburbs around 1963 and they were walking to the school about 2 blocks away. Back home I was considered living close to my school and was nearly a mile away. But in hindsight it (the Detroit school) was a small school of only a 200-300 kids. Most schools in much of the US now are larger just to keep the system wide facility costs down.

337:

I recently came up with The Answer to the Trolley Problem: move the switch just as the trolley hits it, so at least the back wheels derail, and the trolley falls over.

338:

Growing up in Philly in the 50's and 60's, the only time I ever saw a schoolbus was when they took us on a class/school trip, like downtown to see the Phila. Orchestra.

I walked or took city transit from 5th grade on. In shools, you got student tokens, significantly cheaper than the regular fare.

And yes, I read 60% of Americans are overweight to obese. Or, as I like to say, "let's see, you get a riding mower to do your lawn, then pay how many hundred dollars to join a gym, and drive your SUV a mile to the gym?"

339:

The funniest programme on UK TV has experts in this process. But Bob Mortimer is the best.
Thank you. Took a few notes; he is good.


340:

When I first moved to Ottawa back in the 80s I had an apartment near Hogs Back Falls, on the top floor (18ish). So I usually took the elevator, although if the lines were long* I would take the stairs. There was a yuppy-type chap who lived on the second floor who took the elevator down to the ground floor to go play squash. Like, it would have been faster to walk down the stairs rather than wait for the elevator, but he never did.

He did this, as far as I could tell, every day. Keeping fit is important.


*The chap who delivered the newspaper would claim an elevator by the simple expedient of blocking the door while he was delivering, so for half an hour a day things could get crowded.

341:

Off-topic update: Trump Force One/N757AF/T-Bird, the Trump-org(one of them)-controlled 757, is still parked at New York Stewart International Airport, and is still missing (as of 12 Mar 2021 13:20 EST) the port engine (with opaque white plastic wrapped around where an engine should be mounted).
This makes me smile. (It is (additional) evidence suggesting that Mr. Former Guy is less wealthy than he claims.)

Also, QAnon believers and contemporary US "Christian" grifter-prophets are continuing their melding, and their escape from fact-grounded reality (and also losing believers):
Trump President Of Heaven Now, Says QAnon Prophet Guy (Robyn Pennacchia, March 12, 2021)

If you can go in the spirit, if you can see what's in Heaven, who's sitting in the throne; go up and look at the presidency seat in Heaven, see who's there. It ain't Biden; it's Trump."
Of course, it is possible that they all just hate his guts up in Heaven and this is a big prank.
(My neighbor down the street still has a Trump2020 flag flying underneath an American flag on tall pole. Also saw a half-dozen in (far) upstate NY Feb 28 2021. They, including Former Guy, still don't believe/understand that they were(/are being) outplayed, even in the religious domain.)

342:

I'm not clever enough to judge this but it seems a good fit for the neighbourhood - Finding Mona Lisa in the Game of Life - https://avinayak.github.io/algorithms/programming/2021/02/19/finding-mona-lisa-in-the-game-of-life.html

343:

My kids rode the bus to school for a decade, ending this year because COVID. Apparently we are now about 10 metres inside the 'ineligible' ring and the one kid remaining in the school system must walk or be driven.

I am an ogre, but not so much of one that I will force him to trudge through the driving rain and up a 2 km steep hill each morning. I also work half a block away from the school myself, and start at the same time. So I drive him, and he tends to walk home (after 'hanging out' at the school yard or a friend's house). Driving the kid to school goes against all my values, only barely rationalized by the fact that I usually need my vehicle at work.

With the weather turning we will begin riding our bicyles, there are off-road trails through the forest that get us to the school.

344:

And yes, I read 60% of Americans are overweight to obese.

Things are not as simple as you might want to think in terms of correlation/causation.

My kids got driven to school a lot. And took buses a lot.

My daughter played organized baseball/soccer into her later teens. She still runs in her later 20s.

My son played lacrosse in high school and now does regular workouts plus rock climbs. Indoors and real cliffs. He owns the gear. Just turned 30.

Me, I use a push mower.

Now loading that large van full of stuff from the 3rd floor apartment in Texas in July about did me in. 40+ moving boxes full of stuff. But those two beefy looking guys doing the same across the hall also looked worn out. I guess we were both having issues with the 100F heat.

345:

But are they riding [tall bikes] to high school? That's where the line was drawn.

Now that you ask I realize I don't recall either way. Despite living just down the street from a high school named after a distant relative, I don't recall noticing any of the kids riding bicycles that were out of the ordinary. And this isn't a good year to look.

346:

You know, it's actually fairly easy to understand why so many right wing American Q-nuts want an apocalypse. If they think Trump's a good guy, it's just their subconscious lashing out and making them hate the reality they've consciously yoked themselves to. It's just another version of the cognitive dissonance of worshiping a god described as a being of infinite love, who will nonetheless condemn you to an infinity of suffering if you do not perform the arbitrary and capricious acts his self-appointed intermediaries require of you.

Note that White Jesus Wielding a Sword and Killing Brown People is Hitler's version of Christianity, not what's in the bible or even in most churches. But if that's the cult someone believes in, I can see where they want it all to be over, at least subconsciously.

347:

And yes, I read 60% of Americans are overweight to obese.

Don't take that too literally. By the BMI scale I've been "obese" for years, just because I'm 6'3" but not built like Spider Robinson.

348:

I spent a lot of my summers in the 1960s in Vermont, on the shores of Lake Champlain.

Obesity was not a statistic. It was a reality, as I looked around me at US citizens. A lot of them were fatter than any people in Montreal or Quebec City, my reference points. And the fat were more numerous, in proportion to the fat in Montreal or Quebec City. I was in the land of the fat and the free.

I don't know how much of this has changed in fifty years or so. I don't travel much to the US. But when I did go to scientifc conferences now and then I couldn't help notice that I had more fat people around me in the airports, than in Canada.

Oh, they were not "grossly and immensely fat" like Baron Harkonnen, but they were fat enough to make a difference. I never saw such a difference when I was back home, or driving around Europe in later years.

349:

Conflated a couple of different things...

To say the least!

Thanks for that expansion of my throwaway bit about Malibu. It's been far too long since I was in the LA area at all and I don't think I've ever been in Malibu Canyon when the Santa Anas are blowing; I can't imagine it's pleasant.

A burn frequency above 30 years is also also what I read is necessary to prevent species loss in chaparral. That number may vary for similar biomes on other continents.

350:

but they were fat enough to make a difference.

Yep. I'm 30 pounds over what would be a great weight for me. And 45 past that college weight. And most people around me seem bigger than me relative to their height.

351:

Well my trouser waistband size has not changed in the past 40 years ( 34" = 76cm (ish ))
Mass has probably gone up a bit, because C-19, but was steady at 80kg before that - might be 71/71.5kg, now.

352:

I don't know how much of this has changed in fifty years or so. I don't travel much to the US.

Apparently, the average American is nearly 20% heavier than they were in 1960. And women now average what men averaged then.

https://www.washingtonpost.com/news/wonk/wp/2015/06/12/look-at-how-much-weight-weve-gained-since-the-1960s/

When I last visited America, in the 1990s, I was astounded at how much rounder people looked. Not just one or two — sooo many round people.

Not saying Canada doesn't have round people, but the proportion is a lot less than I saw in America.

353:

And in other news, apparently the Kentucky Senate is pushing forward a bill to make it an offence to insult the police. Wonder how the Constitution's Framers would view their modern acolytes*…

https://www.kentucky.com/news/politics-government/article249876428.html

Text of the bill:
https://apps.legislature.ky.gov/recorddocuments/bill/21RS/sb211/bill.pdf


*Sarcasm. I suspect that the original Framers would react to modern 'originalists' the same way a certain carpenter would react to modern Evangelicals…

354:

Insult the police? Why bless their hearts!

Wonder if I'd get ticketed for saying that in Kentucky.

Of course, San Diego's finest until this year were known to give tickets for seditious language to people for the crime of being stopped by the police while brown, so I certainly have no cause for waxing sarcastic. None at all.

355:

Why bless their hearts!

I wonder how many people outside of the US get the joke.

356:
Why bless their hearts!

I wonder how many people outside of the US get the joke.

I certainly don't get it, and googling didn't help.

Can you explain to a poor German guy?

357:

"Why, bless their hearts!" is that very rare thing: a phrase used by USAns with deliberate ironic intent (it's frequently used with intonation that inverts its apparent meaning).

