Back to: Brief announcement | Forward to: Why I'm not on Google Plus

What am I missing?

Two weeks ago, at USENIX Security, I banged on a whole lot about the implications of cheap bandwidth and cheap data storage, by way of lifelogging using devices descended from today's smartphones.

But I am currently thinking that I over-narrowed my focus.

Here's the thing: let us postulate that by 2021, we will have hit the buffers using current microlithography techniques on CMOS -- say at a resolution of 5nm (compared to today's 22nm process). (Below 10nm our integrated circuits experience interesting quantum effects, not necessarily in a good way, due to electron tunnelling.) At this point we're well into the realm of nanolithography. Today's Intel Westermere Xeon server cpu has on the order of 5 million transistors per square millimetre (on a 512mm2 die) using a 32nm production process; my BOTE calculation suggests 80 million transistors per mm2 is likely by the time we get to 5-6nm resolution, giving full-sized chips with up to 40 billion transistors.

What applications are going to hit mass consumer adoption in the wake of us reaching a point where a first-rank CPU of some 40 billion transistors (equal to, say, 16 x 10 core i7's) cost US $250, and low power CPUs (an n'th generation ARM descendant with, say, 2.6 billion transistors -- a thousand times the component count of today's Cortex A-9 ARM architecture) can deliver the clout of a 10 core i7 on a TDP of around 10mW for a component cost of around $1-2?

Years ago, a couple of eminent computer scientists (if I remember the story correctly one of them was Danny Hillis; I forget who the other way) were discussing trends in chip production around 1980, and one of them objected to the other's extrapolation with, "but there's no market for such cheap chips! What are you going to do, embed them in door handles?" And five years later, checking into a hotel, he suddenly realized that he was using a magstripe card to open his hotel room door because there was indeed a microprocessor in the door handle.

But it doesn't take much in the way of embedded logic to operate a magstripe reader and a deadbolt. So what are the doorhandle applications that become practical when low-cost embedded devices are as powerful as today's high end servers?

One trivial possibility is widespread adoption of biometric authentication based on mixed parameters that take quite a lot of processing: for example, that hypothetical hotel room door might open for you by recognizing your facial bone structure and gait pattern as you approach. Again, your car won't have a key; it will "simply" recognize you, both by your face and your voice and more subtle cues such as your pressure distribution as you sit in the driver's seat.

But that's a gimmick. By which I mean yes, it's convenient, but it's not a game-changer: we already have ways of achieving these objectives (hotel room keys — or magstripe cards — and car keys with immobilizer chips). It doesn't fundamentally change the way we live the way that, say, mobile phones or lifeloggers would bring about basic behavioural changes.

What are the consequences of powerful microprocessors getting really ridiculously cheap — applications that just aren't practical today? Things like the library digitizer from Vernor Vinge's "Rainbows End" (which I shall not describe, because it's both a spoiler for the book and a thing of horror to bibliophiles), or infinite focal depth cameras, or giving your lifelogger real-time ubiquitous text recognition (as in, everything textual in your field of vision is scanned, digitized, and indexed immediately). What am I missing that isn't possible today and doesn't substitute for an existing process or technique? Alternative formation: if spimes are artefacts which are the physical instantiation of an entity with a trackable history on the internet, what happens when spimes acquire enough on-board processing power to act as the container for their own virtual existence?



Really small robots, with the sensory and motor control capability of current experimental robot platforms with (often) separate stationary processing. Though any advances in robotics depends on improvements in motion systems - better motors, artificial muscles, battery technology - as much as on processing capability.


Food tracking - where did that orange really come from? Was it sprayed with something that will give you a third nipple in ten years time? Was that chicken really humanely raised and slaughtered? Just how many food miles did that avocado clock up on its journey from the Occupied Territories?

Ad serving in newspapers, a bit like Negroponte's 'The Daily Me', but dynamic. It monitors your interaction with the paper (perhaps even using eye tracking) - what are you spending time looking at? Is it the financial news in which case have you thought about investing your money in Bolivian llama farms; is it the possibly underage girl on page three - in which case you're going to love our .xxx domains, or are you stuck on the crossword - buy our dictionaries or even treat yourself to the English degree you promised yourself all those years ago.


Realtime language translation, done rather well. It can already be done acceptably on Tablets, but leaving a lot to be desired.

Similarly, texting will go the way of the dinosaurs and be sneered at by all teenagers living in the year 2020. replaced by either accurate speech to text, or via subvocalization pickups on the skin of the neck

Iris scan as a logon for all portable machines

A cheap, small and decent headup display/monocle to replace screens on phones/tablets/laptops (long overdue IMHO)

Accurate eye tracking to replace the mouse, possibly coupled with EEG pickups (doubtful)

BTW, Moore's Law will not end in 2020. The number of transistors will continue to increase as wafer scale electronics is finally needed, and later printing onto graphene sheets. How much could you get on 10m^2 graphene sheet?

Also, the cost is going to drop drastically as lithographic tools do not have to constantly push for ever smaller features


BTW, a spime already has a name - a soul


Of course, to me the more interesting question is energy and resource-cheap and non-toxic processors, not just tiny ones. That adds more to ubiquity than tiny processors.

One of the really interesting ideas isn't just processing power, it's etching nanolabs onto chips. There are a lot of interesting uses if you can quickly determine the chemistry of a small sample.

One example: landfill and sewage recycling. Currently landfills are such a heterogeneous mix of garbage that they're hell to recycle. Even the gas coming off a landfill is such a mix that it's generally cheaper to vent it (if it's not too toxic) than it is to get the methane out for fuel. Ditto with sewage--all it takes is one jackass pitching his car battery (or any other toxic waste) into a sewer to make all the sewage unfit as fertilizer.

Cheap, fast, smart chemical analysis could change the cost equation substantially, simply because they could sort and filter things on a small enough scale that their output might actually be recyclable. Think of these as a form of Maxwell's Demon in the recycling stream.


In a move to gratify the (possibly spinning in their graves) soul of old-time conspiracy theorists, the phrase "smart money" takes on a whole new meaning. Whether you do your transactions electronically or in cash, they're closely monitored.

Biometric identification of who handles the cash, maybe drug and explosive detection processes by on-chip ELISA or similar becomes the security forces' wet dream.

And slightly more off-piste - we finally have machines that can extract precious metals from water and similar. We find the price of gold falls (although not hugely necessarily) and if we're really unlucky we also find that one of the scrubbed metals is a vital sub-micro-nutrient that we've suddenly lost.


Vast swarms of tiny drones.


heteromeles@5: "Cheap, fast chemical analysis" == GATTACA. Identifying people by their DNA is, in the final analysis (pun intended) simply chemistry.


In a move to gratify the (possibly spinning in their graves) soul of old-time conspiracy theorists, the phrase "smart money" takes on a whole new meaning.

This gets really interesting, and it allows for some really weird stuff. For example, we currently have a certain cycle that money follows; created by a government, (or a loan) distributed to a bank, loaned to a business, paid to an employee, used to pay the rent, sent back to a bank, used to purchase some groceries, used by the grocer to purchase stock in IBM, etc. Obviously some of these paths are more efficient than others, by which I mean better for the economy, or better for the average human being...

So what if money gained and lost value depending on what it is being used for as a vehicle to encourage savings, encourage purchases, etc. For example, used to purchase a beer, that dollar might be worth 90 cents. Used to purchase a home, that dollar might be worth $1.10...

Smart money might also be able to tell if it was being used illegally, or if it was attached to something of questionable value, such as a mortage run through the MERs system.

The really interesting thing is who gets to define what are the "good" and "poor" uses of money. I imagine S&P being put in charge of programming our smart money, and the thought makes me shudder... but that's OK, because there's always some dodgy type hanging around the bar who will root your cash for a small fee.


Off subject: It's H.P. Lovecraft's birthday today.

Happy Cthulhu Day!

"May he eat you first!"


I was afraid you'd say that.

Take a look at this then extrapolate all the way down to large insect/small bird sized UAVs with a single-shot disposable shaped charge or projectile weapon and enough smarts to hunt down their own prey. Possibly with encrypted mesh networking so they can hunt in swarms -- like killer hornets.

It'll revolutionize warfare, and not in a pleasant way. (Want to clear insurgents out of an area? Sent in a cruise missile with a warhead consisting of a hive of 10,000 of these hornets, primed to hunt down and kill anyone with explosive traces on their hands. Of course, the first order response of your typical insurgent group will be to press-gang pre-teens and toddlers to carry their bomb-making equipment ... it goes downhill from there, fast.)


Look on the bright side - it could revolutionize agriculture. Hunter killer swarms designed to identify and rid crops of pests. No pesticide needed.


GATTACA is one outcome... Maybe. I think Charlie proposed some useful countermeasures in Halting State, and I think they're already being used by criminals. More to the point, microsatellite analysis (genetic ID) is notoriously sensitive to contamination, because much of the data is in numbers of repeats of short sequences, not just a straight sequence with mutations. Stuttering repeats can be hard to amplify well, for the same reason that stutters tend to identify people.

Also, if PCR is involved, it's not that fast (hours, not seconds). Certainly an advanced reader would need less amplified DNA to sequence, but in PCR, each cycle doubles the amount of DNA. Cutting off two cycles won't shorten the process all that much. As anaother example, current DNA chips take hours, because you have to attach the fluorescent molecules to your DNA. You can get titanic masses of data from such a process, but I'm not sure you can speed up the process substantially, at least without a different technique.

Additionally, it's already possible to spoof a DNA sample. Currently, crooks pick up random cigarette butts from somewhere (for example, at the local cops' hangout, a head shop, or a bus station) and leave them at the scene of the crime. In the future, I'm sure spoofing will get more sophisticated. For example, if you've got cheap, fast, DNA analysis, you may well have cheap, fast DNA synthesis, and a crook could use such a synthesizer to drown out his traces at a crime scene


@mike: I am really looking forward to a future where I can review the lifelog of my organically-raised chicken...


I've been waiting for these to hit retail since 2006/2007, so we probably don't need smaller/more powerful processors to do consumer-level HUD. What makes their non-appearance really irritating is Lumus isn't a vapourware company; they make military HUDs as a day-job.

16: CMOS-based DNA sequencer that doesn't need fluorescent tagging. On a related note, thanks for the link to the cheap Nature subscription a few weeks ago, Charlie. ",)


Combine lifelog style recording with plentiful processing power, you could build pretty accurate individual predictive models for each citizen.

This would let you know what they were going to buy, say and do in most situations.

You could also take natural language processing to a next step, be able to detect and eliminate cognitive bias, emotional appeals, flaws in reasoning in near real time. Be fun to run a politicians speech through something like that


I think, though I'm not certain, that the most recent (experimental) DNA sequencing methods don't use PCR to amplify things, but just analyze continual strings of DNA. Something about using electrical charge to feed it through a hole so small that it has to go through as a single strand. I would guess that this implies they unzip the strands, but whether first or during the process I don't know. I also don't know whether this makes it easier or harder to count the number of identical codons.

However, thinking that future sequencing will depend on the rapidly varying segments is probably wrong. Sequencing of a full chromosome is getting cheaper rapidly, so they'll probably just do a complete sequence of everything. This will come up with a REALLY unique id, as even identical twins tend to differ in small ways. (For that matter, even the cells within one body tend to have small differences...stumbles during the copying that result in either duping or deletion of a codon.)

Spoofing is a different problem. And it may be hard to beat the dust collected off a bus, or from the corridors of a hospital, or office. But consider taking a bit of a corpse and finely dicing it. (Really finely. You don't want this to be obvious.) If they start separating our by strength of signal, that could produce a much louder one than whatever you left by accident.


Computers that are not just ubiquitous, but imperceptible.

Microscopic processors like bacteria scattered everywhere, including all over your body. Powered by ambient rf fields, thermal gradients, photovoltaics, whatever; and forming a massive global network (as in Vernor Vinge's "A Deepness in the Sky"). Processors that need no power when not active, but that still preserve their state (like HP's memristor concepts.)

User interfaces so natural to use, that you don't need to think about them at all. Not just understanding natural language, but reading voice stress, muscular tension, heartrate and other subconcious physical cues.

Want to know what the weather forecast is? Just ask the question to the air. Or subvocalise. Or just look up at the sky. The response is whispered into your ear, or projected into your field of view, or a weather map appears as if tattooed onto the back of your hand.


For every real-time verbiage garbage collection, the politicians will buy real-time verbiage garbage generation.

Think "arms race."


Really good VR, to the point that "facetime" will be virtual (bandwidth will still be expensive, so you send a 3d model of your head and facial movements at the beginning of a conversation, and rather less when you talk). In fact, let's extend that to virtual vacations.

Fahrenheit 451 style individually-tailored TV programs. Let's extend that - on-demand modification of books, music, TV, movies, recipes to suit your taste and/or requirements.

Much better virtual modeling, and on-demand manufacturing, of clothes, shoes, etc.

Vast amounts of CPU overhead to implement the latest fad in DRM, copyright/usage tracking of anything copyrighted. You'll still pay $10 for an ebook because that's how much it'll cost the publishers to make sure nobody copies part of it.


Some years ago, Marvin Minsky collaborated in an SF novel about the rise of AI, and one of the applications he described was a robot that increased crop yields and obviated the need for pesticides by manually removing harmful insects from crops and leaving the harmless or beneficial insects behind.

He described a large robot with very fast-moving hands walking along the rows of plants visually inspecting them and clearing them of bugs as it went, and this strikes me as fairly silly. But what if every single plant has its own monitoring system?


I think definite user-interface improvements. We've been hovering on the edge of useful voice recognition for a while, for example, and I think we're going to go over the edge on that fairly soon. We're further off for handwriting, and that may not be useful; but if it is useful, we'll get there soon.

Maybe a whole different paradigm using goggles or overlay for real-world interactions.

Keyboards hanging on a lot -- they're easier on the voice or arms than lots of alternatives, for when you're doing dedicated work at a computer.

One of the game-changers is AI assistants, but who knows when? We have not the slightest clue how to get to there, all our AI ideas are either pure blue-sky and not very well confirmed, or else very specifically targeted, not general purpose.

But we may have computer-mediated human assistants sooner. Seems like the job market is narrowing in on the very best people. That leaves a lot of under-employed pretty darned good people. Time-share personal assistant? There are more and more rich people who might have a use for this.

Definitely severe arms races in facial recognition, tracking, and such. Laws and styles about dress hats and veils. Facial tattoos? Better, face paint (frequently changed). Also in spam-vs.-ham (more generally than in online forums).

Major uprising to shove spam back into the shadows?

Identifying botnet traffic and cutting it off?


The microprocessors-in-every-doorknob story was recounted by Hillis in his talk On game software development which used to be available on, but I can't find it there any more. The transcript of the relevant section below starts at 14:03 in that talk:

One of the very first talks I ever gave in public was at the Hilton hotel in New York City, and I gave a graph kind of like this. It was when microprocessors were first coming out in the '70s, and it was a graph of how many microprocessors there were going to be, and I gave what was at the time a very radical talk. I made the proposition that pretty soon there were going to be more microprocessors than people, and that literally got a laugh in the '70s. People thought that was a very funny idea, that you'd have more microprocessors than people. And in fact, at the end of it I kind of bombed in the talk because at the end of it I did a Q&A session, and somebody asked the question "what do you think somebody's going to do with all these microprocessors? I mean, it's not like you need a microprocessor in every doorknob", and I didn't have an answer for that. But, if you go back to that same hotel today, of course, there is indeed a microprocessor in every doorknob.