358:

a phrase used by USAns with deliberate ironic intent

It's mostly a Southern thing. Non-southern USAns often don't get the ironic usage of it.

359:

The variation I've mostly heard is "Well bless it's pointed little head", usually directed at the annoying victims of "Rectal-cranial inversion".

360:

It's part of a category of "southern compliments" that are really insults. In the "genteel south" you never directly put someone down. You always do it indirectly.

"Bless his/their heart(s)" is basically saying "too bad they don't know what they are talking about".

There is an entire category of things like "most people couldn't pull off that hairdo" and so on.

361:

Since I wrote my predictions on spaceflight going forward, some new stuff has happened.

1. RocketLabs is building the Neutron rocket to challenge the Falcon 9. In terms of mass to LEO, Neutron is closer in capability to the Ariane 4 than the Falcon 9, it shows one of the following

a. The SSLV market is likely to be saturated soon, as could the cubesat market
b. There's an expectation that other companies will build their own satellite internet constellations
c. Starlink will absorb most of SpaceX's launch capacity, leaving room for new entrants to grab the rest of the market
d. Either Starlink-class satellites will replace cubesats, or it'll be rare for cubesats not to launch as constellations

2. Relativity Space and Astra are also proposing Falcon 9 challengers. Astra managed to launch a payload into space, but the fuel was too oxygen-rich to get it into the proper orbit. Relativity Space is a different company altogether. It plans to 3D print the entire rocket, not just the engines. When asked, they hinted at bird-bone-structured fuel tanks. To my knowledge, they've received the most DoD subsidies of any SSLV company that hasn't launched a rocket into space (currently just RocketLabs, Virgin Orbit, and Astra)

3. Now that NASA is no longer dependent on the Soyuz, the Russians plan to resume launching paying space tourists
https://www.space.com/russia-launching-space-tourists-2021.html
https://www.space.com/space-adventures-roscosmos-tourist-flight-spacewalk-2023.html

4. Also, SpaceX has several private flights planned for the next 5 years
https://en.wikipedia.org/wiki/SpaceX_Dragon_2#List_of_flights
They have a contract to send Tom Cruise to the ISS, but I don't see his name on the roster of any of those flights.

5. Blue Origin is planning to spin their New Shepard capsule to simulate lunar gravity while in space. I don't know if this means that they're pivoting NS from a suborbital space tourism company to a more capable sounding rocket? Maybe I'm just reading too much into it?

362:

In the novel I'm working on one character says to another, "Your hat is quite in advance of fashion."

363:

Yep. But I just can't imagine a nice southern lady saying that phrase quite like that. :)

364:

Bingo!

That was one of the points I was trying to make, that outlawing insulting the police in the South has a strong bias towards certain social classes. The people who presumably would insult the police are, by stereotype, poorer and presumably browner than the fine white people who (in theory) know the art of the southern compliment.

The irony that Black Kentuckians might be able to teach them a thing or two about using ambiguous and multivalent language with white people may well escape those fine lawmakers. They are fashionably anti-intellectual these days.

The other point is that "Why bless their hearts!" is such a stereotypical comment that even a Californian such as myself understands in general what it means. Therefore, deploying it against the fine Kentucky policefolk when they are zealously pursuing every aspect of their duty might actually get me a ticket.

Incidentally, "why bless their hearts!" seems to be used more by women than by men.

365:

The Cruise flight contract is with Axiom rather than directly with SpaceX so it will probably form AX-2 or AX-3. It is likely to need at least a good outline of the script to be in place with production of the rest of the film scheduled. NASA has also recently hiked its prices for use of ISS facilities, particularly for "any project that is primarily intended for marketing, advertising, or entertainment purposes.", so that may have caused a budget adjustment.

366:

It's a fantasy story, so she's not from the south. She's a human who's just met an Orc for the first time, and the Orc is wearing a hat that's just a little too much!

367:

AI & "ghosts"
In today's "FT" there is a book-review of the almost-theological infighting between the "Connectionist" & "Symbolist" approaches to & progress (or not) in AI.
The book is: "Genius Makers" by Cade Metz.

368:

Robert Prior @ 317:

That stop for coffee doesn't even require a "side trip" and the AI doesn't give a shit.

Lots of offices use computer monitoring to micromanage their employees. Look at the dashboards being built into subscription software, for example. I would expect that many managers would extend monitoring if they were able to.

But the office still has a coffee station in the break room ... because micromanagers drink coffee too; and have to take bathroom breaks and do all the other messy organic things people have to do. And the law requires certain minimums.

"Outside" staff who have to use a company vehicle (because it's the most cost effective way for the company to get the company's work done) are a bit more difficult to micromanage.

369:

Rocketpjs @ 343: My kids rode the bus to school for a decade, ending this year because COVID. Apparently we are now about 10 metres inside the 'ineligible' ring and the one kid remaining in the school system must walk or be driven.

I am an ogre, but not so much of one that I will force him to trudge through the driving rain and up a 2 km steep hill each morning. I also work half a block away from the school myself, and start at the same time. So I drive him, and he tends to walk home (after 'hanging out' at the school yard or a friend's house). Driving the kid to school goes against all my values, only barely rationalized by the fact that I usually need my vehicle at work.

With the weather turning we will begin riding our bicyles, there are off-road trails through the forest that get us to the school.

Where I grew up the city school system didn't have school busses. There was a city bus that stopped in front of the school & you could ride that for about $1.50 a week.

My dad didn't drive us to school, but because he passed by the school on his way to work, he'd drop us off & we walked home after school (not quite a mile).

The ogre remark reminded me of something smart-ass one of my friend's dad said to us once when we asked him to give us a ride somewhere ... "Normally, I'd let you kids walk, but since the weather is so bad, you can crawl"

We did get the ride, but also a whole bunch of bullshit.

370:

David L @ 360: It's part of a category of "southern compliments" that are really insults. In the "genteel south" you never directly put someone down. You always do it indirectly.

"Bless his/their heart(s)" is basically saying "too bad they don't know what they are talking about".

There is an entire category of things like "most people couldn't pull off that hairdo" and so on.

I think my favorite is "You have a point ... [but if you wear a hat no one will ever notice]."

"Bless his/her/their heart" also implies the person so blessed suffers self inflicted incurable stupidity.

371:

Here's an interesting short film I ran across today.

Antikythera mechanism: https://vimeo.com/518734183

372:

[Old Fogey Mode]The younger generation are just soft - we did that and more.[End Mode]

'Tis true, sir.

373:

Just to annoy the physicists out there, Quanta Magazine, fount of truth that it is, this week published this: https://www.quantamagazine.org/quantum-mischief-rewrites-the-laws-of-cause-and-effect-20210311/

The important gist is that it's possible to set up quantum experiments (using entangled, polarized photons) where it's impossible to tell whether event A caused event B, or event B caused event A. There's also apparently a bunch of mathematics that led to these experiments, so these were tests of theory, not wild observations in search of a theory. They're part of the eternal churn of trying to reconcile relativity and quantum mechanics.

If this is right (titanic if!), then it may turn out that correlation is more important than causation in certain cases, and that causation, like newtonian physics, is a human-scale subset of a more general reality that's noncausal at other scales.

Although it's not discussed in the article, this irresistibly reminds me of the old saw for why FTL implies time travel. The pithy/sarcastic shorthand is that there are three options, relativity, causality, and FTL, of which you can pick two, and the presumption is that our universe has settled on relativity and causality. If it turns out causality is an emergent phenomenon, then we may actually live in a relativistic universe, and...

Have fun getting annoyed with this article. I'm sure some will judge it not even wrong.

374:

I visited the Boston Computer Museum in the 1990s. There was a full scale see-through model of the Antikythera mechanism. I was impressed!

I remember absolutely nothing else of my trip to the Boston Computer Museum.

375:

we did that and more.

In the snow.

Uphill.

Both ways.

376:

""Bless his/her/their heart" also implies the person so blessed suffers self inflicted incurable stupidity."

It's familiar in the UK as a phrase used by grannies cooing over tiny kids who have done something cutely stupid because they're too small to know better, like trying to eat pictures of sweets, or describing some model of how the world works which is egregiously and amusingly wrong but can be readily understood to appear reasonable if you only have two years' worth of observation to base it on. By extension, it can also be used of adults to imply that their mental development has not progressed beyond that level.

However, most of those grannies don't really say anything much these days, and when I read the original comment I pictured it quite vividly as being uttered in a heavy Southern accent by a rather younger woman in an elaborate dress and hairdo.

(Aside: interesting too how the meaning of "cute" has shifted so that I can use it in the above sense of "adorably dumb/helpless" and expect to be understood. It is an abbreviation of "acute", as in angles, and was originally used to mean "sharp" as in "smart, quick, alert" etc.)

377:

The 'bunch of mathematics' is ancient; it was described as a consequence of some the quantum mechanics I was taught in the 1960s. Acausal mathematics has also been studied in both statistics (earlier than QM) and parallel computing (later), and the potential advantages are well-known in those fields. Physicists are notoriously both reluctant to and bad at thinking in such ways. But all I really took away is that they are looking at this aspect very seriously, which is good. The descriptions of actual experiments (e.g. to implement Maxwell's demon) are all still theoretical, as far as I could see, but the blurb wasn't precise enough to be sure.

378:

When I was young, it was normally in the form "bless their cotton socks". Don't ask me why.

379:

Yes. And by contrast, there is also the explicitly vituperative use of the hosiery reference in the expression "rot your socks".

380:

The Maxwell's demon part seems to be theoretical, but there are real experiments where they've entangled pairs of photons A and B such that both A causes B and B causes A, and the outcome of the experiment depends on the lack of definite causality.

The part that interested me is that someone's developed a formalism that allows apparent causality to emerge from a fundamentally noncausal reality, analogous to the way Newtonian physics is an emergent form of relativistic physics that appears at some scales. This may turn out to be a very useful part of this whole exercise, for it helps get past the objection that "obviously, cause follows effect in the real world, therefore it must be a fundamental part of reality." The human real world appears largely Newtonian, but few dispute that relativity is an essential part of reality.

381:

My ISP suddenly became more unreliable than usual, and the IP address was changing frequently as a result. Then the IP changes stopped, but as near as I can tell, the unreliability and router re-connection requirement did not. My theory is the ISP couldn't fix the reliability problem(s) and to stop the complaints, made the IP addresses fixed, or mostly so.

382:

My wife, who is Yorkshire through and through, never heard this story, but finds it entirely plausable.

383:

Problem
If you can have TWO of: Causality / FTL / Relativity
BUT
Causality is only "apparent" at macro scales, like Newtonian physics ...
Then is FTL possible for "macro" objects, like starships & people?

I feel there is a v interesting SF story or three in there

384:

The thing about "insulting cops" is that it means a brief period of creative insults before the red queen race gets underway.

"you exemplify the public view of policing" ... racist murderer with qualified immunity or thin blue line?

"no finer police officer exists in your department" ... someone that shit at it is the best?

Plus the whole list of backhanded compliments "you have your shoes on the right feet" and "isn't that a well-polished badge".

I assume police are exempt from this law, just as they're exempt from "bringing police into disrepute" and so on, because otherwise a lot of "banter" is going to be one per customer. Can you imagine a police locker-room where insults result in fines or loss of employment? "good morning sarge" ... "Do I look like someone who is having a good day? I've had enough of your insults"

385:

Yup.

I should point out that there's still ample room for disappointment. For example, if the top speed of a warp drive is equivalent to 10C, then basically you have a ship that can reach Alpha Centauri or Proxima Centauri in about six months. If the most habitable planet around any of these three stars is about as habitable as Mars is, you've got the same problem that NASA has putting humans on Mars. I'd suggest that, from the pandemic year, that locking people in a spaceship for more than a year (coming and going) is on the outer limit of endurable, even if the food holds out.

That's another solution for the Fermi Paradox, which is simply that good places to land are, on average, further apart that starships are willing and able to fly. In our little part of the Milky Way, it would be better to be able to go 100*C or faster, just to get to the more sun-like stars with apparently more Earth-like planets orbiting them. That's not my favored explanation, but it's certainly possible. Space is very big, and we've got no good evidence of ETs. There are any number of likely reasons for that, and few of them are very inspiring.

386:

Heteromeles @385 : "I'd suggest that, from the pandemic year, that locking people in a spaceship for more than a year (coming and going) is on the outer limit of endurable, even if the food holds out."

Not everyone is an extrovert.

I found these last 12 months very soothing, working from home in my little bungalow. In fact the time went by very fast. Note that I have no personal interest in current space exploration, given the crude level of technologies such as spacesuits.

But all you have to do is make a good pick of introverts for the crew. Whaling expeditions used to last one to two years, starting from New England and going to Greenland and later on to the Pacific Ocean.

https://www.youtube.com/watch?v=DlHhV8ImdNg

387:

Apropos of nothing, you just reminded me of a ridiculous story idea I was unable to persuade to work years ago.

Premise was essentially: ftl practical, ftl implies ctcs, causality violation disallowed due to censorship hypothesis.

You can send someone away faster than light but an expanding event horizon forms that prevents any information coming back that isn't in your past light cone. Throw in an interstellar gate network. Hilarity ensues.

The plot that I was unable to make sensible involved routing of messages to create censored regions in which shenanigans could occur...

Anyway, the answer to the fermi paradox is that aliens are everywhere but since they got ftl we can only see versions of the universe that doesn't have them :)

388:

Forgot to mention, it started with someone sitting in arrivals for 6 months as they waited for causality to sort itself out so they could interact with passport control.

389:

That's a great story just by itself.

390:

Thanks, but it gets more unwieldy the more you think about it.

Thinking about what can be inferred from information that is unavailable, and how to prevent such inferences in such a universe is the path of madness.

391:

Heteromeles @385 :

Agree with Niala. My wife and I love the isolation, and really hope my employer (NOAA) will make working from home a permanent thing. Admittedly I miss out annual midwinter trip some place tropical, but what I miss is warm sea. Not people. If I could take such a trip without interacting with anyone except online, it would be great.

392:

Re: introverts....not really. I'm very much like you, but that's not what I'm talking about.

The closer approximation is more likely being stuck on a nuclear submarine for six months. No windows, no time outside, minimal personal space, constant racket, and 100-200 other people in there with you. Then you land in a weird and bizarrely dangerous place, do stuff there, hopefully survive, and pack the survivors back in the tube for the same trip home.

You can read about it online, for instance https://www.thedrive.com/the-war-zone/34104/nukes-nubs-and-coners-the-unique-social-hierarchy-aboard-a-nuclear-submarine

I'm not particularly interested in making this space suxxx volume 33. If spacefaring or star travel are possible, they're going to be unlike what most of us experience. We're used to the pleasures of reading alone some place comfortable, and subconsciously think that star travel will be like that. If you start learning about life on a submarine or the ISS, it really, really isn't like that at all. But people do live that way, so it's possible.

393:

being stuck on a nuclear submarine for six months

There's probably more open literature about science missions to places like Antarctica. The requirements for the human components are complex and very much not for people who like solitude.

Think of the pics you see of the ISS... it's less about floating 30Mm from anyone else and more about thinking very clearly and methodically through a complex task while someone in the same phone boot exercises at 80%VO2max, then crawling out to sleep in a 50cm diameter tube wedged between an ashmatic vacuum cleaner, the machine that goes "bing" and a very bad steeldrum band. Earplugs are just the start of it...

394:

qualified immunity

In a very interesting reading of the tea leaves QI may be going away. At least partially and/or slowly.

A recent dissent to rule from SCOTUS implies the standard may change. It's a very deep reading of the tea leaves but also significant.

We'll see.

395:

But all you have to do is make a good pick of introverts for the crew. Whaling expeditions used to last one to two years,

But deep in everyone on the ship's mind they KNEW they could head for a port, shore, or island if needed. Once you head out in a star ship, you're committed. Unless we start building Star Trek/Starwars ships.

396:

My wife and I love the isolation,

My wife was force to retire. And my work means I got out several times a week to drop off/pick up a cable, computer, or whatnot. At times front porches hold a lot of valuable stuff.

Me: I'm headed out to make these stops.
She: When do I need to be ready? I must get out of the house.

397:

You can read about it online

To me this one is the ultimate "Can I live on a star ship?". Star Trek it isn't.

https://fas.org/man/dod-101/sys/ship/docs/simlife.htm

398:

I skimmed past an article somewhere that said the Dems have a bill to cut it right back.

the House passed the "George Floyd Justice in Policing Act," ... Enables individuals to recover damages in civil court when law enforcement officers violate their constitutional rights by eliminating qualified immunity for law enforcement.