Actually, keeping agricultural robots macroscopic and availble for visual monitoring might not be a bad idea. Devices DO fail because the Demon Murphy never sleeps, and some of the failure modes might be fatal to humans. It would be easy enough for a failure in an ID system intended to take out, say, rats in a granary, to set a swarm of the things going after ANYTHING warm-blooded.

Hmmm. Zombies don't look like that bad a threat in comparison.


"Watson" on every machine, and on every search server


Objects keeping track of their own status, able to tell you not just when the milk has spoiled, but when the lightbulb is about to burn out, the sunglasses screws need tightening, or the pants pocket has developed a hole.

I'm also always interested in how new technology can make it possible for people to more fully indulge their paranoias. Charlie's killer-hornet notion made me think of how worried some people get about deer ticks spreading disease -- or even germs from coughing passersby. Could those hornets be used like Maxwell's demons, orbiting around you and preventing the Bad Actors from physically contacting you?


The problem with ideas like this is, they're always the (heh) germ of an arms race. I remember the prelude to Dean Ing's novel "Soft Targets" which went something like:

A hard target is one surrounded by guards, stout walls and barbed wire. A soft target is a hard target that has been plucked, quivering, from its shell.

Or as my wife started reminding me when the Crypto Wars started: "Safemaker, safebreaker."


The final demise of free range kids, as well-meaning but overbearing parents adopt the tools of the surveillance state to monitor their kids' every waking moment.


Easy space exploration and space telescopes. Build nanosats equipped with a block of aerogel (to survive the impact), a wireless transmitter (to talk to neighbors), a small versions of your desired sensors, a tiny solar panel, and a tiny ion engine. Then scatter a cloud of them in the general direction of your target stellar object. Extract useful measurements statistically; rather than getting results from single high-precision trials using expensive equipment, get results by comparing the results of a million trials carried out by a thousand nanosats.

Same approach works for telescopes (can anybody say "long baseline interferometry"?) and phased arrays. You could put a cloud of these up in orbit and it'd be able to do face/gait recognition on anybody not under a roof.


No more actors except in luddite productions, because real time generation of CGI actors and voices becomes substantially cheaper than hiring real people.

As part of this writers suddenly become much more valued (hah! you wish!)

Also the division between games and reality becomes a lot narrower when you can walk into a room and be presented with a 3D view processed so fast that it can adapt to every movement of your eyes.


Instant recognition/description of the nearby environment for blind people.


With that much processing power available? More like artificial vision, including infrared and ultraviolet if desired.

Same thing with hearing. I'd love to have TWO good ears, including the ability to turn them off with recognition of "emergency reactivation" in case a fire alarm goes off while I sleep.


Yep.... first thing that I thought of when I came upon this thread and scrolled down to your post.

BUT, not written, text, translation but a 'sit on shoulder' Parrot - pieces of Eight Style- that would take the piratical spoken source and TRANSLATE to semi lip syncopated from the speaker to Unfortunate Foreigner that is not of OUR nationality.

Lots of Game changes here for the language of new asylum seekers country that Requires that you Speak the language of the host country. After all we aren't prejudiced against people who need to use a Zimmer Frame are we?

So ..a person using a translation Parrot - or Shoulder Dragon as in the Sci fi Con costume fans costume of choice ? - would needs must be a handicapped person, eh?

Charlie hasn't thought of this one ... the tourist trotting about Edinburgh with a Translator - Strange World tm Guide - sitting on his/her shoulder.

Out Gracious Host has thought of VR Glasses of course- for a New view of the City of Edinburgh - but he hasn't thought of a Parrot. .... ' Avast there Me Hearties! Pieces of Ten! Pieces of Ten! ... ' first on the left to ..sponsored Bookshop if its a 'Free ' hired Parrot.


"...extrapolate all the way down to large insect/small bird sized UAVs with a single-shot disposable shaped charge or projectile weapon and enough smarts to hunt down their own prey...."

And like the natural world, swarms of predators can be used to attack these machines. An arms race, to be sure.


If you think nano-drones are bad, here's another:

For some time both SF and thriller writers have used DNA targeted poisons as tools for killing. I don't think they exist (yet!). But, imagine if you could identify one or more genetic markers indicative of ethnicity. You could accomplish genocide of an order not previously envisioned.

5nM "robots" could be absorbed through almost opening in the body. Aerosol anyone?


Actually I don't think biometrics will ever take off.

Firstly the processing power is well within the realm of mobile phone/netbook computer power already. No laptops are around that can be unlocked by your face. Things like highend cars could already afford to implement that kind of thing today. It might add a few hundered dollars to the cost of a $30k car. I think hotels would easily be able to fork over more money for higher end biometric doorlocks today if they wanted to (remember those magnetic ones would have been expensive once).

The problem is that a camera can't tell the difference between a person and a photograph of that person. Even if you add some kind of high resolution Kinect style depth mapping to it, the just requires a 3D printed 'bust' rather than a photograph. They would have to add some even more advanced sensory stuff to it (maybe an infrared laser to check if the skin performs the correct subsurface scattering that plastic wouldn't, maybe using laser to listen for a heartbeat), all of that can be worked around and your fancy biometric lock will always be less secure than dongle with some kind of publickey cryptography built in (which might work via RF). It also drops the issue of having to authenticate someone since you can just give them the 'key', but if it is stolen you can disable that key.

We could end up with flying automated drones infesting cities the same way we have pigeons. Think of a totally automated, solar/ambient-powered quadrotor drone for a few hundered bucks (or less). Capable of running a full robotics platform like ROS on the drone itself with all the sensory, automatic mapping and so on. They could be programmed to just fly around until there battery runs low, find somewhere out of the way to land (such as a building roof), recharge for 2 days then proceed with another 40min of flight (or more if battery tech picks up). Add in some kind of wifi, or enough intelligence to return home after a month of flying around. How would the .gov stop people releasing them, they might not be able to track the signal if you relay via some other drones and the effort required would be to much. Chances are they would have to be outlawed like laser pointers.

Distributed drone networks. Lets say I buy a drone and sign it up to an open community run cloud service that takes control of it's operation when I'm not using it. One that lets someone put a drone basically anywhere in a major city within a matter of minutes. With multiple drones we could get a realtime, 3D Google Earth. Stick a TB flash drive in there and you have RFC1149, or mobile flying data drops. Just jump online and request that the drone with that library of 4K cinema HD rips flys overhead. People can stick open drone recharge stations on their roofs so the drones don't have to use solar, the stations themselfs could store all the data from the drones and upload new files that people have requested like some kind of real life physical p2p. Not to mention the possibilities for people selling things like drugs. How do you deal with people trying to fly one into your bathroom. Maybe we need drones designed to take out other drones, or laser systems to shot them down like that one they made for mosquitoes.


"What are the consequences of powerful microprocessors getting really ridiculously cheap?"

First thing that popped into my head was 'animism.' Both as a 'thing-in-real-life' and as a belief structure. Think about how the majority of people interact with their computers today and project that out into a world where everything is a computer, connected to a ubiquitous network. It'd be charms against evil magic instead of a firewall inside two generations.


Another idea. We could see the ultimate form of DRM.

Rather than a movie selling on a disc it could contain an entire player that streams to your TV.

A video game could ship as an entire game console. ebook's could sell like physical books (or maybe as chips that implement the entire ebook->pixels rendering and plug into an epaper display). That way pirates would have to physically open the devices (which would be nothing but a solid block of epoxy surrounding chips and some self destruct systems) and then they would be dealing with some kind of custom hardware (look how long a PS3 jailbreak took, we could see each game publisher shipping their own custom cpu architectural automatically generated and optimized for that specific game). Although movies could still be ripped even if it was via frame capturing.

Of course it would stop any kind of online distribution of the products, but if the companies selling the stuff can totally stop piracy that way we might see them make a push, simply choosing a few highly popular games to sell only inside their custom game system as a special feature for people who fell the need to own a physical object and if that takes off pushing everything that way.


No holds barred ubiquitous law enforcement for both state and commercial interests.

If it becomes commercially viable to embed substantial computing power and communications into nearly every product, it will be only a matter of time until product manufacturers enter into agreements (under coercion or otherwise) with other entities who will want to use that computing power to monitor the public or use them as a revenue stream.

The state security uses of hardware that can monitor populations 24x7 for any signs of descent are mostly obvious. Walls with eyes can identify vandals. Smart doorknobs with speech recognition can report state subjects for subversive or treasonous activity, such as planning to vote for an opposition party or failing to show proper respect for the police. Less obvious consequences include politically motivated functionality degradation. Recording devices could refuse to capture police brutality or record audio of corruption. Keyboards--and even pens--could be designed not to record politically questionable manifestos or allow disloyal subjects to communicate readily.

The commercial uses for ubiquitous monitoring are less well explored but considerably more wide ranging. A constant of the modern economy is that businesses from all sectors will, if given the opportunity, conspire against the consumer. Consumer products manufacturers who find themselves with more processing power than they need in their products will be quite happy to sell that excess power to other commercial interests. For example, any device that processes ambient audio and has surplus processing power could be sold off to music rights holders to use as a royalties enforcement device. Sing 'happy birthday' within earshot of your voice activated door lock and you'll get a bill for royalties. Hum a pop tune within earshot of a cell phone and get sued for creating an unauthorized derivative work.

Other commercial uses for ubiquitous intelligence require no inter-business collusion. Computational intelligence everywhere would enable the spread of inkjet cartridge-like brand lock-in to all consumer products. Imagine light fixtures that require brand-specific light bulbs, with compliance enforced by encrypted communication between the fixture and bulb. Or imagine light bulbs that only work with certain electricity providers. More insidiously, think of consumer products that selectively degrade in response to consumer activity. Cell phones could brick themselves if their owner criticizes the manufacturer, the network provider, or any cause either manufacturer or provider bankrolls. Fridges could shorten then lifespans or spoil your food if they detect you're an environmentalist. And so on.

Ubiquitous intelligence means ubiquitous surveillance and ubiquitous control. Like most technological advancements, this is always a good thing for the wealthy and powerful but never a good thing for the public.


If such chips and devices are cheap and ubiquitous, we could see a return to a world that our ancestors lived in - every tree and rock (or artificial object) will be "alive" and have agency. You could hold conversations with the environment and feel steeped in an enveloping mind.

The need to do routine tasks will disappear. Everything will just work and maintain itself. Almost all coordinating work might disappear. The cops in halting State and Rule 34 would have a completely different function - as almost any crime could be solved almost instantly.

The natural environment, especially the unnatural farms and gardens will be highly monitored and controlled, perhaps even to even plant and animal, even the insects.


Artificial Weather TM - weather control with swarms of airborne robots. Creating clouds and rain where needed. Reflecting sunlight to cool the ground.


That's actually impossible. Or rather, it's not useful. (Consider the bandwidth limitation imposed by trying to describe the environment to someone. Now consider the problem -- an AI-hard one -- of prioritizing which aspect of the environment to describe first.)


Rule 34?


Let me turn this around and ask if and how the kind of itty-bitty, ubiquitous technology being discussed here might (note and emphasize "might") help solve problems we currently think are important and hard. Such might be, just to prime the pump:

  • How brains work to produce organized behavior/ intelligence/consciousness.

  • How that DNA stuff produces hugely complex and huger structures like hands and, yes, brains.

  • Of course, prime factorization, everybody's favorite.

  • Stabilizing the plasma in a GWt magnetic confinement fusion reactor.


Taking it a different way I see the production of really good (and quite freaky) software capable of acting in ways that mimicking intelligent animal/human behaviour. With Moore's law grinding to a halt the focus in the industry will switch to making software that does a lot better job with the tools it's given.

In conjunction with really cheap, really powerful chips you could get an awakening of the world around us where every product has some bizzare yet useful (and importantly artistic and intuitive like an iPhone OS) software to interface with.


Actually, I own a laptop that can be unlocked by my face. The software is crude, and it does have a practical ceiling of usefulness, as you point out, but it's been here for a while. My laptop shipped with a copy of FastAccess installed; it's a nice little parlor trick, but not exactly something I'd use as a single form of protection ... and this wasn't a really high-end laptop, so I'd guess that FA and similar programs are readily available now.


You're missing bitcoin, Charlie. Although you've not yet included any major pure software changes. What if trade, or at least banking, can be run entirely across the web?


I can tell you as a parent that while I monitor, I don't necessarily act on what I monitor. Remember that the "blind eye" is a very useful technique.


Realtime language translation will probably be standard on the 2020 model of the Google Phone, both voice and visual.

A phone is a useful interface for translating street signs and restaurant menus; probably not so great for books. If there is still a demand for translating books and official papers 9 years from now, look for either single-purpose translators the size of a paperback book that function like a cm thick piece of glass, but provide augmented reality translation, superimposing translated text over the (dimmed down) original; or, this might be the killer ap for AR goggles.


One thing that would be developed quickly is not microprocessor based, but is a cheap EMP generator to fry them. There is information on the web to guide the development. The small microprocessors will be very vulnerable to EMP as shielding of any size would defeat the idea of smallness (actually would they survive in the normal electromagnetic environment?).I see criminals being the first to deploy EMP devices but if the microprocessors were as ubiquitous as others think, and given some ideas above, many other people would deploy the devices as a means of self defence


More environmental consequences:

--If we start seriously investing in planting trees and sequestering carbon in the soil, we'll need cheap, effective monitoring equipment, something we don't really have right now. If the world goes in for bio-char, we'll see proliferation of techniques for monitoring the soil. While determining the amount of charcoal in the soil is somewhat cheaper than determining the total amount of carbon in the soil, it's still a pain. Similarly, with trees (especially in dryland habitats) most of the biomass is underground. Currently the technology to monitor this kind of thing (root cameras) is fairly crude and extremely expensive. These will benefit from better, cheaper instruments.

--Continuation of the arms race between poachers and conservationists. Even now, poachers are part of the international black market (and are often heavily armed), while conservationists are using civilian drones, RFID tags in rare plants and animals, and similar law enforcement gadgetry. If micro-UAVs and similar gizmoids become prevalent, we may well see the Serengeti wired like London is now, and every wild tiger, rhino, and elephant has an armed UAV as a babysitter.

--For agriculture, cheap sensor nets may be really useful for things like measuring nutrient flows, pest monitoring, and similar things. A soil test for a cup of soil currently runs up to >US$100 for total nutrients from a lab, as one example. Getting that kind of detail and accuracy in a $1 sensor would change things dramatically.


The dark side is the potential for disruption. A world with ubiquitous, communicating devices is going to be easy to disrupt with jamming, machine predators, rogue code and information that misleads and traps the unwary.

It really may feel like good and malevolent forces warring with each other.


"Single-shot"? Why bother? They're cheap. Vast swarms of small kamikaze drones.

(As you probably don't recall from my LJ, I have an entire as-yet-unpublished near-future novel about the first stage of this.)