I presume at least one Democrat senator will vote to kill it, but you never know.

https://www.techdirt.com/articles/20210308/14502246383/police-police-supporters-ending-qualified-immunity-makes-being-cop-too-hard-somehow-defunds-police.shtml

399:

"Insulting the cops"
Is going to get a lot worse, over here, after yesterday's disgraceful scenes.
See also: this image - taken about 400m from my home ... Wrong colour scheme, of course, it should have been Purple/White/Green, but good for Stella!
At the SAME TIME Patel's proposals to restruct peaceful protest even more are being put to the Commons v soon.
Not sure you could make this shit up ....

Moz
Will that apply to the other well-known US police theft protocol .. ummm ... "Civil Forfeiture" ??

400:

I presume at least one Democrat senator will vote to kill it, but you never know.

Takes 60 to pass such things. So one Dem either way means the difference between 10 or 11 short.

Senate rules are arcane and complicated. A simple majority (on major legislation) only can happen during budget reconciliation which only happens down a very specific set of rules about budget items. Which is why the Senate parliamentarian said no to the minimum wage part of the last big bill. It didn't fit the rules of reconciliation. And while Schumer could have overridden the parliamentarian there would be a political cost so he decided to not use that nuclear hand grenade at that time.

Which all leads to a parade of House bills trying to force Senators into bad votes for election ads for 2022. Pelosi and others are trying to keep this parade focused and under control so they wind up with a nice set of talking points for the election. When the R's ruled the House they did so many such bills it was hard for them to be seen as serious. And the number made it easy for both the D at the time and R for a while Senate leadership to ignore them as such things tie up precious floor/debate time to bring a bill up.

401:

Re: 'Star Trek it isn't.'

Quite a few places esp. open concept offices, refugee tent cities, pre-COVID major downtown city streets and subways are noisy. After a while you just zone/tune out all but the few pertinent noises. Ditto for interpersonal interactions. My guess is that this reaction is why large city inhabitants tend toward rudeness/callousness: their brains auto-delete from consideration anyone and any activity they don't already know/expect.

I moved from a large to a small city almost two years ago. Public interactions here tend toward more positive types of sociability: people treat everyone like 'people' with much less BS too. Hadn't expected this for some reason, just thought that apart from size/number of interactions, things would play more or less the same since the incidence of idiots, yahoos, sociopaths, etc. is comparable. Maybe there's an optimal city size.

402:

Some years ago I was a referee for a friend who applied to the British Antarctic Survey. They sent me a several page questionaire to fill in which could be summarised as "Would you spend six months in a shipping container with this man". I must have said the right things because they sent him to Antarctica for 2 1/2 years.

403:

I wonder if "After a while you just zone/tune out all but the few pertinent noises. Ditto for interpersonal interactions. " is another introvert/extrovet difference. I find that very difficult, and especially people - if I am in a busy open plan office I can't maintain a train of thought. Possibly that's also why I don't like socialising in large groups, because I can't filter to the conversation I want to be part of? I have found noise cancelling headphones a great boon, and am loving working from home and not having to deal with colleagues except when there is a work reason to.

I've certainly noticed that virtual gatherings tend to be dominated by the loud extroverts I would usually avoid, and those of us who would be having a quiet side conversation if we were in a pub don't get to. I would not cope well with being somewhere I can't get away on my own or with a small group, being forced into larger groups all the time would be hell for me.

404:

I am not surprised - if I recall correctly, that is an expected consequence of most models of quantum mechanics as regards quantum entanglement and the wave/particle duality (at least one of which I assume is used in the experiments). If they have demonstrated it conclusively, it does exclude the few causal models, but they were generally disbelieved even in the 1960s. My point is that this confirms existing belief - it doesn't teach us anything new. If that blurb were a little less vague, it might convince me otherwise.

The interesting questions are the old, old ones (a) can they pass information and hence create a causal loop and (b) are there any time/distance constraints on such dual observations? To the best of my knowledge, the answers are "probably not" and "that's a matter of pistols at dawn".

405:

Absolutely right. People like me, who can handle solitude indefinitely, would go bananas in short order under such circumstances - yes, I remember my schooldays :-( Most people who can handle such crowding can't handle solitude.

406:

Yes. I posted some constraints a while back to avoid causality violations. One of them was, if I recall correctly, if the gates are a distance D apart and their relative speed in that direction is V, the minimum transmission time is D*V/c^2. Of that form, anyway.

Despite the dogma, you CAN have relativity, causality and FTL - just not unconstrained FTL. One of the main reasons I think that it is such a disgrace that the quantum transmission time issue is being swept under the carpet is that any answer to it would provide an immense amount of information on the relationship between the two (including whether such an ansible/gate is forbidden). Oh, yes, there are a zillion claimed answers, but no conclusive experiments.

407:

Well, I was responding to your line: I'd suggest that, from the pandemic year, that locking people in a spaceship for more than a year (coming and going) is on the outer limit of endurable, even if the food holds out.

However I agree that the pandemic situation is not a meaningful approximation of a spacecraft. Moreover, I had known for a long time that I am not a good candidate for a submarine crew. I am a good candidate for something like lighthouse duty, where you are alone for weeks or months. But spaceships are more like submarines than lighthouses.

David L @ 396:

Me: I'm headed out to make these stops. She: When do I need to be ready? I must get out of the house.

LOL! Never happens at my house. More like "How can I optimize the errands so the next run is in two weeks instead of in one week?"

408:

SFReader @ 401 : "My guess is that this reaction is why large city inhabitants tend toward rudeness/callousness"

It doesn't work for written intercourse though.

Last year I had to deal with a good number of people in New York City, by email, and they were all, with a not a single exception, overwhelmingly polite and obliging.

409:

Re: '... by email,'

This suggests some sort of pre-existing relationship/set of interpersonal rules that ping their brainware to act human vs. being just another head bobbing along in some vast anonymous human tide.

410:

SFReader @ 409: "This suggests some sort of pre-existing relationship/set of interpersonal rules that ping their brainware to act human..."

Fortunately we have among us some individuals which contantly remind us to act human:

https://www.deviantart.com/schmoedraws/art/190316-Finger-872114430

411:

I'm not particularly interested in making this space suxxx volume 33. If spacefaring or star travel are possible, they're going to be unlike what most of us experience.

From your mouth to Blog's ears. This place definitely needs a new set of strange attractors.

412:

I recall a SF story about a stay-at-home type who was secretly recruited by the Galactic Federation to be a starship explorer. His brother was a gregarious outgoing personality, training to be an astronaut in Earth's nascent space exploration program. The takeaway from the story was that the meek would inherit the Universe, leaving the extroverts at home because space wasn't for them.

People in space for long durations far from Earth and with no comforting escape capsule docked at the end of the ship might require choosing a very specific emotional type for crew. In extremis it might require selective breeding and gene engineering to make a version of humanity that can survive and prosper in such an alien environment.

413:

And then, there's "Way Station." by Clifford D. Simak. It won the Hugo award for best novel in 1964.

The hero of the story is a stay-at-home type who lives alone in a secluded farm house and secretly tends the machinery of the hidden way station of an interstellar transportation network. The network was built by the aliens who recruited him and never revealed their presence to the rest of humanity.

414:

Re: ' ... choosing a very specific emotional type for crew. In extremis it might require selective breeding and gene engineering'

Or empaths* - not hive-mind types but folk who can register and process others' feelings and have an in-built aversion to deliberately upsetting others. If you can avoid small emotional upsets very early, chances are you will also avoid having someone bottle up their feelings and having their feelings fester to the point where they explode. Also not the handwavey mind-reading version of empathy - just mindfulness and ability to read appropriate signals. Such a crew would also probably be more likely to suss out alien psychology (because they'd be alert for it) therefore establish better, more accurate communications.

As for the physical environment - could be a mix of physical engineering (design of the space ship to maximize perceived and actual usable space) as well as breeding smaller humans who would need less space and/or humans who need very close physical contact with others to feel comfortable.

* Yeah, empathetic does not automatically equate compassionate.

415:

First, breeding smaller humans? You mean, like women astronauts, who have been shown to deal with isolation better than men?

And then there's the point of... small starships? If we can build a starship, we can build big enough that the size of a human isn't a problem. You're still thinking right now technology... and NASA and others aren't worried about that, the way that, had I been qualified with degrees, I couldn't have tried for the astronaut program in the sixties, since the *maximum height* was 5'10" (1.78m).

416:

So far as living on a spaceship goes, I tend to look at what has worked on small islands, especially in Micronesia. There is some literature on this, mostly aimed at ex-pats and people working in the islands to try to help them adapt. Much of it is applicable to spaceships.

The one part I always get curious about is that both Polynesian and Micronesian societies are strongly birth-order ranked. The putatively highest ranked person is the eldest child of the eldest child going back however far. The lowest ranked theoretically are the youngest child of the youngest child going back however far, although after a certain point the people at the bottom often seem to stop keeping track because, really, why bother? And sometimes mom really doesn't remember or won't say who the father is. But these rankings go across generations, so for instance the eldest child of a second child doesn't rank as high as the younger child of the eldest child.

I've often wondered if this is a bug or feature. It doesn't obviously cut down on chiefly competition, since the younger brother who thinks he's better than his older brother and tries to overthrow or kill him is a stock figure in stories and histories. But on an everyday level, does having an elaborate ranking system help on average, hurt on average, or is it just one of those things that people do? These are people who live on small islands and sail tiny ships, it's not like they can get away from each other. Does having an elaborate pecking order help in this particular situation, or not?

The parallels with starships and off-planet colonies should be obvious.

417:

In a way it doesn't matter, because it's incompatible with even our weak implementations of merit ranking. And from what little I know, most Polynesian societies have a few ways out of the ranking order, mostly the "being really good at something" sort (and not just killing people).

418:

Also, I suspect anything that reduces status fights is going to help. Even the most modern and equalitarian nations tend to have rigid hierarchies in ships and other situations ("but I am your king!") For all that various militaries claim "a duty to disobey" they tend to be vigorously to the contrary in practice... and there are whole forests of examples of claimed social norms not applying in practice, often due to hierarchy effects.

Just because we've moved on from a formal "three female witnesses equal one male one" doesn't mean that we've also removed the various related factors that lead to 6% being an upper bound on number of rapes that lead to conviction.

419:

An alternative is the way astronaut corps in the US and Russia do it. There are basically two ranks of astronaut (trainee astronaut and astronaut). There are mission commanders and mission controllers but those are per-mission command positions, not permanent rank. They prefer to keep a flat hierarchy and rotate positions around and critique how each mission went.

The difference, I think, is that space flights so far tend to be closer to Earth, shorter in duration (up to a year, of course), and intensively monitored, while a starship that doesn't have an ansible equivalent is going to be temporarily or permanently out of communication. Probably having a stable hierarchy is a good thing in long, independent flights. Is it worth transitioning that hierarchy into things like land ownership? I'm honestly not sure, and that's what makes it a fun question.

I happen to agree that the Oceanic societies, like every other one, have all sorts of complexities, workarounds, and coping mechanisms. Birth order isn't everything in islander cultures, but it seems to be about as important as it is in the British aristocracy, and for much the same reason--it tracks ownership of resources.

420:

I think in the gap where people aren't living in each others armpits less rigidity is inevitable, and I suspect variety as much as flexibility is necessary, if only in the "can choose which ruler to live under" sense. Given a planet that's even explorable I can see half the ship wanting to move there just for the joy of not being subject to Captain Bullshit every second of the day. Then a month later wanting to come back :)

Longer term it would depend a huge amount on the size of the colony. Even 1000 people on Marsbase One are going to be looking at the Earthly aircraft carriers with considerable envy... so much space, so many people, and shore leave! Even Antarctica is better, and not just because you can go outside. But being able to swap between the ship and the ground might be really handy, almost as handy as being able to get new stuff into the ship (and possibly, dump old stuff).

Going off the documented smell of the ISS I can imagine a literal change of air after 5+ years in transit would make a world of difference. Just freshly distilled nitrogen and oxygen... imagine!

421:

Vernor Vinge has an amazingly prescient scene in his space opera A Fire Upon the Deep. Our heroes are fleeing the Blight, a sort of reified computational virus that can infect most things with even moderate computing power, including human and many other organic minds. They receive a message from a spacecraft that's following them, and the crew look normal, without the Borg–ified zombie look that's the sign of infection by the Blight. Yay! They have allies!

But then they think the video looks too good, so they check the data rate and it's far too low to support full (3d?) video. Uhoh. They turn off their local AI enhancement of the signal and it drops down to a blocky, jerky rendition of an obviously infected crew. No help there …

Anyway, I said astonishingly prescient because it essentially anticipates the problem we're having with AI image processing. If we were viewing things in confusing circumstances we would probably notice something; but its failure modes are different from ours and its apparent certainty leads us into thinking that it's perceiving things clearly and isn't subject to error.

422:

Given a planet that's even explorable I can see half the ship wanting to move there just for the joy of not being subject to Captain Bullshit every second of the day. Then a month later wanting to come back :)

I liked Cherryh's 'Foreigner' series in respect to this. (Mild spoilers.)

In the books, there was a human ship which got lost in hyperspace, and found a planet they could land and perhaps live on. Some did, some went away, and there is conflict when they get back. The "Not being subject to Captain Bullshit" vibe is strong there.

423:

Firstly, please do not confuse isolation and being in a confined group. The GROUP may be isolated but the MEMBERS are confined in that group. Secondly, these are matters of statistical distributions, not categories, so all sexist rules are bad ones.

Very few women handle isolation well, as doing so is associated with those of us on the Asperger's spectrum, and that is much rarer among women(*). Most women may well handle the confined group situation better than most men, but the evidence I have seen isn't entirely convincing.

(*) My wife goes doolally if she doesn't talk to anyone for two days (discovered when revising at university), and I like nothing so much as a week completely on my own and away from people.

424:
Takes 60 to pass such things. So one Dem either way means the difference between 10 or 11 short.

That's something I deeply don't understand about the US political system.

This 'it takes 60'-rule doesn't seem to have bothered the Republicans during the last four years, where they never had 60 senators or more (52 from 2016-18 and 53 from 18-20). But still McConnell could basically do whatever he wanted.

So why does it suddenly become relevant now? Is there some special provision in the US constitution saying that Democrats need 60 votes in the Senate do get anything done while Republicans only need 51 to dominate everything?

It's something I can't wrap my head around. Can anybody explain?

425:

AIUI getting legislation through Congress takes strategic horse-trading, arm-twisting and dollops of pork. Trump was not that kind of President.

Most Acts of Congress are pretty small beer: update a law here, rename a federal property there. These tend not to be controversial, and are usually passed to please some constituency or do something for Motherhood and Apple Pie. The elected representatives in both houses have no great difficulty in organising votes to get these passed, and as long as they don't touch a Culture War issue nobody in the Senate cares enough to filibuster them. So no problem.

However if the President wants to get some "signature" piece of legislation passed, like ObamaCare, that is another matter. The minority party know that any cooperation they provide is only going to help the majority party at the next election, so the incentive is to block everything possible.

AFAIK the only piece of legislation particularly associated with Donald Trump was the Tax Cuts and Jobs Act of 2017. This cleared the Senate by 51 votes to 49, but because it was a finance bill it went through via "reconciliation" (just like the recent Covid relief bill), so the filibuster wasn't an issue.

Apart from that everything Trump did was via executive order, which is why it was all so quickly reversed when Biden took over.

426:

Re: 'Even the most modern and equalitarian nations tend to have rigid hierarchies in ships and other situations ...'

Yeah - we need to get away from a one-size fits all approach to how we live esp. listen to and become more comfortable with recognizing and minding best-for leaders and strategies by situation.

COVID-19 has helped some of the public get comfortable with this, e.g., for medical advice re: pandemic, Fauci is the leader; for socioeconopolitical, it's Biden. And both Fauci and Biden publicly acknowledge that they pay attention to people who although organizationally more junior are more expert within specific fields. The one all-knowing-being/authority is not just obsolete, it can and does hurt us.

427:

Does having an elaborate pecking order help in this particular situation, or not?

If it helps, I suspect that would be because it's not a pecking order in the avian sense, in that your place on it is pretty fixed and so you can't fight/scheme your way to the top. (Outside a narrow range, like the chief's second son.)

428:

Re: 'First, breeding smaller humans? ... small starships?'

Yes - as part of physical and psycho-social resource planning and management. And we really don't need a 'breeding' program to get smaller sized humans. Size-management of the human body as documented in different cultures is as simple as: adjusting caloric intake by trimester; reducing broad spectrum antibiotic usage that kill off 'good' gut bacteria; optimizing the amount and mix of various types of carbs, proteins, fats, etc. Oh yeah - the non-extreme version of this approach* also happens to correlate pretty well with overall improvement in health and longevity. Go figure!