H3g3m0n @ 39

I read your "ultimate DRM", and the thing which springs to mind for me (as someone who grew up in the 1980s, at the beginning of the whole "computer games" thing) is my old, LCD display "Donkey Kong" game. Which was exactly what you're talking about there: a single box which played a single game and which only played that one game (and would only play that single game for as long as the batteries held out - no upgrades, no changes, nada).

Or are you suggesting something closer to the notion that the consumer would be sold a "console" which is essentially a battery framework which supplies power to a game cartridge or disc which contains effectively the entire console-plus-game on a chip? I suspect the game programming companies would be of two minds about that - on the one hand, they'd be able to pretty much produce console-independent games (as in, it doesn't matter which brand of "console" you're using, you'll get the same gameplay, the same sorts of sprites etc) rather than having to either subtly vary their product to suit the market (as per the arguments about Xbox vs PS3 versions of certain games) or the console manufacturer's perceived market. On the other hand, they'd be heading back to doing a lot of very basic interface and system programming (which is currently pretty much handed off to the console manufacturers) rather than concentrating solely on the gameplay stuff. On the gripping hand, the types of optimisation which could be achieved by knowing from the start the type of hardware you're using and the particular chipset you're coding for could lead to some rather interesting changes as well (what springs to mind is the sort of graphical optimisation that SquareSoft obtained over the course of three different PS games - Final Fantasy VII, Final Fantasy VIII, and Final Fantasy IX - as they got more knowledgeable about what could be managed within the hardware constraints of the one console; they managed something similar on the PS2, although not quite as graphically spectacular, considering the varying differences between their first PS2 games - eg FFX - and their final generation ones, such as Kingdom Hearts II).

Of course, the other fun bit would be the standard computer industry mantra that "standards are wonderful things: this is why there are so many of them to choose from". I suspect instead of standardising on a single form factor for their little hardware block, each game design company would have their preferred version. Which means the game "console" of the future would probably look rather like a multi-outlet power adaptor designed to work in sixteen different countries simultaneously. Oh, and we'd still have the standard fun and games where stuff made in or for the US would explode in a shower of sparks and magic smoke if it were plugged into the power supply anywhere else (because the US, for reasons known only to their initial planners, works on 110V while the rest of the known world works anywhere between 220 and 240V).

My strong contention is that the future will look a lot like today, because humans remain human, and we will of course do just about anything to keep our lives pointlessly complex.


Wouldn't information processing this cheap and ubiquitous put the big bandwidth companies out of business? No more Sprint, Mediacom et. al. acting as gatekeeping rentiers. Not when you can have millions of thimble-sized repeaters scattered across the land like so many acorns.


Everybody's got their own customized possessions, right? Sharing that stuff with other people can get to be a real headache if the basic settings conflict, so people tend to hang out mostly with other people who's stuff can be shared easily.

The endpoint is when people's stuff actively start steering them towards the "right" romantic partners and discouraging the wrong ones. Much drawing room comedy ensues as the strong-willed teakettle is at cross-purposes with the equally formidable dining room set in regards to whom young Bertram should be dating.

Come to think of it, I believe Philip K. Dick already did this one :-) To a lesser degree, so did H. R. Pufnstuf.


andrewD - Oh, that's a good one. I was just thinking that person on person crime would be a thing of the past, and then I read your comment.

Heuristic matchmaking. Lifelogging allows the creation of an inexpensive app that develops realistic, complete personality templates. The matchmaking algorithms being developed by eHarmony and are perfected to the point that everyone's ideal soulmate is identified for them.

Which is great for young people, not so great if you're 50 and pretty happily married with kids going through college and a house, but not blissfully in love. Oh, the temptation just to meet your soulmate.

Or the complications if you're from a traditional culture, where your parents want to arrange a match.

Matching someone with their ideal soulmate is not necessarily a good thing, either, as all of the Bonnys will meet their Clydes.

"Watson" on every machine, and on every search server

And to use another cuurent IBM project as an example, Smarter Cities. Instead of the current model of a panopticon, with effectively dumb cameras everywhere, smarter sensors - CCTV cameras which can run face-detection software in realtime, networked tracking software using this so that as you walk out of frame on one camera, the tracking knows where you're coming into frame on another. Traffic lights which do video analysis to alter their timings for optimal traffic flow, again with distributed algorithms relying on multiple smart nodes operating in concert. Water mains with smart sensors to spot leaks (if you think this isn't an issue, live in Dublin sometime - we're living in a nation famous for its rain and we get water shortages because of leaks in the water mains...). And of course, the improved error detection hardware that we've already seen in some public phones (especially as these die out, because you need the phone to call for repairs rather than paying people to inspect them regularly, and also for collecting cash and so on), will start showing up in more and more equipment as it gets too complex to service easily; likewise, as we get more and more room to play with hardware on-chip, expect more and more redundancy as well to bring down MTBF and maintenance costs (and thus TCO).


I worry about the implications of massive computational power for oppressive states. With enough computational power (it is not clear to me that this particular level of computational firepower is sufficient tho), as well as good enough natural language processing and synthesis capabilities, you could bend content around to the POV you're imposing. Imagine a great firewall of china that edits content, in realtime, to support the official line. Talk about an echo chamber...


Super cheap CPUs + computer vision(&smell?) + robot arms = automated robot fruit pickers for crops that currently are picked by hand, like lettuce or strawberries. I'll leave to others to work out what that means for immigration.


Several commenters have already offered Panopticons. They'll be operated by insurance companies, and come with behaviour modification a la Thaler and Sunstein's Nudge. Want life insurance? Give our software these permissions in your "extended identity"/sphere of control.

Subtle environmental cues will subconsciously persuade you that you don't want that dessert, that you want to go for a nice bracing walk, that you want to hand over your savings to the nice insurance company, etc.

One consequence: drastic reduction in size and scope of the whole criminal justice system - Nudge-type mechanisms will control impulsive crimes before you feel the impulse. Another consequence: many fewer accidents of all kinds. A third: the complete helplessness of people who wander into places with no bandwidth.


Vebayste, I don't want to rain on your parade, but our current processors are already near-useless in space: the transistors are so small cosmic ray damage renders them useless within weeks or months. Devices a couple orders of magnitude smaller would only live minutes.

I think the most practical application will be making things that are currently dumb as rocks slightly smarter. If your car was being driven around by someone it didn't recognize, wouldn't it be nice if it let you know? (Maybe it's just your brother.) If you lent out your new hardcover book, wouldn't it be nice if started nagging the borrower after a couple weeks? (And call you up after six weeks.) How about if your mum's pill bottles and her alarm clock and her fridge and her clock checked up on her and let you know she was having a bad day?

None of that's even difficult. The issue is figuring out how to monetize the damn things, to create a revenue stream with them. Because that is what's going to happen, and the time to figure how the hack the process or pass a law against it is right now.


Smart Dust: nano-processors that feed on ambient energy gradients and come in a variety of flavors all network meshed back to report data. Flavors: moisture, temp, alkalinity, pressure, visual

Combine this with tattooed circuitry:

And you could have an external sensory network feeding you data via your tattoo circuits as pressure on your arm, warmth, or flexible display tats.

However really "smart dust" comes in when you take a concept like this IBM chip: and use memresistor tech to make high-density neurons with memory, cheap enough to scatter like dust. Then the world does come to life with intelligent agents embedding the world. Given how high tech is magic to most now, imagine the typical non-tech user inhabiting this world...

Perhaps heating a house/work site with the heat from millions of processors built into the walls (think embedded in plastic with interconnects and rolled out as sheets behind the sheet-rock) would be feasible. There's your cloud computer.. need more heat? You can sell more computer time in nano-transaction jobs to run in the wall computers.

Earlier posts mentioned CGI replacing actors. Take it up a notch with super high level video compression that is totally algorithmic in nature. Entire scenes (even live sporting events) are broken down into a CGI description of the world and transmitted to your TV to be generated on the fly.. who cares that the people in the stands are just library models at certain detail levels. Think the tennis Hawkeye replays but in this you cannot tell it from live play.

(1) flat displays where all the memory and storage of a current state of the art 8-core 3ghz server fits into the cardboard thick panel.

(2) Lego style computer blocks that snap together memory, sensor, display, etc.. blocks Yes, primitive examples of this exist in research labs, extrapolate out snapping together whatever computer resources you need

(3) genetically modified plants that take up smart dust during growth to grow solar powered computational tree farms (the cloud's in the forest!) No genetic modification necessary if you can tap photosynthesis to power the smart dust.

(4) The migration of vast computer resources to colder climates as super high-density computation becomes ubiquitous and the waste heat generated becomes the new target of climate change legislation.

(5) highly accurate weather forecasting as current super-computer level computational power becomes evenly spread across the world. Think weather-underground stations but with current weather forecasting simulation power in each station and smart dust drifting in hurricanes and tornadoes.

What wouldn't be effected.. You can stick with the concept of super-dense chips for extrapolation, or add things like molecular pumps on the surface, the ability to change surface reflectivity, nano spectrographs, molecular gates, etc.. The ability to have cheap throwaway computation tied to the ability to manipulate an external element and feed off ambient power is a game changer. Perhaps not full nano-tech as envisioned by sci-fi but quite implementable at 5nm with nowhere to go smaller.


Subdermal, networked, customizable lEDs. Tattoos become animated, luminescent, and responsive to the environment. That nicely detailed Japanese style tiger on your arm turns bright red and pulses when the heat index outside your house is lethal. The tribal tat around your wrist disappears when your blood alcohol level passes the legal limit. You could have a keypad and screen that you can call up on your forearm. All your mobile apps on the back of your hand.


I was sure you'd mention quantum computing and the threat that it poses to current generation PKI (and secret key) encryption in a talk about network security. Is there some context I'm missing? Am I behind the times already - is quantum crypto research already addressing that threat?



That should be LEDs, not IEDs, but those too.


Recent research has found ways to break "unbreakable" quantum encryption, though classic encryption is still vulnerable to quantum codebreaking algorithms. The jury is still very much out on what benefits quantum computing will actually provide; as yet there aren't many useful algorithms, and QC can't improve the performance of classical algorithms.


Smart Dust based fusion power plants.

Control the confinement of individual hydrogen atoms from water to ensure they are accelerated and meet at exactly the right angle to provide fusion on an atom by atom basis. Marriage between microfluidics and smart control to do what tokamaks can't on a brute scale.


I think you're right about that particular application, but I also think that the level of integration we're talking about will dramatically affect enabling technologies for all kinds of handicapped people, including the blind. WIth the kind of massive signal-processing capability 100 billion transistors can provide, you could make a lot of real-time decisions about the nature of nearby objects imaged with a couple of cheap cameras and a wide-angle doppler sonar depth detector, and generate speech for warnings about the step down 2 meters ahead, or the dog approaching from the side, or the vehicle turning towards you on your left side. Navigating blind usually involves a relatively one-dimensional view of the world, projecting your path ahead and finding out what lies along it. If your navigational widget gives you a running commentary on what's ahead, coupled with occasional warnings about things approaching from the sides or behind I think that would handle most of the requirements. It would be even more useful if it knew your route and included Google map information in the mix.

And hearing aids could use a lot more computation to do a better job (I know this from firsthand experience: I have a pair of hearing aids that were state of the art 5 years ago, and I've been following the advances since). It's still very hard to distinguish one of several streams of conversation in a group (the "Cocktail Party" effect), or to tell which direction a sound comes from.

Computer aids for cognitively-impaired people (Alzheimer's patients, those with Down's Syndrome or Fragile X Syndrome, memory disorders, stroke damage, etc.) have been proposed for some time, and current smartphones have almost enough juice to handle them except that the user interface must be more flexible than any available now (Augmented Reality will help with that, but probably won't be sufficient). More processor power will make that easier.

what happens when spimes acquire enough on-board processing power to act as the container for their own virtual existence?

I'm not convinced that on-board processing will replace simple IDs in a large percentage of spimes. RFIDs will always be very much cheaper than processor chips (when a hundred-million transistor processor costs $0.10 in million quantity a 256-bit RFID (larger than existing commercial products) will probably cost less than $0.001 because the processing for the processor is much more complex and takes longer, because the manufacturing of the processor requires fabulously expensive fabrication equipment which has to be amortized, and because a large part of the cost of the processor is testing (lot testing isn't usually sufficient for a reasonably low DOA rate). So most things, especially low-manufacturing cost items or components of other items will keep their data in the cloud.


erm @ 34 It's called: BABELFISH !

Curmudgeon @ 40 Like most technological advancements, this is always a good thing for the wealthy and powerful but never a good thing for the public.

You're in the USA, right?

Lastly. NO-ONE has mentioned this technology in co-operation with FABBING. This would ensure much greater precision in all 3 dimensions - which is the bugbear of fabbing right now. Then what? Really well-made, made-to-order apecific objects and tools - in your garage/garden shed/loft .....


I'm with @52, life logging for more than humans. Start with whales, tigers, and elephants. Funding can come from lifelogging for pets, so people don't have to worry about their cats and dogs wandering off.


If you have game-specific chips you're going to run into the cost of designing the hardware. It's a different sort of problem, the costs of setting up the chip-fab are significant. This is more than a write-once PROM.

Design costs are going to have to fall, or will have to be spread over billions of chips. Just making the masks for the etching can run to a million dollars.


"Actually, I own a laptop that can be unlocked by my face."

I just had a flash of someone slamming their face down on the keyboard, and the computer reading the pattern and height of the keys they hit...

With enough processing power, you can get into the age of "subtle security." Once you program the suite of processors and detectors with the right inputs, any intruder will have to deal with finding out the actual challenge first, before they can even hope to defeat it without being detected. It could be anything from the things mentioned upthread (face, gait, whatever) all the way down to particular hand gestures or breathing in a certain pattern - and with enough variation in processing and detection, it would be very hard to defeat without leaving obvious traces or setting off alarms.

Of course, a bad guy with enough breaker software and hardware would just dump a couple of (literal!) buckets of processors onto the job and sit back...


Wouldn't information processing this cheap and ubiquitous put the big bandwidth companies out of business?

See my USENIX talk for my take on that. Shorter version: no, unless we get really good at routing -- and then we run into trust problems: if you're doing your internet banking over an open network, how do you know whether the folks on one of the intermediate nodes are poisoning your DNS cache and running a transparent proxy for the banking server so they can do a man-in-the-middle on your SSL connection?


It seems that trojans on customer PCs are the predominant way that banking accounts are compromised these days.

Running a transparent proxy for a bank server still requires that your end-user fail to notice that the connection is not actually signed correctly, which is something that most browsers will whinge about in fairly strong terms. Of course, if you can compromise one of the signing keys of the SSL cert issuers then you can do whatever you want, but that's a bit hard for your average criminal gang.


One thing you missed - which has major security implications - is the coming collision of robotics with ubiquitous communications.

Robotics hasn't taken off yet, although great progress is being made in the movement-and-sensing department (as can be seen in this extraordinary robotic seagull: This is because computers are presently too dumb to be trusted with any difficult task. However, communications allows us to sidestep that problem.