* Becuz there's always someone yelling 'what about the XXX where they starved people!! Again: I'm not into extreme either-or scenarios because they plain just don't work. Unfortunately finding the Goldilocks zone for a complex species is kinda tricky.

429:

This 'it takes 60'-rule doesn't seem to have bothered the Republicans during the last four years, where they never had 60 senators or more (52 from 2016-18 and 53 from 18-20). But still McConnell could basically do whatever he wanted.

Actually, as Paul indicated, he didn't. Outside of budget bills and judges, not much of substance got done. There were some epic failures of Senate R's to shut down ACA/ObamaCare and do anything with immigration reform.

At one point in 2017 or 2018, when many Senate R's were arguing to get rid of the 60 so THEY could get things through McConnell made a statement along the lines of:
"There's no where we need 60 where we have 50 at this time."

As to judges they sailed through with 50s because the D's changed the rule for them from 60 to 50 back when the R's were obstructing Obama's nominations.

As to understanding how all of this works refer back to my comment about arcane rules. The Constitution states that the House and Senate get to make their own rules. And they do it by simple majority. And most of these rules are in place to make the "should be easy" easy, and "should be hard", hard. And the general principle is rules carry over. Changing them is done with trepidation.

The House is starting to get fed up with nutcase Greene of Georgia. The conspiracy lady on steroids. Lately she's been using the ability of any member to call a roll call adjournment vote at any time. Which means everyone show up (Covid-19 restrictions mind you.) and vote in person. Basically it is a way to tie everyone's personal time calendar into a knot. She says she was elected to do such. Her fellow Rs are beginning to think of tying her up and tossing her in a closet. Doing such once is a some ways a principled stand. Doing it every few days is just being an ass.

430:

An FYI

Second Moderna shot noon Saturday. Got tired earlier than normal and went to bed around 8pm which is crazy early for me. I think between night time and naps I've slept 30 hours since then. This morning I feel mostly normal again.

Others I know who've had one of the 2 mRNA versions in the US report similar.

431:

I read that article about Clip. Those were spectacularly bad examples. If you take a Granny Smith and the network can identify it as an apple, and then slap a label on it that almost completely obscures the apple, you can hardly expect it to identify it. I'd argue that most people, not having seen the unobscured apple, would have focussed on the label, and called it something like "a post-it with the word IPOD on it" rather than "an apple with a Post-It on it". The poodle with the $ signs is a little different, but it's an image not seen in nature. If you asked me to describe it, I'd say it's a dog AND a person, with a lot of $ symbols all over it. If you asked me what it _signifies_, I'd have to give "Piggy Bank" fairly high marks.

I'm sure there are better examples, but those two were journalistic cherry (apple?) picking.

432:

The problem is that during an emergency or a war, a good decision now is much better than a great decision ten minutes from now. So, ideally, you do "exercises" about battles and emergencies, using experts as referees and report writers. You also cycle the people you've picked for commanders through the lessor slots and hopefully they pick up the right kinds of expertise.

433:

Niala @ 374: I visited the Boston Computer Museum in the 1990s. There was a full scale see-through model of the Antikythera mechanism. I was impressed!

I remember absolutely nothing else of my trip to the Boston Computer Museum.

On my last trip to the Smithsonian's American History Museum (sometime last century) they had a computer exhibit that included the first computer "bug". Had the log book on display

https://www.nationalgeographic.org/thisday/sep9/worlds-first-computer-bug/

434:

MSB @ 424:

Takes 60 to pass such things. So one Dem either way means the difference between 10 or 11 short.

That's something I deeply don't understand about the US political system.

This 'it takes 60'-rule doesn't seem to have bothered the Republicans during the last four years, where they never had 60 senators or more (52 from 2016-18 and 53 from 18-20). But still McConnell could basically do whatever he wanted.

So why does it suddenly become relevant now? Is there some special provision in the US constitution saying that Democrats need 60 votes in the Senate do get anything done while Republicans only need 51 to dominate everything?

It's something I can't wrap my head around. Can anybody explain?

Think "Heads I win, tails you lose!" Republicans have been able to break the rules (and laws) with impunity.


435:

JBS
Us, too
Here's our version:
Think "Heads I win, tails you lose!" RepublicansBozo's crew of semi-fascist lairs have been able to break the rules (and laws) with impunity. ... so far.
We will see how that plays out over the next few months, I think.

436:

The Sarah Everard murder, the police over-reaction in London & Brighton ( Everywhere else, the plod had the sense to stand back ) the arrogance & lying of Patel the "clamp down" ultra-right in the tories......
THIS is our BLM moment
It's going to be a rough ride.

438:

No. And I note that you don't seem to have read the rest of the post you're responding to.

If we can build a starship, we can build a big enough one. Even NASA hasn't required astronauts to be 5'10" or under since the what, 80's? 70's?

439:

Getting back to Musk the God Emperor of Mars Technoking of Tesla, there is an interesting line of thought in this article in The Register.

TL;DR Starlink (the satellite-based Internet service) has the potential to let Chinese citizens pole-vault over the Great Firewall. What will China do about this? What will Musk do about what China does? And what will India do if what China does interferes with India's access to the Internet?

Personally I suspect that the way India is going the Indian government is going to be really happy to see Starlink blocked in India too; they don't want their citizens having a free and unfiltered Internet either.

(Technoking: is that a verb? I technoke, you technonke, he technokes, we are technoking... )

440:

Update on my own # 435/436

REALLY chilling moment this morning ..
BBC_R4 were interviewing the woman photographed handcuffed & laid-down yesterday ... ( Patsy Stevenson )
Her comment was that the individual Plod, afterwards were perfectly polite & almost-apologised, which was actually the scary bit: "We are only following orders" - yes, really.
Pastor Niemöller's remarks about "not protesting" - until there was no-one left to protest for one's self come to mind, &/or the Soviet equivalent: "you today, me tomorrow"

In spite of authoritarian brute Patel's remarks, as I type this, Parliament Square is filling up ....

Thoughts?

441:

Here's a link to the relevant figure in the paper:
https://distill.pub/2021/multimodal-neurons/#figure-14
Unlabeled and labeled (with three different labels) Granny Smith, laptop computer, coffee mug, rotary dial telephone, plant pot.
(Somebody clipped an image of the upper left of the figure.)

442:

The constitution has no mention of the filibuster. But it does say the Vice President votes in the case of a tie. However, if a filibuster is allowed, the VP's tie breaking vote is meaningless--it will produce a majority vote but not one reaching the filibuster target. So this filibuster rule is anti-constitutional, rendering a key part of the constitution useless. The Senate should not be able to change the constitution this way without a proper amendment. Not that the current Supreme Court would have the courage or desire to so rule.

443:

The original idea of a filibuster was that a Senator (or possibly Congressperson) would stand up and start to speak, then refuse to yield the floor, sometimes for hours or days. The effect of this was to halt Senate business for the length of the filibuster. It was also possible for two or more Senators to divide up the filibustering duties; one senator would yield to the other senator, and so forth, resulting in filibusters that could conceivably continues for days, weeks, or even months. The point here was that if you really cared about something, you could filibuster it, but only by actually doing the work of talking incessantly about whatever came to mind for ten or twenty hours straight. (Southern Democrats were famous for their filibusters.)

In order to stop a filibuster someone had to have sixty votes.

Unfortunately, these days someone just says, "I'm going to filibuster that" and everyone sits around and asks themselves if they have sixty votes, and if they don't the threat is honored... If they'd make anyone who wants to threaten a filibuster stand up and bloviate for hours that would take the wind out of a good many sails! (And I think that letting Republicans get tired and start to free-associate would be wonderful thing.)

444:

Different usage of the word "isolation". In that paper the term appears to refer to moderate reductions in human contact. In some definitions it means a reduction in direct human contact to zero or near zero.
(Astronaut corps are generally pretty small and could select for whatever personality/mind types they want, including on a per-mission basis. There are enough Asperger's females to draw on.)
Also, re Zoom and other web conference services, many people are not very sensitive to lag jitters - if one is sensitive (I'm aware of jitters down to about 5-10 ms) then lag jitters favor extroverts re conversational turn-taking. (They are more willing to break the turn-taking conventions).

445:

From the OED: The action of isolating; the fact or condition of being isolated or standing alone; separation from other things or persons; solitariness. The paper whitroth referred to was about the ISS and, if there is one thing that the people on board are NOT, it is isolated. It is the environment that is isolated (in the very loose sense you describe).