You know all those articles on how it is better to be a plumber that a computer programmer, because local activities like plumbing can't be offshored? Uh huh. In the future, a robotic plumber will be delivered to your house, and it will be operated from wherever labour is cheapest. The parts of the job which don't require intelligence will be gradually replaced by software. One guy will supervise a few robots doing the same job, just as today one guy supervises a few auto-checkout machines in Tescos.

The current platform battle is over who controls your mobile phone OS. the platform battle of 2025 will be over who controls the OS of your housebot. Companies will provide cleaning, cooking, eldercare, repair, and entertainment services - and maybe even local manufacture - via the bot, just as companies today provide information services via your computer or mobile.

This doesn't even require much new science; it's just an engineering evolution of tech we have now. So it will happen in 10..20 years.

The security problems are kind of obvious....


Don't we currently ship all browsers with a shared secret?


is quantum crypto research already addressing that threat?

The threat is fairly well-understood, and has been for years; the real problem is constructing a quantum computer, which is somewhat less well-understood.

The only public announcement of the sale of such a device is the D-Wave machine recently sold to, IIRC, Lockheed -- but it's arguable as to whether it's a true quantum computer.

Doubtless the NSA knows more than this, but they're not talking.


"What are the consequences of powerful microprocessors getting really ridiculously cheap"

Powerful microprocessors are really ridiculously cheap. Why are we not living in Science Fiction land already today? Most major corporations and governments today have access to Bigdata, why are they so seldom using it for useful stuff or implementing evil master plans?

Mostly they are just bumbling along trying to maintain growingly complex system portfolios and organisations, trying to keep chaos from penetrating the gates. if we consider Parkinson’s Law as just a specific case of the second law of thermodynamics, we can understand that people are too busy to actually implement the internet of things.

We could have done a lot of very cool things just with current technology and processing power, but we never seem to have the necessary implementation bandwidth.

Schools and kindergartens overflow with clothes that kids have lost or forgotten. If we had embedded a RFID tag on all clothes the parents would have been able with a mobile phone and an app to register the ID to a specific kid. Whereupon teachers could just use their phone to read who a forgotten item belonged to. Why have not done this years ago?

Following that thought if we RFID all clothes we could revolutionise washing of clothes. First the machine could automatically sort the bin of dirty clothes according to washing instructions registered by the fabricator. After washing and drying, the machine could automatically sort according to registered owner. You would no longer need a washing machine in every household, but have a fully automatic service in every neighbourhood. Thus reducing space needed in each apartment, and also through cutting down the "cycle time" that a clothing item use from wardrobe and back you could reduce the needed inventory of clothes and thus also cut need for space and dead capital in your wardrobe.

Preferably I would of cause like my laundry delivered by pneumatic tubes. Pneumatic tube is another one of those superior technologies we never seem to have time to implement. Think about what a delivery system and waste disposal system using pneumatic tubes would do your personal logistics and the need for roads in a community.

But think what we could do with both pneumatic tubes, and an internet of things…


Smart Dust based fusion power plants.

That sounds, if you'll forgive me for saying this, science fictional (for Star Trek values of "science fictional").

My take on fusion power is that, even if cheaper technologies such as Polywell fusors turn out to be viable, there are huge obstacles to deploying it which ultimately make it barely advantageous over fission.

(Fusion releases energy in the shape of neutrons and gamma rays. To get useful power out, you need to capture these. The result is that your reactor structure and primary heat transfer system will become intensely radioactive and accumulate high level waste products through transmutation and secondary activation of their components. They'll also suffer from the same neutron-driven embrittlement as conventional fission reactors. They're not even a useful anti-proliferation measure, as if you install hunks of U238 inside the primary containment you can use those neutrons to breed plutonium. And this leaves aside any consideration of where the reactor fuel comes from. The easiest reaction to drive, D + T, requires tritium, which doesn't occur in nature and is mostly produced by neutron activation of deuterium; but the reaction barely releases enough neutrons to breed fuel for the next stage. Aneutronic fusion or fusion using other elements -- 3He or Boron -- takes order-of-magnitude higher temperatures than the 100M Kelvins required by D + T. These problems are hard ...)


Why would it be world-changing? Remember what Gibson said: the future is already here -- it's just not very evenly distributed.

Tracking? life-logging? wildlife monitoring? all present already. Would any of these things, magnified to global scale, mean a qualitative difference?

(Setting aside people's paranoia about the Panopticon, which is kind of stupid, or Vinge's ideas about law enforcement, which are more so.)


The drones are intriguing, but I am not sure how useful a military technology is that could be stopped by chicken wire or mosquito netting.


Machines that operate at unstable equilibria.

Wacky medical stuff. End of all natural diseases, cure for cancer, etc, etc. Obvious, boring. Add meta to all your omics, do what's being done now for individuals to whole systems. Whole environment metabolism monitoring.

Full audit of the Earth's proteome. There's got to be some cool stuff in there. Molecular simulation for high throughput determination of protein function.

What if you maintain the biological environment at an unstable equilibrium? Weeding is an example of this. It's easy keep a garden weed free, harder to weed a weedy garden. So this sort of thing, but at a much finer level of detail. Or more exotic things. Fragile but very efficient systems.


The drones are intriguing, but I am not sure how useful a military technology is that could be stopped by chicken wire or mosquito netting.

Logical fallacy. Viz:

"Pistol bullets are intriguing, but I am not sure how useful a military technology is that could be stopped by bullet proof vests or helmets." (Enter, stage left, A Tank.)

Wars are fought with a combination of arms, and we won't see small swarm-drones deployed without a countermeasure in mind for the basic defenses, like chicken wire.

More to the point, the whole dynamic of the arms race since 1945 has been towards smaller numbers of vastly more accurate weapons. We've got smart bombs and very accurate artillery and with things like the XM-25 we're seeing intelligence pushed down to the level of squad support weapons such as grenade launchers ... but the logical end point of the individually smart bullet (the minimum sized weapon for disabling or killing an individual) isn't here yet. I note that in a contemporary conflict on average on the order of 10K to 100K rounds of ammunition are expended per enemy death. What are the implications of reducing that to one round, one kill? Even if it's a $100 bullet, it's going to work out much cheaper overall (and put less strain on the military organization's logistics chain).


The first drone blows a hole in the wire; the next hundred thousand stream on through...


Power supply and interconnectivity need to come in conjunction with the disposable processors. If all the processor-embedded packaging and products in warehouses, shops, cupboards and landfill sites were cloud-computing nodes while not being used for their primary purpose (an active standby mode I suppose)then there would be ever-increasingly powerful, very low cost processing on demand for everybody and everything to tap into.

Many devices will become dumb terminals that outsource their processing requirements, but there will always be a need for self-contained processing ability for devices that are used out of contact with the processing cloud. These may however become quite unusual and specialised pieces of kit that anyone living in a populous area would simply have no need for.

So who gets to use how much processing power? I guess there will be processing markets and processing brokers.


So who gets to use how much processing power? I guess there will be processing markets and processing brokers

There already are.


@ 78 Ok, but DARPA are already there, with an artificial HUMMINGBIRD. See: Here and again, here Scary. Only question is what is the endurance of these things?

Charlie @ 83 Percisely - look at the timetable of Steam-Power developments, back at a previous thread. Like how long from the Newcomen engine to the marine application of the Parsons Turbine?

Carlos @ 85 They fly in flocks - some are bombs - which detonate on encountering the netting, letting the rest through .... Oh, I see others spotted that one too.

ANOTHER area where "Universal_Logging(TM)" has not been mentioned, which I find very suprising, especially on this blog is, erm, err, shall we say; "Rule 34" ??


Monitoring costs in environmental work: the biggest limit in most environmental studies is sample size. You figure out cost per sample (time, money, and resources), then you fiddle with the sample design until you can figure out what a sample size that small can tell you, if anything.

Example: For my PhD, I was working in part on soil biology, and so one whole project was limited by soil analysis, which took 20 about person-hours per ounce to find and visually ID organisms. At that point, DNA analysis of the same sample would have cost about $500/sample, so I trained undergrads to help me with the grunt-work. If DNA analysis cost $5/sample and extraction took 2 hours (getting DNA out of soil is a chore), then I could have processed at least 10 times more samples, which would have radically changed the study.

That's just one of a myriad of examples. A more extreme example is Google Earth. A decade ago, people paid a lot of money for aerial photos. Now, they're not only free, in some areas you can also get historical photos records, also for free. We're starting to use this with volunteers (also free) to find environmental violations that are visible from the air.

All these areas face quantitative limits, and there are many things we can't do due to lack of resources. When technology enables us to do them, it makes a qualitative difference. That's the point.

93: Idea: Breadcrumb routers. If you need to stay connected to a particular network without physically being there, leave a trail of tiny routers behind you at 50-foot intervals.

"Only question is what is the endurance of these things?" Depends on how good their actuators are and how much their controls concentrate on power consumption, but I'd guess around an hour. Once graphene and nanorod ultracapacitors get figured out you'd probably get five times that.


Matt Ruff's Sewer, Gas, and Electric includes a DNA-based targeted genocide, along with frozen Walt Disney, a resurrected Ayn Rand in a hurricane lamp, and a bunch of other oddities. Odd book.


87, 88: Actually, I'm looking at it in terms of countermeasures. Warfare is a capital-intensive business, sure, but defensive measures are by and large much cheaper. You're assuming drones will play the role of the tank in WWI. I'm assuming chicken wire will play the role of barbed wire in WWI.

If you want, you can assume "smart" chicken wire -- say, by using short-range jammers -- or "advanced" mosquito netting (Kevlar fibers). But simply "more" chicken wire might suffice, the same way that "more" anti-tank ditches or "more" bunkers often suffice.

I don't suppose I have to bring up the use of then state-of-the-art detection technology in Vietnam, which I think is the classic case of the transistor era, or how well artillery fire removed the barbed wire from the trenches during World War One -- roughly equivalent to the hundred thousand drones against a chicken-wired position.

Also, I'm not sure what problem micro-drone technology is supposed to solve better than the alternatives. First, explosive power scales as the third power of size. Something twice as small has an eighth the boom. Second, fuel for flight scales as the third power of size. This has scale problems too, depending on the method of propulsion. Why not "smart" bullets?

Now, these doohickeys combined with IED technology -- the "smart" micro-mine -- they seem more threatening to me. Not world-changing, because I'd expect the development of similarly "smart" mine-sweeping units, but there the countermeasures will likely be more expensive than the deployment, which can be drop and forget. That's the side of the equation you want to be on.


Considering roughly 100% of browser SSL warnings are false positives, we're teaching people the wrong set of reactions to rely on that to save them from MITM attacks. Plus, even if it did work, the SSL cert. market is completely broken. Tim @79: All browsers are shipped with dozens of root certs, and nobody will revoke inclusion priveliges of a cert. vendor with insufficient customer verification or other problems, leading pretty directly to the latter point above.


I only read to about comment 50, so bear with me if I'm repeating some ideas, but my own thinking tends to run perpendicular to most of fandom. I'm too pragmatic to think that governments or corporations will do things just because their customers think it's a cool idea. I tend to look at money, politics, and law first when considering technology, and not the gee whiz that most tech geeks do.

My first hope, and a rather boring one I know, is a phone that has approximately the same processing power as the one I have now, but will hold a charge for days if not weeks, and won't get hot.

The same technology can also be used to make batteries more efficient, so a smart phone that can hold a charge for longer than a day would be great. Putting nanoscale structures inside batteries to increase their life is a technology that's being experimented with now.

On the subject of phones, I expect faster and more efficient Internet bandwidth, since you can slap traffic control just about everywhere on the cheap. Additionally, I'm expecting the very notion of using wires to connect to the Internet in urban areas for individuals to be a relic of old dinosuars like me. Even now, in urban areas in Chicago people are using 4G phones to get their internet access.

People in more remote areas will still need satellite service or wires. At least in the US they will, but for most Americans I expect that wireless connections to the Internet will be the norm.

Surveillance, in urban public places at least, will be ubiquitous and cheap, and online most like. Cameras and WiFi are already cheap, but processing the video feed is still on the slow side. It should be possible for a small business owner to buy a DIY store surveillance kit that lets them slap a bunch of small, battery powered, WiFi enabled cameras in their shop. Not only that, but feed back to a computer with massive solid state drives or the Internet at large. It should even be possible to build a VR simulation of distant places with a rig like that.

The applications are endless, and a mix of good and bad. Virtual tourism, real estate home tours, spying on spouses, security in places like stores, homes, and prisons, and a real time version of Google Maps.

There are some interesting legal issues here. What's your liability if someone kills themselves in front of your cameras that you put online, and the Internet being what it is, the video goes viral? Will the aggrieved family have some cause to sue you for their emotional torment? What expectation of privacy should people have?

The only technical snag would be people's willingness to share their video feeds, and that seems to be a social barrier that gets lower every day. The only legal/ government snag will be the willingness of governments and police forces to have ubiquitous surveillance that isn't under their control.

It becomes even more interesting if the technical and social hurdles of A/R are overcome. If you've got a GPS enabled phone with high bandwidth plugged into a set of cameras on your head you'd be adding to that full VR experience of Google Maps, or whatever.

I don't expect the money to get smart, not in the US at least. People are emotional about their currency, and won't put up with "smart" currency that tracks them. There might be RFID on the money to track it, but no chips that change the value of the cash depending on what you buy.

Besides, most economic activity happens over computers now anyway. I do expect the concept of "chipped" ATM or Credit Cards, and government IDs to be everywhere. It's a technology thats only limited by the cost outweighing the benefits. Soon, people will just be waving their phones at each other to transfer cash. That's already started.


Guided bullets:

DARPA is on the case.


It's not impossible, it's just that the descriptions of the environment need to be based on short-range high-frequency radiation...perhaps light.

Describing the environment to blind people requires more recognition of the environment than anything else. The "description" problem, however, is that a verbal description is nigh unto useless. You need either a direct feed into the visual cortex, or a mapping procedure that preempts a normal sensory channel. The back has been used, the tongue has been used. Neither has been totally satisfactory. The back is low bandwidth, so you need a very large display area. The tongue has involved a wire hanging out of the mouth among other problems. And even so in both cases the resolution has been markedly inferior to a 1940's TV.

So what you're talking about here is not something impossible, but something that requires a different tool. A neural interface. This isn't a "cheap computation" problem, but it's not impossible, either. And it could certainly be facilitated by cheap computation. And by parts of the environment that identified themselves.


How about some of the old stuff, that never happened, finally coming true? Fully automatic factories?

Think of a Henry Ford style production line from the 1930's. Big leap forward from craft production, but there is still stuff going wrong, gumming up the works, requiring craft work interventions to fix problems. Fast forward to the Toyota production system of 1990. For car making you can keep up the pressure on getting production flowing smoothly. If you never have a molding with a piece of flash that stops it fitting, if the bolt holes in the pressings are never misaligned, etc, you cut another big chunk out of production costs.

I think that is where we are today with manufacturing: if the component parts going into your factory are flawless you can move to very high levels of automation. I suspect that is part of the attraction of high density fibre board in the manufacture of furniture. It is more uniform than wood, sufficiently so that you can import car industry techniques that depend on buying a roll of steel and having it perfectly uniform.

We haven't so far seen 100% automatic factories, running lights out, on any important scale. Today's dumb robots require uniform materials, such as steel, or homogenised materials, such as fibre board, and even then cloth defeats them. Any little hick-up or flaw in the inputs causes the machine to stop. And a jam in a robot factory is costly. You have to clear the machines and restart.