As that paper said in "Behavioural health ..." and several places elsewhere, it was talking about living in isolated and confined environments, not isolated individuals. It was shoddy in "Relevance of clinical disorders ..." where it said "may make women more susceptible to the stress of isolation" when it was referring to the "the separation from family and other important relationships ", which is very different.

The ISS is completely different to sending an individual astronaut on a loop around Mars - or, indeed, an individual prospector disappearing into wilderness for 3 months. It is those people that are isolated.

Isolation within a community is another matter, and few people handle that at all well.

446:

I have had some experience spending large amounts of time in small spaces with other people. In the early 90s I crewed on a 31' boat across the Atlantic with 3 others (and a dog). Two of us handled it reasonably well, one spent most of his time asleep, and the captain went bugfucking insane interpersonally while still doing a creditable job of navigation.

The Age of Sail has plenty of examples of large numbers of mostly men crammed into tight spaces for years at a time. Rigidly enforced discipline and a shared and constant desire to not die probably helped a lot with keeping things together. Many of those men weren't there voluntarily either.

I am currently reading 'Children of Ruin' by Adrian Tchaikovsky. Their terraforming vessel was deliberately crewed with borderline obsessive scientist types who were happy to spend a decade focused on their hobby research while also doing the work of the ship. Some shenanigans ensue with long impact (i.e. if your hobby is to breed octopi for intelligence and sociability then release them into a water world, look what happens 3 millenia later). Nevertheless, the function of focused research as a way to keep many intelligent people occupied for long periods in tight quarters is a good angle. The use of 'cold sleep' for long voyages also helped - set up an experiment, go to sleep for a decade then spend your awake time reviewing the results...

447:

In spite of authoritarian brute Patel's remarks, as I type this, Parliament Square is filling up ....
Thoughts?

(Am American to be clear)
A lot of things out of the US 2020 protest playbook (with similar tactics and responses seen elsewhere as well). Even seeing "crisis actor" accusations. People in the UK appear to be alert to previously-used/historical tactics, which is great.
If it's played right, P. Patel's political futures will take some damage. I hope she is misreading the polity as more authoritarian than she thinks. (I.e. that she consumes and believes skewed right-wing information sources.) (In general the protest police response is not a good look for Tories, viewed from outside the UK.)
Re "Police, Crime, Sentencing and Courts Bill (but check for a newer version)"[1], language like this is worrying people:
(ab) in the case of a procession in England and Wales—
(i) the noise generated by persons taking part in the
procession may have a relevant impact on
persons in the vicinity of the procession, and
(ii) that impact may be significant, or”

The purpose of protest is to impact politics, politics is made of persons, and protest is often used by people without power (so they believe) to otherwise affect politics.

[1] This looks like a useful augmentation of UK media coverage: LIBERTY’S BRIEFING ON THE POLICE, CRIME, SENTENCING AND COURTS BILL FOR SECOND READING IN THE HOUSE OF COMMONS

448:

ever see the film 'virtual nightmare'?

449:

Re: 'Others I know who've had one of the 2 mRNA versions in the US report similar.'

Glad to hear you're feeling 'mostly normal'.

I've seen several sources reporting that the second shot can really knock you out especially on the second day post-jab. It's such a common reaction that some hospitals advised their staff to schedule their second shot for the day before their day(s) off.

450:

Now you're being ridiculous. Other than the first flights into space, show me *anyone* talking about sending one person on a multi-month flight, such as around Mars.

Not going to happen. They will *always* want more than one, in case of emergency.

451:

I know that. So. What.

I am pointing out that you are posting nonsense, because you are abusing the term "isolation" in regard to a person. The fact is that, statistically, men are better at handling isolation, whether or not women are better at handling confined environments that are isolated.

452:

Re: 'You're still thinking right now technology... and NASA and others aren't worried about that, ...'

Yes, I did read that post.

The size of the humans has impacts beyond the size of their container (spaceship, rooms) - specifically, air, water and nutrition. It's more a medical health vs. comfort concern - although both would be important if you're locked up inside a tin can for years.

453:

You're still assuming small ships. I'm assuming that by the time we build starships, we'll have solved a lot of the issues around recycling and life support, and that we will build large ships.

Let's see, I just looked up, and see that an Ohio-class boomer is about 171m long, and has a crew and officers of 155.

In my novel, set about 150 years from now, the research starship is in the neighborhood, or larger, than 1/3rd of a klick, and has a crew of 25 (including the captain) and a research team of about 70. Oh, and a friend who works on research vessels out of Wood's Hole says that's a larger team than she's ever been with.

And there are expandable bays, for all their equipment.

This is what I'm assuming. What are you assuming?

454:

The Age of Sail has plenty of examples of large numbers of mostly men crammed into tight spaces for years at a time. Rigidly enforced discipline and a shared and constant desire to not die probably helped a lot with keeping things together. Many of those men weren't there voluntarily either.

Sounds like you went through one of those standard disaster at sea survival setups, but fortunately the disaster didn't happen.

That said, there are some problems with using Age of Sail as a proxy for extended space travel. One is some critical bits of the life support system were outsourced to the surrounding ocean and Earth, and another is that, with some exceptions, the voyage between ports is on order of weeks to months, not months to years.

And this is if you assume that the spaceship will land at the end. If it's a vessel that's never intended to land, life support will have to last for the life of the ship.

Another point is that, to a first approximation, 90 percent of the biomass on the ship will be non-human. In other words, it's going to be a flying greenhouse* to a first approximation, not a wooden ship or a corporate skyscraper in space. Unless some sadist insists on using those pink growlights, the light needed to grow the plants, and the resulting green, will help maintain sanity in the crew. But we're still so programmed to think of spaceships as ultra lightweight machines that we don't think about what a long-term life support system will look like.

That's why I tend to look at small islands for analogs to how a starship might work, although a submarine also has certain parallels.

*Probably the plants will be watered and fed aeroponically (roots in humid air, not hydroponics or soil). This is basically a safety measure, since water spilling under variable or no gravity is just a versatile and expanding set of problems waiting to happen.

455:

Ongaku @ 442: The constitution has no mention of the filibuster. But it does say the Vice President votes in the case of a tie. However, if a filibuster is allowed, the VP's tie breaking vote is meaningless--it will produce a majority vote but not one reaching the filibuster target. So this filibuster rule is anti-constitutional, rendering a key part of the constitution useless. The Senate should not be able to change the constitution this way without a proper amendment. Not that the current Supreme Court would have the courage or desire to so rule.

The Constitution says in Article 1, Section 5, Paragraph 2:

Each House may determine the Rules of its Proceedings, punish its Members for disorderly Behaviour, and, with the Concurrence of two thirds, expel a Member.

That's where the Senate's 60 vote rule for Cloture comes from.

It's anti-democratic, but it IS in the Constitution.

456:

SFReader @ 449:

Re: 'Others I know who've had one of the 2 mRNA versions in the US report similar.'

Glad to hear you're feeling 'mostly normal'.

I've seen several sources reporting that the second shot can really knock you out especially on the second day post-jab. It's such a common reaction that some hospitals advised their staff to schedule their second shot for the day before their day(s) off.

Those strong reactions appear to be a "virtue" of youth. You're more likely to experience them if you're in your 20s or 30s (and maybe into your early 40s). Geezers like me (& you?) aren't likely to have as much reaction.

And even the "knock you on your ass reaction" is better than getting full-blown Covid. Would you rather be sick for a couple of days or sick for months & maybe die?

I'm 71 and the only reaction I had to either Pfizer shot was a bit of (quite minor) soreness at the injection site the day after.

457:

Geezers like me (& you?) aren't likely to have as much reaction.

Not so sure.

I'm 67. I had some all over muscle ache. But not enough to keep me from mental work or even light physical activity. But the "ok, go take a nap" feeling I had for 2 days was the big deal for me.

And in talking with some more 2nd shot folks tonight, reactions are all of the map.

458:

That's "The Claustrophile" by Theodore Sturgeon. Another essential feature of spacegoing people is the ability to be entertained by one's own thoughts.

****

My impression is that England has an unusually noxious tabloid press. Is this true? If it's true, any ideas about how it happened?

459:

I see the Tory Viles have finally achieved 1990 (https://en.m.wikipedia.org/wiki/1990_(TV_series))
Next up, Patel’s Public Control Department.