But what happens when computer power gets cheap enough that industrial robots, that see and feel, can respond quickly and adequately to what they have seen and felt. What happens when they can chisel wood and stitch cloth? What happens when a sewing machine is not a power tool for a clothing worker, but more like a web fed rotary printing press: a big roll of cloth on one end and a stream of shirts coming out the other? What happens when robots can see how a piece of mold flash is stopping a part from fitting and trim it with a paring knife? What happens when the robot can look at the bolt holes, seen that they are not quite right, and jiggle things, or do a little filing?

These questions all went out of fashion. In 1960 automation was going to eliminate factory jobs and cause mass unemployment. The predictions didn't come true. Instead robots were unable to compete with cheap Chinese labour, or pretty much any labour with eyes to see and brains to respond to what it sees. Computers have had eyes for years, but brains have been too expensive.

Charlie's question is what happens when computer power gets cheap. My anwer:

EITHER brains for assembly line robots remain expensive because the code to get them to make shirts, assemble washing machines, etc, is too expensive to write and maintain. Welcome to a big revival of research into artificial intelligence.

OR old questions, about automation and unemployment that went out of fashion, come back into fashion.


Talk to me like I'm a venture capitalist. This sounds great if you're a PhD candidate in soil science. But what's the killer app?

Let's assume that overall funding will be $1 per person on the planet -- roughly the size of the CDC's current budget. What results will make it quickly worthwhile? Here, I mean as quickly as installing electronic keys in hotel room doors, or scanners on a checkout line.

(There are strong reasons why investors and institutions have short time horizons, as much as you or I might dislike it, so the "scientific knowledge will pay off in the future" line is not going to play here.)


You don't use "smart bullets" because of the problem of friction vs. mass. And a "smart bullet" can't hang around waiting for a target to present itself.

OTOH, I don't think explosives are the assassination tool used by a smart drone. Think instead of poison of one sort or another. There are toxins that would be quite effective for this use. Or, if you were feeling less lethal, LSD. Opiates probably wouldn't be effective at the low dosage that any one drone could carry. Neither would any of the more traditional poisons. But there are a few from the microbiology lab that would. Some quite quickly, some quite softly, others quite painfully.

The chemical used should be effective for the intended use at microgram quantities. That means either a disease, or something unusual (if LSD still counts as unusual). Things that aren't recreational will currently be exotic. And notice that diseases tend to be very slow, so if you want a quick effect, you go directly to an effective toxin.

This does mean that you need to penetrate the skin. Mosquitoes, however, do this all the time without being noticed, if ever, until it's too late. Usually not until the bite starts itching.


What kind of a detection range are you assuming for your killer hornets?

At the moment, the various airport bomb sniffers require you to hold still, stand next to the sensor, and have compressed air pumped at you. However kewl the processor, I can't see this sort of stuff having a useful hit rate out in the wild because there is just too much stuff. We're already in the position of finding man-made chemicals everywhere we look just because the instrumentation is good enough. We're already in the position of being able to RT-PCR anything...and every damn one of its contaminants (see today's Observer for one of the outcomes - the XMRV CFS cockup has led to scientists who were right facing the full Steve Milloy/Animal Rights nut playbook).

In a battlefield environment, there's going to be traces of explosive...everywhere, even before the insurgents get a spray can of liquid rich in nitrogen and a long pole. In practice, I presume you'd want a wide-area survey sensor to get a handle on the environment.

Which gives me an idea. EMP devices can be pumped by conventional explosives. Which can be detonated sympathetically. Your sensor should find a sizeable bomb open to the air, but that may be the last thing the whole system does, carrier aircraft and all.

I'm rather surprised we haven't met a guerrilla tactical EMP yet. Not that much harder than a good compressed gas EFP, surely? But then, blood counts for something. They probably prefer to kill men rather than fuxxorise their computers.

On other suggestions, I've yet to see any evidence that biometrics will ever beat the Iron Law of False Positives (ILOFAP), and certainly not that its problem is one of "damn, not enough welly" rather than "damn, I wish I was smart enough to come up with a better specified problem". See also auto-indexing splodge (you know splodge - images, video, audio - that stuff).


Fueled drones have the problem of, well, fuel. If my back-of-the-envelope derivation is correct, they have a time in flight that's sort of like the delta-v in the rocket equation: it's proportional to ln(massinitial/massfinal). Not insurmountable, but a major design consideration. Are they going to be able to hang around as long as a sniper?

Solar- and battery-powered drones have their own problems, although the idea of these drones operating like a cut-rate version of the mythical Project THOR -- cyber hawks dive-bombing someone's head from ten thousand feet up for a kinetic kill -- is rather appealing in a Hitchcock sort of way. Or maybe Phantasm.

I suppose there's beamed power. Maybe, but if you're beaming power to a drone, why aren't you beaming power at your target? eliminate the middleman!

I'll note that the conversation has now shifted from the "swarm of smart drones outperforming artillery" to "chemical assassination weapon".


The most likely weapon enhancement is a smart sight. It looks for a Human signature (optical or FIR) and when the trigger is pulled only fires when the target is properly lined up. Intelligent spray and pray


I don't expect the money to get smart, not in the US at least. People are emotional about their currency, and won't put up with "smart" currency that tracks them. There might be RFID on the money to track it, but no chips that change the value of the cash depending on what you buy.

Too late.

Firstly, €50 notes have had RFID chips in them for years now. The only reason lower denominations don't is that the first-gen RFID tech was too expensive. Expect this to change imminently.

Secondly, while it's not cash that changes its value in your pocket, some British banks are experimenting with contactless short-range payment cards for making small purchases: the idea is that if you want to pay for a bus/subway ride, or a cup of coffee, you shouldn't need to authenticate yourself to the bank with your chip'n'pin card, but can just tap the card on the contactless reader. It's taking the contactless payment tech in, e.g., London's Oyster cards, to the next level.

The twist in the tail is that the contactless "cash" card is the same card as your actual bank debit/credit card -- it has a short-range "wallet" that you can top up at an ATM by making a "withdrawl" from your bank account that's then stored in the wallet for these minor purchases. IIRC there's a ceiling of around £25, so if someone does find a way to pick your virtual pocket they won't empty your account. The idea is that, ultimately, this could replace cash for a lot of low value transactions.

Second, fuel for flight scales as the third power of size. This has scale problems too, depending on the method of propulsion. Why not "smart" bullets?

Yes, but the small drones are operating in a different aerodynamic regime and so the scale factors are different. This is similar to the difference in flight mechanics between a hawk and a dragonfly.

Wouldn't information processing this cheap and ubiquitous put the big bandwidth companies out of business?
See my USENIX talk for my take on that. Shorter version: no, unless we get really good at routing -- and then we run into trust problems:

I'm not sure what you mean by "really good" at routing; given enough nodes the problem becomes a trivial one in the sense that bottlenecks become unimportant. Cheap repeaters equals cheap cable, ergo, more - perhaps many more - cable services (assuming that municipalities will no longer enter into exclusive contracts with just one or two providers). Especially if they can use each others pipes, which they can, given that these nodes are rather clever little devices.

In re security: data is pretty secure right now . . . if you want it to be. The problem is that for most people, "security" is something that's fiddly and eats up time and resources and is really rather pointless. Until it isn't of course. I suspect that if you doubled most people's computing power, they would not double the resources given over to security.

The same technology can also be used to make batteries more efficient, so a smart phone that can hold a charge for longer than a day would be great. Putting nanoscale structures inside batteries to increase their life is a technology that's being experimented with now.

Well it would have to be a different type of small scale machining, but yeah, that's really about the only way to improve battery performance. Bear in mind that the chemical reactions inside a battery aren't substantially less energetic than what you get with air and gasoline. But in a battery those reactions are a surface effect and in a gasoline engine the reactions take place in a volume.

So, more powerful batteries. Now if we could just make them dirt cheap . . .


I'm not sure what you mean by "really good" at routing; given enough nodes the problem becomes a trivial one in the sense that bottlenecks become unimportant.

You're talking about replacing the telcos. The telcos provide backhaul bandwidth over hundreds or thousands of kilometres with repeaters along the way and routers only near the end-points.

If you replace that with ad hoc networks of peer-to-peer junk, the number of routers in the signal path is going to skyrocket, because your packets will be going through at least an order of magnitude more folks' networks. Which means your latency is going to rise steeply, unless you can optimize your routing across multiple networks (some of which may be configured differently -- e.g. to prioritize gaming packets or video streams over speech traffic, for example).


I'm not seeing scale favor smaller drones. You're still in the Newtonian drag regime.

Dragonflies fly much less quickly than a hawk, even though they consume about the same percent of their body weight in food per day.


If your nodes know their locations in space, your routing problem is significantly easier.


So what if money gained and lost value depending on what it is being used for as a vehicle to encourage savings, encourage purchases, etc.

Tricky! Useful, though. I've actually participated in an (internet virtual) economy something like that. Money that you earned and 'welfare' money spent the same, but your welfare account shrank in realtime. You couldn't get rich by piling it up, pass it around, or invest it - just spend it. Money you actually earned could be saved, given to others, etc. It worked in the context of the VR where we used it.


Of course, the first order response of your typical insurgent group will be to press-gang pre-teens and toddlers to carry their bomb-making equipment ... it goes downhill from there, fast.

Mm, yes. I've thought of this too. You're right about the first response of an insurgent group (and the second is to get videos of occupation drones blowing away school kids onto YouTube), but I worry that the first thought of a terrorist group is to load a few hundred of these into a stolen car and drive through a city with the window open, tossing out time delay seeker missiles as they go...


Personally, I'd like to see internal labs-on-a-chip as a method of decreasing wait times on cancer screening, etc. I think (and hope) that we'll eventually arrive at a time when biopsies seem barbaric and wasteful, because there's internal, ubiquitous monitoring of things like estrogen and lymphocytes.

This would also be useful for telling if somebody had faked an orgasm. I can imagine some paranoid asshole boyfriend hacking their soon-to-be-ex's phone to see if her inner space app had logged extra dopamine or not. Ditto prudish parents of teenagers. And let's not forget the mental health industry, either -- they'd love an internal mechanism that twigged patients to how much serotonin they weren't producing.

I'd also like to see some sort of drunk driving test embedded in a vehicle. Not a stupid breathalyzer, either. That's just for legal limits, and does nothing to tell if somebody is high on something other than alcohol. If we're theorizing biometrics that can recognize faces and gestures, etc, there's no reason it couldn't also do the same for pupil dilation, wavering steps, excessive sweating, etc. I'm writing a design thesis on the customs clearance process for travelers, so stuff like affect detection and ID verification has been much on my mind lately.

I also like the "where my stuff comes from"/"where is my stuff?" side of things. One of my favourite moments in MAKERS, for example, was when the characters figured out how to Google their homes for missing ephemera. I'd like it if my things came with their own tags, though, that logged into my house whenever I brought them home from the shops. That way I wouldn't have to do the "In which ring of Hell is my aloe vera gel?" dance at the start of each summer.

The no-contact cards sound a lot like a SUICA, which I'm all in favour of. I wish the SUICA card would just take over every major city, already.

The animism idea has been done, and well, by Karl Schroeder in VENTUS.


I am reminded of Neal Stephenson's "The Diamond Age" where New Chusan(?) has a cloud of drones on its perimeter that acts as defense & immune system.

With everything, I wonder about failure modes...


"Stinking smoke wafts up from braziers made of skulls set around the edges of this room. The walls bear scratch marks and lines of soot that form crude pictures and what looks like words in some language [Goblin]. To the left lies a pile of rubbish and rubble heaped into a crude dais. The dais has upon it an ironbound chest that has been painted with a goblinlike face. Furs and skins of unknown origin are strewn haphazardly about the floor before the dais. A mindflayer is charging at you..."


"Oh, too bad, you didn't see that rabid rat just behind you. It bites you. You die. The mindflayer throws a hissy fit because it didn't get to kill you and take your experience points."


I think there are several advantages to small drones. First, square-cube advantage to solar-powered (or at least solar-assisted) drones: the drone mass, including the parasitic structural mass, goes down as the cube of the linear dimension, while the available area for solar cells goes down as the square. Second, smaller drones operate at lower Reynolds numbers; they're moving slower but they can hover and move slowly without stalling. Insects turn out to be very energy-efficient at flying; the tradeoff is that their wings have to be built to operate at high beat-rates, up to 250 beats-per-minute.

I would expect small drones to be used in missions that don't require loitering over large areas, for instance in urban areas, surveilling known locations, tracking and attacking humans on foot or in traffic, or attacking intruders entering secure areas. Given that more and more of humanity is getting packed into denser and denser urban concentrations, I would think that a weapon that can maneuver around and through buildings, stopping and waiting at various points along the way, would have a lot of uses on both sides of asymmetrical warfare.


One of the major limitations of 3D printing just now is that any electronics that needs to go into the final product must be built separately and installed in a separate manufacturing step. The most favored electronic module currently is the Arduino, which is fairly low tech, and low cost. This results in the requirement to design new I/O circuits for most products, and to code new control software for all of them. The low amount of memory in these modules means that the software has to be pretty lean to fit, and the kind of product design that we see in mass-produced systems isn't common, because it requires more sophisticated programming that is usually available.

Given cheap, large-scale control processors1, and some sort of universal, programmable effector/sensor electronics, it should be possible to print most products with a common set of electronics, the only differences being the printed wiring in the product (that's doable now) and the control code in the processor. Then the processor programming and insertion steps would be physically the same for a large class of designs, and could be handled by relatively cheap extensions to a low-end 3D printer. By reducing the amount of manual intervention in the printing process this would make home use or rental of a printer at your local Kinko's copy center much more attractive to a larger number of people.

1. Let's say, a processor with the power of an ARM Cortex-9, a gigabyte of flash memory for program and persistent data, and a few hundred megabytes of RAM for computational memory. On top of that, a runtime software environment that contains thousands of pre-compiled modules for common (failry complex) functions, so the designer can just wire up combinations of them2.

2. Yes, I'm aware of the arguments against dumbing down programming that way (I started some of them, several decades ago). Note that a lot of control programs for microprocessors in products like traffic lights, machine tools, and suchlike are written in Relay Ladder code, just as if the processor were composed of a bunch of electromechanical relays, and nobody seems to mind. The paradigm is familiar to the engineers with experience in the application area, and if they can get results with something that doesn't require retraining, projects cost less and take less time.


I mis-wrote my theories about RFID in money. We do have RFID in American money, but an earlier poster had made a comment about putting computing power on the dollars/pounds/whatever, and I don't see that happening. Banks will just prefer to set up cash cards as you describe.

As far as I know, the powers that be don't use the RFID to track individual notes everywhere they go. The cash register at the local store is not equipped to track the money.

My current employer, a temp firm, fills up a cash card like the one you describe with my wages.

It's going to create whole new and creative areas of criminal enterprise.


US currency does not (yet, anyway) have RFID. It has strips or dots of metal in it, but they're not RFID.


"If you replace that with ad hoc networks of peer-to-peer junk, the number of routers in the signal path is going to skyrocket,"

This was actually tried as a kind of "grass-roots ARPANET" back before the mists of time. Slogan was possibly "ARPANET for the rest of us"

As far as I recall it was called "Vine-net" or something like it, and they managed to stretch it from California to just across the rockies before the number of hops meant that packets sent from BSD systems with default ttl of 15 timed out before they ever arrived.


Okay, here's the venture capitalist end:

What does soil affect:

  • Biodiversity, specifically, how many plant species you can cram into a square meter. This matters for things like cost-effective ecological restoration, polycultural agriculture, and similar issues. Currently ecosystem restoration efforts are similar to 19th Century medicine, in that a restored area is often about as functional as a peg leg. In other words, if we destroy a forest, we can't replace it. We need to be able to restore the soil as well, but currently, we can barely determine what's there. Wouldn't it be great to be able to replace a forest after you've destroyed it? It would revolutionize the restoration industry, and that industry (IMHO) needs a revolution.
  • Polycultures (>1 crop per field) has been shown to be more productive per acre than monocultures, and also to have fewer environmental side effects. Currently, polycultures depend on human expertise, and can only be done on small farms. Scaling polyculture up to industrial agriculture demands much more sophisticated knowledge of what's going on in the soil, which in turn depends on cheap, sophisticated sensors, especially if you're working on an industrial scale with few humans in the system.

    Global warming: soil is a huge carbon reservoir, but current industrial farming releases carbon from the soil into the air. It's possible to get carbon back into the soil (through no-till farming and other advanced practices), but monitoring levels of soil carbon and determining how long they stay resident in the soil is expensive: currently you send a sample to the lab (US$50/sample for a 1 cup sample at To monitor soil carbon, we need to get that cost down to something like 50 cents per sample or lower. If costs are that low, it would be easy for landowners to verifiably demonstrate that their lands are a carbon-sink, and we could set up reasonable market mechanisms to get the carbon out of the air and into the soil.

    That's off the top of my head. Hope it helps.


    I do all my banking over the web other than getting money from the ATM (I normally spend cash locally) and depositing my private disability check (I could have it direct deposit, but then it would come into my account six days later.) Oh, and when change fills the bowl on the foyer/litterbox table, I take that in to go through the machine.

    Did you guys know that Wells Fargo is going to charge their debit card holders $3 per month? As this says, they're doing it because of the new law that sets a fixed amount that banks and such can get for debit card use. The banks and other financial bodies are all probably thinking of something like that. If my credit union did that, I'd get money at the ATM (free), but pay my rare off-web bills with a credit card.


    Dragonfly weight is about a gram. Not a lot of solar assist in an urban or indoor environment, which most secure areas are -- my guess is it would be chrome on a mass-critical device, but maybe not. It seems clear, though, that it represents a tradeoff. Probably the tradeoff point can be calculated.

    Anyhow. I am going to assume a time range of about an hour at one-third of a dragonfly's top speed (30 km/hr). This is based on watching the things buzz around by the lakes where I grew up, and assuming biology has found a sweet spot in the design space. Unless technology vastly improves, these things aren't going to be able to "live off the land" the way actual dragonflies do. If you want, you can assume the mission takes place at night.

    We have more tradeoffs. If the target is five km away as the dragonfly flies, that's half of its useable lifespan gone. We can speed it up, but there's a power curve. So it has thirty minutes to achieve its objective. It can land, but it will take energy to take wing again. It can hover, but that's power-intensive too.

    I'm going to say half the mass of this drone is useable payload, not counting whatever signal it can broadcast. I think that's generous. If we are talking about things that go boom, well, 0.5 g of PETN is probably not going to blow a hole in the chicken wire. If it lands on someone's face and detonates, sure, that would be a bad day for the poor bastard. But if it lands on a solid door, would it even move it off its hinges? Half the blast would necessarily be outward, too: it's not like those videos on YouTube where they stuff a few grams into a tree trunk or a brick wall.

    You can mess around with the numbers, but they seem pretty unprepossessing to me. How many of these things would one need before they move from "nuisance" to "threat"?


    How about if your mum's pill bottles and her alarm clock and her fridge and her clock checked up on her and let you know she was having a bad day?

    Look here: there's already one developed. Be sure to look at the image. I wouldn't mind staying in one of these when I get to that point, although I don't really have relatives.


    This is already a personal method of assassination. People are "accidentally" struck by the end of an umbrella held by a stranger and a day later is seriously ill.


    Biodiversity and global warming are problems with time horizons longer than most investors have patience for. But let me introduce a phrase into this discussion: "charismatic flora". The famous, the lovely and rare, the ones with special economic importance. People interested in those plants are the people you should be thinking about promoting the idea to. A Mondavi who can monitor every vine, a Weyerhaeuser who can monitor every tree. There was a university somewhere in the South that had its historic tree poisoned by a vandal recently; I would bet they'd be interested in something like this. Sure, they're Red Staters, but even they will apply themselves to a problem when it starts directly affecting them.

    Yes, it doesn't help the broader issues... initially. But it gets the production lines and protocols set up with the help of deep pockets. Then you point out the next steps.


    From The Game Players of Titan by Philip K. Dick:

    It had been a bad night, and when he tried to drive home he had a terrible argument with his car.

    "Mr. Garden, you are in no condition to drive. Please use the auto-auto mech and recline in the rear seat."

    Pete Garden sat at the steering tiller and said as distinctly has he could manage, "Look, I can drive... Start, darn it!"

    The auto-auto said "You have not inserted the key."

    "Okay," he said, feeling humiliated. Maybe the car was right. Resignedly, he inserted the key. The engine started up but the controls were still dead. The Rushmore effect was still taking place inside the hood, he knew; it was a losing argument. "Alright, I'll let you drive," he said with as much dignity as possible..."


    RE: the platform battle of 2025 will be over who controls the OS of your housebot. Companies will provide cleaning, cooking, eldercare, repair, and entertainment services...

    The question I continue to have is at what date will we have more than sweeper bots?

    What do you suppose the day will be between now and 2025 when we'll have a bot that either acts as chef or house assistant?

    And what would this bot look like? I can't imagine a walking bot, nor a bot that's any larger than the current iRobot sweepers.


    Living in hong kong we already have an "octopus" card that does this already most major shops 7-11 supermarkets vending machines and all public transport run on this.

    99.9percent p Of population carry one


    Are we at the point yet where a typical laptop's webcam's image of the user can be processed quickly enough so that the laptop always knows exactly where on the screen the user is looking?

    (I used to think this would require all sorts of crunching and extrapolation of the shape of the user's eyeballs would be required, but then I realized: You have the screen, so you can post graphics to it and then look at where/how it's mirrored.)


    Fair enough. I agree that charismatic fauna and flora can be helped by better monitoring. The issue there is that many of them (e.g. elephants, rhinos, orchids, cycads) have an increasing value on the black market, and money to protect them is decreasing in the current political climate. The value there is if you can have a monitor to replace a warden or a guard. That's not great, but it helps.

    As for the examples I pitched, I was trying to pick infrastructural technologies, where the technology was vital to enabling new industries to work. There's a lot of money to be gained on the front end of such a market, when the sensors are being proved and widely adopted.

    There's also money to be made in the second phase, when such sensors become commodities. Those who can manufacture quality sensors in bulk could have markets all over the world.

    Anyway, that's my thinking. Besides, if investors can't find ways to make money solving the biggest problems we face, our civilization is doomed anyway.


    Hearts and Minds in War.

    What does the military have trouble with? What does the military want in terms of tech?

    Does Hearts and Minds not = roads & bridges, food, jobs, and structure?

    Roads & Bridges: Don't they equal a stable net today? And would that not demand a managed power source? I would like to believe the amount of processing power described just might fabricate a physical minimally hardened grid (don't ask me how) similar in nature to what the US had as far back as 1970 or so.

    Food: I honestly don't know what Monsanto is up to though I would guess offering cheap franken'food is somewhere on their "to do" list for second and third (and first) world countries.

    Jobs: Difficult, we're running out of those here in the first world, yes? Who was it that claimed "death to AI/Tech" will be the bluefaced warcry for WWIII or are we already there?

    Structure: I would argue that future structure amounts to a 24/7 (brilliant AI) net moderator that acts as a psychiatrist and bouncer (using NUDGE tech) - only when the AI cannot control the citizenry does the military get involved, identifying the trouble makers and possible terrorists - and hopefully separating them from the antagonistic journalists, taggers and street artists. Structure also equals potable water and toilets. We need a brilliant iRobot OS to plug into a third world nation state and automagically give the country working plumbing that, within a year, covers the largest towns and outlying areas.


    Control in War.

    What is the military good at? Killing. What is the military NOT good at? Crowd Control.

    UAV's have changed the face of War, allowing for 24/7 tracking of various elements in a war zone. Current next step is a lethal strike from a UAV or an AC-130 gunship strike or somesuch (ground-troops at last resort).

    Future Next Step:

    24/7 UAV identifies Hostile Group.

    "Minutemen Swarms" dropped from 24/7 UAV's.

    Swarms stun Hostile Group for "TAGGING".

    Those Tagged will be further PUSHED to comply with the larger national net.

    Those TAGGED near a SECONDARY HOSTILITY are then detained.

    Protesters are hit with non-lethal swarms dropped from 24/7 UAV and again "TAGGED" or, if identified as repeat offenders: terminated.

    PUSH tech again drives all toward working plumbing, working electrical grids, good food, and a working (net) society.


    RE: War. Excuse me: PUSH tech = NUDGE tech. I got my terminology skewed...


    Molecular modeling folding@home style where nano-scale resources on a vast scale get routinely scheduled to design better drugs or vaccines.

    Fluid dynamics aerodynamics, pipe flow dynamics, battery cathode/anode best fit simulations, etc.. Where enough power and resources exist to evolve optimal designs.

    High Frequency trading where algorithms discover methods to manipulate profit hundreds of trades out in the future. Where billions of processors can watch every trade, anticipate ripple effects and gear up to capitalize on trading events 10-20 years out, let alone in the millisecond regime.

    infectious transmission modeling Where the sea of movement information tracked by smartphones, travel information, and climate gets processed to simulate optimal vaccination strategies. Then use molecular modeling to exhaustively iterate through optimal vaccines, ship the appropriate number to the optimal locations and finally notify the medical establishment. Eventually totally automated by medical systems trolling hospital databases.

    Retail simulation and modeling RFID tags on everything track your movement through the store, where you pause, what your eyes focus on for what duration, etc.. The model that represents your purchase history and inclinations becomes as valuable as any purchases you might do. Markets spring up to buy or steal those models.

    smart dust surveillance Scattered smart dust that lies mostly quiescent snooping electromagnetic and visual spectrum. Passing ship/plane/car/satellite wakes them up briefly to extract data Note: could easily scatter onto hair or clothes of the target, or around some corporate facility


    Entertainment; As you sit in the cinema the microsensors in your seat continually check audience satisfaction and rewrite first the secondary subplot and then, on the basis of that data, the plot to meet audience expectations. And after a hundred iterations, every movie ends up as The Terminator.


    @ 131 And how will CATS react to house-bots? Yes, I know, I've seen the YouTube of a cat-riding-a-Roomba!


    Ok, I totally skimmed the comments, but here's a practical note on modern "digital hearing aids". My Mum recently got one, and has been staying with me on holiday (I'm fortunate enough to live in what's generally considered a scenically beautiful area). When she's using her aid, she's asking me to turn the Tv down, and it's set up to default to a comfortable volume for my "ok" hearing.


    Something which would be easier and cheaper: really good software-defined radio.

    In fact, networking in general might change so that the specialised router or switch would be a thing of the past. Everything's a Linux box with one of those is-it-a-terahertz-laser-or-is-it-an-NIC devices you were talking about.

    Perhaps the Internet's structure shifts towards Van Jacobsen's idea of content-centric networking - if processing is that cheap, there's little objection to putting forward deployed application servers and CDN nodes in street cabinets, under your TV, in buses, etc. In which case there would be a strong case for a shift to a content- or function-centred addressing architecture. IPv6 autoconfig gives you routable addresses and either the routing infrastructure or whatever replaces DNS in this scenario tells you where your nearest source of whatever is.

    Also, if routing happens on big CPUs, that relaxes various assumptions - you don't need to worry about TCAM burn and the routing table just has to fit in RAM.


    I don't think the assumed scaling is reasonable, and a bundle of other effects cut in way before the scale Charlie talks about.

    The first is a short channel loss of gain. This was pointed out by Likharev in the mid 1990s - once the channel length gets done to 10nm, a FET shows no power gain, even if for an optimised, back gated design.

    Some applications, like memory arrays, can live without power gain, as the address line drivers and readout circuits can be made using larger design rules to make up the losses. But general purpose logic does need power gain, and so it seems unlikely that design rules beyond about 15nm will be useful.

    The next effect is the failure of voltage scaling. For many design rule generations, finer lines went hand in hand with a drop in operating voltage. The increased number of devices was compensated for by reduced power device, partly due to reduced current, and partly due to reduced operating voltage. At about 45nm, voltage scaling stopped, as the scatter in threshold voltages due to the small number of doping atoms in a tiny transistor area became significant. In fact, for some manufacturers, the core voltage at 45nm has to be higher than it was at 65nm.

    If voltage scaling breaks down, then there is an unpalatable thermal problem trading off against reduced clock frequencies.


    I hoped someone else was thinking along these lines.

    Think of the combination of smart drones - able to recognise a specific area and to home in on specific targets - with some form of hard to remove RFID marker.

    Take a bank for example, or high end clothing/jewellery store. Smash and grab raid comes in, clerk hits button, everyone in the immediate vicinity gets tagged by the swarm. Store owner and police together have each half of a code to deactivate the marker, and cause it to biodegrade faster for the innocent bystanders. Raiders can be traced by setting up sensors across the road network, combined with cctv and an uplink to the relevant forces.

    Or take a riot/public disorder situation like earlier this month. Send a cloud of drones up over an affected area and either tag everyone indiscriminately and sort it out later, or with smarter drones, tag those specifically with say accelerant residues or glass fragments, or if you get really clever, with the combinations of pheremones and sweat that indicate anger/violence/satisfaction etc, to identify the active participants.

    Again, it allows the police to manage the situation, while making sure they can both round up the active participants easily and combined with cleverer sensor linked cctv, trace the paths taken by criminals making it possible to identify fences.


    This technology is already creeping into smartphones. Considering the technology and the convenience for all involved I expect the smartphone of 5-10 years time to be a travelcard, wallet, key (though only for keyfobs until normal locks are replaced) etc. Though there would have to be really good backup/cloud software to avoid the disaster of loosing your phone. This is the biggest issue for me, if everything is done by your phone then loosing it would leave you right up the creek. Can't call for help, can't travel home, can't get into your home, can't take money out to pay for stuff.


    ISTR that there have already been SF stories written about this happening, or more sinisterly (is that a word?), someone's details being deleted from the central database that controlled the money, transit system, and "key access" records.


    We are; this is the point at which I mention a friend whose webcam tracks whether he's leaning forward or backward and scales the contents of his browser window to match. For tracking accuracy reasons, however, gaze tracking is done via IR reflections from parts of the eye (the cornea and the back of the lens), rather than visually apprehending gaze direction. Most webcams have IR exclusion filters built-in...