460:

Very few women handle isolation well, as doing so is associated with those of us on the Asperger's spectrum, and that is much rarer among women(*)

I am an Aspie. I am not entirely sure about my wife's neurological state, but she is definitely not neurotypical. Well, I do not care about human contact. She hates it. When she needs to figure out some device, she would rather faff it for hours online than to call customer support -- talking to people[1] is that painful to her.

[1] That is, people she does not know. She is fine communicating with me and with our friends.

461:

My impression is that England has an unusually noxious tabloid press. Is this true? If it's true, any ideas about how it happened?

The blessings of Sir Rupert de Faux, perhaps?

462:

The one part I always get curious about is that both Polynesian and Micronesian societies are strongly birth-order ranked.

Yeah. Right.

Mana depends very heavily on whakapapa. It's the source of mana. And it means something if a language has no way to just say "my brother", only ways to say "my older brother" and "my younger brother".

But Mana also depends on what you *accomplish*. Maui was the youngest brother.


463:

Maui was the youngest brother.

I heard the voices in my head start the Roman's sketch right there.

"what has Maui ever done for us, a dead fish and a sunken canoe"

But sadly they stopped there.

464:

It is fashionable to ascribe it to that cause, but it was just as bad a hundred years ago and more; it was already that way when his old man came over from Australia looking for shit to stir up. There seems to be something about British newspapers that they always had to be run by Lord something but it was always Lord Arsehairs or Lord Knobcheese and never one of the nice ones.

Also I think a lot of it is because back in the days when people had to read or else forgo entertainment, you could sell them any old cheap shit, and so it was sold in vast quantities, often in two or even three editions of the same paper at different times of day. And even so they were still dependent on advertising. So it's the same old story really.

465:

This interesting AI goof happened to me today: When I asked Wolfram Alpha percentage of US population over 65 it told me 14.5%, a perfectly reasonable answer. Then I asked it percentage of Nevada population over 65 and was surprised to be told 3.33 billion people. This appears to be, instead, the total population of Earth in the year 1965.

I can see what it's doing differently but it's striking that very similar inputs get parsed so differently. Given Nevada's population of about three million people we can expect the correct answer to be around 450,000; software using these answers as input would be unlikely to notice when an answer was three orders of magnitude greater than the total population...

Sometimes you're lucky to get as close as "iPod."

466:

Nancy L
England has an unusually noxious tabloid press.
All too true - but it always was poisonous & vicious ( See Dorothy Sayers' social criticism of this in the Wimsey detective novels ) ... but, of late, it has got viciously right-wing political as well, though the "Express" & the Hate-Mail always were that way inclined. Now the torygraph & the Times have joined in & the Scum has been spawned onto us.

timrowledge
Yes ... you know why, of course?
Couple this with the social turning point we have just seen ( See me @ # 435, 436 & 440 ) ... and.
The ultras inside the tories, the semi-fascists are now, they think, fully prepared for when the shit really hits the fan over Brexit, come about June/July, when the food-supplies & products from the EU run out & we get food riots.
Like I said, it's going to be rough & unpleasant.

467:

Greg Tingey @ 436: The Sarah Everard murder, the police over-reaction [...]

With reluctance, I have to disagree. The lockdown rules are there for good reason. The virus doesn't care about your politics, and a demo is a classic super-spreader event; lots of people in close proximity shouting at the tops of their voices.

Compare the coverage of these events with e.g. this one, which featured rather less sympathetic politics (at that time the UK was under restrictions but not the full lockdown).

Officers waded in to break up part of the crowd and seize the sound system just before 3pm and used batons against protesters, leaving some bleeding with visible head injuries.
The Met said 16 people were arrested “for a variety of offences”. Nine police officers were injured.

No close-ups of bleeding covidiots being sat on by police officers. Presumably they were less photogenic.

The police are supposed to be politically impartial. If they start choosing between good protests and bad protests then we really are in 1984.

So what exactly were the police supposed to do? Decide that good protests should be allowed while still stopping bad protests? Allow all protests and possibly see a resurgence of Covid (a decision which is rightly above Cressida Dicks pay grade)? Or wade in and start arresting people who are breaking the law?

Once you decide to start arresting people in a demo it is going to get ugly fast. There is no way to politely and gently put someone in a police van if they don't want to go and are surrounded by a crowd of sympathisers.

What would you have done if you were the police commander on the ground?

468:

Our libel laws have a lot to do with it, too. It's relatively expensive to pursue an action for libel, and damages have always been relatively limited unless the libel is really bad (for public figures, the "actual malice" test makes it very hard to sue a newspaper for libel).

Even if the libel is really bad, damages tend to be in the forms that a deep-pocketed newspaper owner can afford; money and a written apology, but generally no need to use the newspaper's platform to broadcast that apology. Even when an apology had to be put in the paper, it'd be extremely rare to require it to be as prominent as the original libel, let alone more so.

Additionally, we used to permit newspapers to pay into court and produce a form apology in writing for a libel, and thus prevent a trial. This allowed the more, shall we say, salacious newspapers to run close to the winds, knowing that the penalty for overstepping the mark was negligible (no trial for competitors to report on, and the defamed party would have to shop around for a competitor to publish the apology and explanation).

This all adds up to a recipe for toxic activity - just know where the edges of legal are, and pay out a small amount on occasion when you step over an edge. The risks are easy to bound, so why not take them?

469:

A kindly reminder about Nuremberg trials and gallows?

470:

It did say that it was counting females with the name Nevada. Adding a “state of” before Nevada gave a more plausible answer.

471:

Coincidentally one particular journalist/columnist has been made to eat crow today for statements she made (not AFAIK in a paper - I think they were on social media, given that an abject apology has been Tweeted today).

472:

See, for example, Lord Copper in Waugh's Scoop.

473:
but, of late, it has got viciously right-wing political as well

"of late"? Supporting Mosley and Hitler wasn't right wing enough for you?

474:

I visited the Boston Computer Museum in the 1990s. There was a full scale see-through model of the Antikythera mechanism. I was impressed!

On a related note, I just bought a new computer. It's the fourth(?) generation of a series I last tried out (and couldn't really use because it was too big and cumbersome) back in 2018.

It has an ARM 6 CPU and 128Gb of storage. And it weighs 27.5 grams (roughly one ounce) because it's a pen. (Livescribe Symphony.) Basically it remembers anything you write or draw on its special paper -- on non-special pattered paper it just works as a regular pen -- and can synch it to a smartphone or desktop computer app with handwriting recognition over bluetooth. (You're not tied to buying notebooks from the company store: you can print your own special paper on a laser or inkjet. PDFs of the dot pattern its IR camera uses for recognition and registration are available.)

Anyway: my watch has a cpu. My ear buds each have a cpu. My goddamn pen has a cpu. My phone has multiple 64 bit cores and a baseband processor (and can kick sand in the face of a late 1990s supercomputer.) Hell, my credit card and debit card are both EMV cards, which means they contain a secure microcontroller (typically based on an 8/16-bit RISC architecture).

Upshot: these days I don't go to the corner shop without carrying/wearing 7-8 computers.

475:

Thoughts?

My actual thoughts are unprintable, but given that the gender balance in comments on this blog seem to be about 95% male, I'm not going to hork up a trollball of a blog essay on this topic any time soon.

Let's just say it's less of a BLM moment for the UK and more of a #MeToo moment -- and England is getting a failing grade from its most senior female politicians.

For context: consider the monstering Nicola Sturgeon got from the Tories in Holyrood while testifying over the Salmond sexual harrassment prosecution. And it was a monstering: she was interrogated on the record, in public, for 8-9 hours, about her response to her male predecessor's alleged offenses. I know you don't like the SNP, but strip it of the partisan aspect and let it sink in: female executive is hounded publicly, with front page newspaper coverage of calls for her resignation, over something a privileged make predecessor did wring.

The UK's gender politics are utterly toxic -- even before we talk about the current wave of (US religious right inspired) transphobia, which is always a wedge issue for homophobic backlash in general (the trans community are seen as relatively unpopular and weak, hence an easy initial target).

476:

Also, re Zoom and other web conference services, many people are not very sensitive to lag jitters - if one is sensitive (I'm aware of jitters down to about 5-10 ms) then lag jitters favor extroverts re conversational turn-taking. (They are more willing to break the turn-taking conventions).

It also favours aspergers/ASD types, who don't pick up on social conversational cues and so tend to talk over people anyway.