    You're missing bitcoin, Charlie.

    I don't think bitcoin is weird enough in itself to make any real change to anything. It's just old-fashioned currency. Ripple might have been more interesting, but it's sadly crippled by using the word "dollars" throughout the narrative.

    Ubiquitous networking makes possible a much more complex and nuanced system of money than the old fiat currency or the even older gold standard.

    Wonder what it will be?


    Not to mention the fact that those pesky, evil, money-grabbing telcos invest quite telephone-number-magnitude sums of money to ensure that many millions of sheep and fast tracts of trees and wild flowers also have at least a reasonable access to the network. (and have to by the way as a condition their licence commitments)

    I know this because I am in fact employed by one of these pesky evil money-grabbing telcos, and can say for sure that we have cell-sites that can go days without significant traffic... Just on the off-chance you want to facebook that amusing picture of you and your mates half-way up a welsh mountain.

    Oh yeah, and I wish I knew where these pesky evil money-grabbing telcos are hiding all that filthy lucre, given my share options are in the toilet, we haven't had a proper pay-rise in years and my laptop hasn't been replaced in even longer.


    I read a comical short story years ago set in a Niven style ringworld. The protagonist jumps on a maglev to visit a friend but get's off at the wrong stop (several hundred km from home and his friends). As he tries to get back on his implants crash and wont reboot, thus he's left penniless, without communications and has no ID. He has to live on the streets and walk home, which takes months.

    In reality that would be my worry, how to cope if someone steals your phone (that also acts as your wallet, keys and travelcard). One idea could be similar to Natwests emergency money scheme, basically if you loose your wallet you can ring them and they will cancel your cards and give you a special code. With this code you can go to an ATM and withdraw cash. Perhaps an idea similar to this would be that you can go to any phone shop and tell them you've lost your phone, they can cancel it (wiping it in the process) and give you a new handset with all the software loaded from a back up in the cloud. Not sure what you'd do if it was late at night or you were far from a shop...


    Relevant paper from the Cambridge security lab.


    There is one element in Rainbows End that people rarely discuss but that depends on extreme miniaturization and ubiquitous high-speed wireless communication: the frequent offloading of processing onto other people's hardware (or municipal hardware, etc). There was this pervasive assumption (that turned out to be, shock and horror, fairly untrue) in-universe that these tiny devices were so secure (because of the trusted computing mechanisms built into them) that it was entirely valid and safe for relative strangers to take over your machines, whether they be on your body or on the street.

    While I don't consider this extreme to be reasonable, I consider it at the same time to be an interesting and potentially useful idea in other contexts. If in ten years your door handle has a processor faster than the one in your laptop today, and your shower head has one too, it stands to reason that rather than lugging a laptop around you can lug something around that's essentially a dumb terminal and use these various nodes floating around to do the hard work. From the perspective of people producing these devices (if they are compatible with each other, which I suspect will eventually become practically a requirement in order for them to be practical) there will probably be, for security reasons if nothing else, a requirement that the kinds of operations they perform be provably non-turing-complete. Say, perform only finite sequences of simple mathematical operations, with no flow control. If the flow control is handled by the dumb terminal then it's much harder to pin the load on your door handle indefinitely and have it melt.


    One thing that would have to change is how systems deals with failure. The more things you have, the higher the chance you need to deal with failure.

    With massively-parallel comms-bound architectures (as most "active dust" architectures would need to be), you're looking at a need to design for you comms, your processors and your storage to have bit failures frequently.

    This would, if nothing else, put a bound on the usable computation you can get out of things. If your typical computing environment are a couple of thousand low-power CPUs with an expected life-time of three years, you're looking at losing a couple of CPUs every day and if your computational model can't deal with failure, you'd be in rather a bad shape. It's probably possible to structure things so the "machine" you work against is a virtual machine, of some sort, that mimics being reliable and simply stay hidden from the actual computational substrate.

    I am not entirely sure how that ties in to the rest of the interesting speculations, but...


    I think this would require a modification to the multi-processor paradigm:- 1) You have to assume massively multi-parallel processing rather than "just 2 or 4" parallel streams. 2) With the assumption of 1 to 3 processors failing every day, you might well have to devide the farm into 2 or 3 sets for doing "more critical" work, and use a "voting system" to establish the correct answer when a processor fails during a critical process.


    That was at Auburn University, and considering that they not only have an athletic rivalry with a school that has a few deranged fans, but also have a number of related majors (horticulture, agronomy and soils ...), they would probably be very interested in technology of that, um, nature.


    I wonder what the metabolic load of an immune system is. What if we could replace that with a shared, artificial system. Get some nice economies of scale going. Some security implications, of course.

    Regarding drones, shoot them out of the air with lasers. Relevant TED talk:


    It's one of the more interesting areas of practical CS - and one which the Google architecture has had to address. I was chatting on Saturday to one of their sysadmins about how code written for or ported to the Google software stack has to cope with the concept that any particular 'machine' may vanish at any moment, but an application as a whole must continue operating without a glitch.

    (Ingvar above, if he's the Ingvar I think he may be, would be probably aware of this as an issue.)


    I was just thinking that cheaper, more efficient computers will aid in the continuing wiring of the developing and rural world.

    The issue of evanescent connections will certainly play a role there, as will pervasive man-in-the-middle hacks in ad hoc networks. We may also see a "sneakernet" backbone of cryptography keys as an inherent part of security in the developing world. Families and clans may become defined by their security arrangements, as much as by their blood relationships.

    One interesting question: will they become technoserfs, or something else?


    some British banks are experimenting with contactless short-range payment cards for making small purchases: the idea is that if you want to pay for a bus/subway ride, or a cup of coffee, you shouldn't need to authenticate yourself to the bank with your chip'n'pin card

    When I was in Shanghai a few years ago, you could use your subway card (one of those RFID tap-and-pay affairs) to pay for taxis, restaurant meals, groceries, etc. It was actually more useful than a credit card for ordinary transactions.


    I realize this is just going to get buried, but I wanted to point out one critical difference between now and 1980 and the doorhandle metaphor. Take 6 letters out of doorhandle and start imagining what a server's worth of embedded chip is going to do in your hand.

    You were talking of data security in that excellent tl;ra (too long, read anyway) post from USENIX. Having a cellphone/cloud manage your security will always make it less secure. So you'll follow the route that keeps data secure in your brain, namely keeping it inside the ol' corpse. We wont have cellphones, we'll be them. We wont have cloud storage of lifelogging, we'll become cloud storage of lifelogging.

    Especially if you can fit 90 years into 60-whatever grams of theoretical diamond memory. I'm sure we could replace an organ or two with a smaller, more efficient device (mechanical, nanotechnological, bioengineered, etc. or combinations of such things) and use the extra space to host our own storage solutions. We'll store a hash with secure, trusted third parties (at least three, preferably more) as a way to verify that we haven't hacked our lifelog to provide ourselves an alibi. We would only be required to provide excerpts for law enforcement/identity theft that match the hashes.

    We've already got the basics for much of this, all that remains is the will. As battlefield intelligence needs grow, I suspect we'll see DARPA looking into lifelogging and cybernetic enhancements related to such before too long. Real time satellite is great, but when you can combine that with near-real-time eyeballs of the same areas (or recordings from the recent days if communications are spotty) you can algorithmically search for enemy tactics and other subtle cues that would otherwise go lost on soldier's minds in the heat of battle.


    I think of it as ubiquitous intelligence. Doorknobs wouldn't necessarily be smart themselves but they'd be in constant communication with truly smart devices and databases. They'd not only know who you are and whether or not to let you in -- they'd open the door, adjust the lights and audio to your known preferences, etc. In fact, one thing that might become rare is the issuance of a command. I think that we could probably consider it likely that all societies would be surveillance societies; a consequence would be that everything everywhere knows what you like and when and how and... Wouldn't it be nice if at the moment the thought occurred to you that a nice cold Pilsner Urquell would taste good, a bottle is delivered to your hand at your preferred temperature? Your environment might know you well enough to anticipate your needs and desires. Actually, the more I think about that the more frightened I get...

    In fact, perhaps only the wealthy and powerful would know privacy. They would likely need it as warfare has the choice of changing to a personal level, rather than continuing on the wholesale basis it operates at now. If you started by executing your opponent's chief executive and continued with, say, one member of Parliament (or Congressman) per day, how long would the war last? Naturally, it would pay the powerful to be anonymous and hidden; at least, identifiably whereas common folks such as you and I might no longer care.

    There would be no blind, no deaf, no handicap of any sort that we know what causes it and have a suitable power supply available. In fact, for most of us how we perceive the world would be a matter of choice: you might see the world as a Tolkien universe while your sibling could prefer Star Trek. We could easily lose track of what is 'real' and what is mediated to our preferences.

    Personally, I'm rather anticipating seeing most of my contemporaries as hobbits -- or as a panoply of Star Wars aliens. Maybe the ones I don't like would be in Imperial clothing, while my friends would be rebels. Or vice versa. Knowing more than one language would become a rather quaint eccentricity, as instantaneous translation is the norm. I don't think we'd have to carry magic phone or shiny wrist-watches, though: I think that most of this stuff would be embedded, using your body as the power source, and you would literally become a node in the network.

    I wonder whether it would be illegal, perhaps highly illegal, to remove your implants and drop off the network? (Yes, I know: stories have been written.) Hacking would be considerably more... interesting, I suppose, as well as having the potential for being more personally annoying. Imagine everyone you see in your moderated experience looking like your ex, only exquisitely beautiful and naked? Or perhaps everyone who sees you sees you naked and 60 pounds heavier? And with hairy warts. And a tiny little... Yeah.

    Anyway, if you can postulate intimately interconnected intelligence everywhere then you can, I think, postulate living in a literally magical world, where anything is possible. That'd be different. :)


    Sorry if this is a repeat: I'm late to the party.

    If you take Lifelogging and add lots of processing power to OCR/transcribe all your spoken/written words and add a dash of tracking of every single object, article, video and audio you interact and then add a giant glob of bayesian analysis, which requires lots of processing power, and then you have a model of how you will interact with the world.

    If this model exists then it can suggest things to buy etc. but more interestingly, you can create a copy of it to interact with the world when you don't want to. Don't like the caller, send a copy of you to deal with it. Don't want to disappoint your mum/brother/wife/sister make a copy to send them emails, birthday presents, etc.

    Imagine what happens when you forget to shut down one of these copies and it lives for a long time, and then makes a copy of itself...


    Electronic door knobs became popular in hotels very fast because there was this infrastructure of skilled locksmiths and locksmith services companies already in place.

    Locksmiths could already repair small, finicky things so they easily made the transition to repairing or replacing circuits inside the locks, and to keeping watch over that tricky connection between dgital and "ancient" metal hardware.

    Cell phones changed a lot in society because there was already an infrastructure of electronic technicians who could repair and replace the circuits inside the cell towers.

    If you want to know which future predicions will take off and change the world look at the human infrastructure needed to repair and replace them. You always need this pre-existing human infrastructure.

    Cars have the service network needed so there will be a massive amount of sensor based and communications based features available in them. They won't be able to drive themselves (real roads are too complicated) and they won't be able to understand complex voice commands (language is too complicated since it depends on culture) but they will be able to do absolutely everything else, including taking over the "anchor" or base station role of the home PC.

    If you want to know which future military technologies will take off just start by eliminating all those predicted weapons which have the slightest risk of making the rate of death by friendly fire rise(like homing bullets)or the slightest risk of being taken over by foes, either remotely or by hacking after capture. Officers will not let those things anywhere near a battlefield. This is the reason why DARPA and others paid so much attention to things like packbots and robot helicopters for delivering supplies and medical evacuation.

    On a related note, thanks for the link to the cheap Nature subscription a few weeks ago, Charlie. ",)

    You're welcome.

    Ubiquitous networking makes possible a much more complex and nuanced system of money than [...]

    That system of money/payment, whatever it will turn out to be – Bitcoin or not – plus desktop computer-level processing in smartphones that absolutely everyone on the planet will have, might make for a pretty interesting extrapolation of Amazon's Mechanical Turk.

    Right now most Turkers come from the US and India (IIRC) but if and when the poorest places on the planet get to participate in that same labour market, some economically weird things might happen.

    And also botnets at an unimaginable scale.


    How about glass octopuses? I have two of his, they're very good.


    Erm, a chef robot came out two years ago.


    I'm not sure why everyone thinks the barrier to drone development is processing power -- we already have cool quad-propeller drones. The problem is they get 10 minutes of flight time on a charge. Small autonomous robots have a power problem, not a processing problem.

    I don't want another billion transistors per millimeter, show me an order of magnitude improvement in energy storage and then I'll get interested.

    I'm not sure why everyone thinks the barrier to drone development is processing power -- we already have cool quad-propeller drones.

    But most of the processing for the sensors and motor control of those quad drones isn't on-board; it's in computers at the other end of a radio link. Yes, some of them can hold position or fly in a straight line to a designated point autonomously, but that's nowhere near enough for the sort of missions we've been talking about. Image recognition and goal-seeking control won't fit in an Arduino just yet.

    Regarding drones, shoot them out of the air with lasers.

    Arms Race: cover the drone with a reflective coating, hell, just wrap it in aluminized Mylar 0.5 mil thick, won't weigh more than a couple of grams (that film is 19 gm / m2 and is available by mail order).


    Security compromises will stop being something you hear about on the news, and start being a simple fact of life. There is graffiti on the wall outside your house and somebody has reprogrammed all of the street lamps to flash on and off. Your credit card details get stolen three times a week, and the banks are relying on spam filtering technology to stay ahead (you're still trying to dispute a charge from last month which they claim is consistent with your spending habits). People come back from holiday to find that a removal company was paid to send all of their possessions overseas and the locks were all set to let them in. The police are jumping at every shadow on their CCTV feeds and somebody has rigged the cameras on the high street to add suspicious bulges to the jackets of black kids.

    And none of this will get reported because it happens every day; microprocessors run every aspect of life and bored teenagers control them. The general public will continue to insist that "if you do nothing wrong then you've got nothing to fear", the companies that make the devices will continue to insist that they are secure, but will expend all of their resources on new security features and none on attempting to reduce the number of exploitable bugs beyond reacting to exploits found in the field.

    In the late 20th century, general low-level public fear and disorder was based around gang crime and guns. Sometime in the 21st, it's going to be based around losing control of the hardware.


    @ 171 We're STILL afraid of gang crime and guns. And with good reason. Did anyone else see the film-clips released by WMP of an-all-alike-dressed-in-black-gang firing shots at police, and deliberately targeting one of the best pubs in the country, the heavily "listed" "Bartons Arms" on the suburbs of Brum? A link showing this can be found Here

    Quite obviously The Bartons - A truly wonderful boozer, in which I have drunk) had told these scumbags to piss off, and they were trying to trash the place - reports suggest that arson was tried, but failed. I think the late 20thC problems you list are still with us.

    Yes, some of them can hold position or fly in a straight line to a designated point autonomously,

    I remember seeing something like that on TV and going "wow!", but that was more than a decade ago.. :-)

    No, the sensor / motor control processing is done on board. I believe that is the thing that has made quadcopters such great hacker toys – let go of the stick and it won't immediately destroy itself. And it doesn't take more than your average 8-bit microcontroller to do it either.

    Now we have hobbyists building quadcopters that can almost take that part for granted, and who get to play with identifying a moving target and landing on it autonomously. We are already pretty far along, and batteries and software is probably where things need to happen, not primarily processing power.

    Image recognition and goal-seeking control won't fit in an Arduino just yet.

    Not the original Arduino platform perhaps, but that is already more than half a decade old. Modern updates to it already use far more powerful chips. (And just the other day, I noticed that there is a BSD Unix that targets a 'duino form factor platform.)


    Where process scaling breaks down is the subject of huge ongoing debate. You may be right, but you may also be very wrong!

    My 2c worth is that moore's law is not acheived by process scaling alone anyway. The industry as a whole achieves the roadmap, but the effort required from the non-process guys to get chips to work with even the current 40nm transistors is huge and increasing. Just because we min out the number of layers of atoms in the gate does not mean that all the other groups involved will stop improving - even manufacture can still get better, reducing process variation, eda/physical design can improve increasing density on the silicon, low power architectures can come in making super dense packing of chips practical (through silicon vias to allow chips to be stacked are avalaible today, it's heat that limits density of chips from being veritcally stacked into a tiny space).

    I'd guess the improvements on the table could be good for at least the eqivalent of one process node's scaling, if not two, once the device physics puts the brake on CMOS.


    Actually, I expect self driving cars to slowly phase in. I'm thinking more along the lines of motorways with sensors embedded into every catseye or a guideway down every lane. You bring your car into the system, and turn on the autopilot which puts it into the control of the motorway. When you approach your destination, the system will automatically alert you and bring the car across to the appropriate exit, at which point the driver regains control of the car.

    Basically means you still have a human in control for the complicated stuff of navigating feeder roads, driveways etc, but you can automate the repetitive stuff of long distance driving and alleviate a lot of the fatigue induced accidents and congestion issues.

    Accidents would still occur, if less commonly, but since the motorway would be smart enough to know of them, it would be smart enough to automatically route traffic around them and to alert the necessary departments. And you would significantly reduce the tailgating/speeding/swerving/generally bad driver behaviour from the mass of drivers and improve travel times. Convenience wins out over excitement every time, and the car would be able to give an expected arrival time based on speed, distance, and what the system reports on road conditions.

    You could even keep the multi-lane setup we have now, one primarily for heavy traffic, one for short distance trips, one for long distance trips or something along those lines.


    That sounds like the kind of massive infrastructure spending that noone in the western world is interested in any more.

    More likely is that the current set of crash prevention and adaptive cruise control mechanisms already available in high end cars grow and merge slowly into a fully automatic car. The main advantage, from the POV of the car companies and governments, of all of this is that there is no need to change the law, the driver is still liable (and needs a license and to be sober) even when they are just sitting there. If there ever is a crash they were "in control" and should have acted to prevent the collision. The advantage to the rest of us is that these collisions will become much rarer.

    But I'll still have to get a taxi back from the pub :-(


    "As part of this writers suddenly become much more valued (hah! you wish!)"

    Nah, because there'd be AI writers.


    If you've got AI writers, then you've got AI substitutes for virtually all intellectual labour focussed around language skills, memory, planning, and theory of mind: which means almost all of them ....

    ... At which point we'll all be out of a job and nobody will be able to afford a movie ticket.

    (Or rather, it means such radical social change that current conceptions of employment will be about as relevant as corveé duties owed by mediaeval peasants.)

    Now we have hobbyists building quadcopters that can almost take that part for granted, and who get to play with identifying a moving target and landing on it autonomously. We are already pretty far along, and batteries and software is probably where things need to happen, not primarily processing power.

    Being able to deploy myriads of tags (up to and including self-enclosed spimes) would greatly simplify the appearance of intelligent behaviour. Instead of landing on an identified moving target, you could have it say "I am the top of the moving target! I am the top of the moving target! I am the top of moving target! I am . . ."

    Same thing for autonomous driving - the problem is immensely simplified if tags like "I am the car in front of you!", "I am the car behind you!", "I am the edge of the road!" etc are liberally strewn about.

    One side effect of this would be the increasingly sharp and increasingly wide divide between rural and urban areas. In the city, transport is electrified. In the country, it's mostly still gasoline. In the city, your business is almost entirely done online and anything you order is shipped to you within minutes via pneumatic tube. In the country, you still have to slog about on foot to do your shopping or have a human deliver your stuff to your door. In the city, robots are cheerful intelligent fellows who know you on sight and will address you by name. In the country, robots slow, sullen and dull-witted. And so on and so forth.


    Bitcoin is an encrypted currency, and seems to present a reasonable possibility of making avoiding taxes much easier. This way, you can take payments over the internet without it showing up in your banks transaction logs. Of course, there is a transaction log, but it's not at all clear what can be learned from it, and how difficult in a practical sense it would be to police bitcoin transaction using it. If this is just one more problem to governments already struggling with the internationalisation of the economy, then how much less influence will they have over the world?


    I think a valuable profession for the future will be whoever is in charge of inventing new jobs to keep people occupied, the kind of people who can't leverage their creativity into a job, i.e. most of us.

    We might find the old codology of astrology etc worth keeping around just to give people something to do, a friend of mine is a teacher but is also big into all the reiki nonsense of faith healing, if technology took her job she could switch careers. Ain't no robot going to tranfer ki or whatever it is they're supposed to be doing when they lay hands.

    There's already some precedent in the official icelandic stance regarding elves and such.

    And we already have professional mmorpg players...


    We're already there, and many places have been there for centuries. My favorite example is the Greek who worked part-time as a fisherman, farmer, hotellier, fixer, and handyman. As the author drily noted, "How many jobs does he have to lose before he is unemployed?"

    The idea that someone has to give us "jobs" is part of the problem. A lot of us work part-time on a bunch of things, trying to make ends meet, although the government only cares whether we're working for someone else or not.

    You have to distinguish between two states: one is that machines do everything for humans that is necessary for human life. Somehow, I don't think that's going to happen, because humans are good some things that machines are bad at (for example: dealing with sloppy problems). However, many jobs are highly repetitive, and actually harm the humans that perform them. Turning those jobs over to machines that won't be harmed doing them is a good thing.

    There are also classes of jobs where machines are cheaper only when they are massively subsidized, and they are able to externalize their problems onto someone else. In these cases, we may be forced to de-industrialize some sectors just to get the whole-system costs down.

    This morning, I was discussing an example with my partner: would you be willing to double your food cost if it meant that you lived longer and had fewer health problems? In the US, we've gone for cheap food at all costs, where other countries have higher food costs and longer lives with fewer health problems. While this is a simplistic example, it is a case where taking some of the automation out might be cheaper.

    This morning, I was discussing an example with my partner: would you be willing to double your food cost if it meant that you lived longer and had fewer health problems? In the US, we've gone for cheap food at all costs, where other countries have higher food costs and longer lives with fewer health problems. While this is a simplistic example, it is a case where taking some of the automation out might be cheaper.

    But a typical American could also halve his food cost, live longer, and have fewer health problems. It isn't a stark choice between Dollar Menu cheeseburgers for the poor and organic salads for the affluent: simple plant products like rice, beans, barley, lentils, oats, onions, carrots, cabbage, and whole potatoes are low in fat and salt, high in fiber, can provide the macro and micro-nutrients needed for good health, and are widely available*. They're also less expensive than any packaged, prepared, or restaurant foods. The money you save by eating potatoes instead of potato chips can be shifted to more healthful treats too, so you can eat some fish, nuts, fruit, and leafy greens in addition to the cheapest staples. For 3 years I lived in a household at the federal poverty threshold and never ate restaurant food, only rarely ate prepared foods. It costs too much to eat badly! The global poor certainly don't (couldn't afford to) subsist on instant noodles and fruit punch.

    Eliminating agricultural subsidies would change the total output and composition of US agriculture. I am much less certain that it would give small producers, less reliant on machines, a large share of production. The returns to scale and labor-vs-capital cost tradeoffs that brought consolidation and automation in the first place would still largely be intact, it seems to me. It's not clear that unsubsidized agriculture, even if it did encourage smaller-scale production and less corn and soy, would benefit people stuck in food deserts. De-industrializing food production and raising food prices seems to me neither necessary nor sufficient to change American eating habits and health for the better.

    After all that, I agree that there are some machines that (at least in some contexts) do net harm and are only profitable to operate because they're not accountable for their externalities. Every coal fired power plant in the developed world, for example. Many fishing trawlers, for another example.

    *Yes, I am aware of the "food desert" phenomenon and agree that it's real, but it cannot explain the widespread bad food choices among the American working poor and Americans in general. According to the USDA, as of 2009 2.3 million households (out of 112 million total) live more than 1 mile from a supermarket and lack a car. That's at most 2% of households forced to eat badly; the other 98% have choices and choose poorly.


    explain the widespread bad food choices among the American working poor

    I've more and more over the year come to the conclusion that some people make a collection of interrelated choices which will keep them "down".

    If you are short of money why is there a case of beer in your house? is your PPV bill higher than my entire cable bill? does everyone in your house have smart phones? do you have the turbo option on your internet package? do you have 4 TVs in your house? do you have a really nice ride? do you go to first run movies several times a week. are your kids wearing Air Jordans? and so on.

    Food choices are a factor in the lives of people like this. They buy for their tongue instead of their health. You can do both but when you only do the first you tend to have lousy results over time.

    Yes, I am aware of the "food desert" phenomenon and agree that it's real, but it cannot explain the widespread bad food choices among the American working poor and Americans in general.

    Do tell. And where are all those stay-at-home adults who will be presumably cooking up dinner out of your "healthful nutritious spuds"?

    I've more and more over the year come to the conclusion that some people make a collection of interrelated choices which will keep them "down". If you are short of money why is there a case of beer in your house? is your PPV bill higher than my entire cable bill? does everyone in your house have smart phones? do you have the turbo option on your internet package? do you have 4 TVs in your house? do you have a really nice ride? do you go to first run movies several times a week. are your kids wearing Air Jordans?

    And the people I know who make those kinds of choices don't tend to be on the lower rungs of the income ladder. They tend to be people making $80 K/year or more.

    Something tells me this wasn't what you were thinking of ;-)


    Call me one of those nutritious home cooks. That was pretty much my standard shopping list, and fortunately I'm in an area where kale is available year-round and relatively cheap. I also splurge on organic peanut butter.

    As I noted above, the idea of de-industrializing agriculture to raise people's standard of health is tremendously simplistic, which, slightly unfolded, means it's another monstrous topic entirely.

    You have to put that comment in context. I was responding to someone who was saying that anything that humans can do (except procreate) can theoretically be done better by a machine, therefore it will be done better by a machine, therefore we need to train people to do useless things.

    I think that argument is bogus, because it starts with fallacious assumptions and ignores the very real costs of substituting machines for people. And I'm not a Luddite either. When we're fighting wars to guarantee the energy supplies for industries that keep food cheap, I think we need to sit back and re-evaluate a bit. The side-effects of that cheap food policy are tremendously expensive.

    Moreover, the available evidence suggests that small farmers are as productive and more efficient (both economically and environmentally) than big agribusiness. Big farms simply have more political clout.


    See "Access to Affordable and Nutritious Food—Measuring and Understanding Food Deserts and Their Consequences: Report to Congress" by the USDA, 2009. According to their research, only a small minority of Americans are stuck in food deserts without a vehicle -- 2.3 million households at most, about 2% of households. According to the 2009 supplemental data (derived from Census Bureau data) of the Working Poor Families Project, there are 9.6 million low income working families in America. Yet according to the 2009 Pediatric Nutrition Surveillance System, nearly 1/3 of preschool children from low-income families were overweight or obese. This is much higher than it should be if the formula is simply "prevalence of obesity among low income families = number of low income families in food deserts / total number of low income families." I am making a mental leap here, perhaps indefensible, in using overweight/obesity statistics as a proxy for frequently eating bad (processed, calorie-dense) food. As I believe was discussed on this blog some time in the past year, obesity has been growing in many nations among all income groups since the 1990s, and it doesn't show any one smoking gun causal factor.

    In 2007 UCLA's Center on Everyday Lives of Families found that that use of packaged and prepared foods is common among Los Angeles families, but it was used to provide individualized food choices more than anything else. Families that used packaged foods lightly had no statistically significant difference in overall meal time from heavy packaged food users. The only savings was 10-12 minutes of direct hands-on food preparation time.

    If you can microwave a cup of noodles you can microwave a potato. The hands-on food prep time for many staple foods is trivial and requires minimal kitchen equipment (like a crock pot or rice cooker and one working electrical outlet). Again, this is what I did all the time when I was low-income.

    I fully recognize that I have heaps of advantages and privileges that made it easier for me to do the smart thing. For example, I had access to a lot of reading material from a young age. I grew up in a school district where they actually taught those "little extras" since cut from many school budgets like household budgeting, nutrition, and cooking. My parents reinforced the same skills at home. I've never been confronted with social discrimination based on my race, sexuality, or perceived class. I was low-income for years, but I still had a big edge over the impoverished neighbors that I eventually left behind. I think it's not that my equally poor-by-income neighbors didn't have the time or income to cook rice and beans instead of Easy Mac and hot dogs, it's that there was a whole missing framework of experiences and skills that would convince someone that relying on inexpensive staples instead of packaged and prepared foods is 1) even something to consider 2) a good idea 3) achievable even for people with limited time and budgets.

    *Less than 200% of the federal poverty threshold.


    Sigh. You may safely assume that I know at least as much about this topic as you do. Further, you didn't answer my question.

    Now, if you aren't going to answer my question, just say so. Don't duck, weave, dissemble and otherwise try to distract my attention away from the fact that you haven't done so. That's rude, disrespectful, and insulting.


    Your question contained an invalid presumption, that you need stay-at-home adults to prepare potatoes instead of packaged foods that say "microwaveable" right on the label.


    And the people I know who make those kinds of choices don't tend to be on the lower rungs of the income ladder. They tend to be people making $80 K/year or more. Something tells me this wasn't what you were thinking of ;-)

    I've seen this in folks living pay check to pay check. At $30K per year and likely more than $100K per year. Annual earnings aren't the issue. It's a desire to always want more than you have because you are "owed".

    Working on a community pool board and having a child go through a high school band you get to rub elbows with all types. And when folks drive up to a volunteer event in a new $40K to $60K car, talk about their and their spouses well paying job, then talk about how they just had to spend $1K+ on their kids junior prom night and the new car they bought them for their 16th and ... Then they complain about having to do volunteer work so they can afford the pool or band or whatever and make it very clear that this work is below their status...

    At times I have to wonder.

    If you want to meet a wide range of people send your child to a magnet/bussed public high school and have them join the marching band. One that competes and travels. (Very USA here folks.) Then you join the band boosters. You'll get all kinds of incomes and other diversity in the mix and the best workers definitely cannot be told from the income levels. In any direction. Neither can you predict which kids will be overweight from the incomes. So there is something else at work in the "I deserve more" mindset.

    A community pool tends to give you a narrower slice of the population but still the above rules seem to apply.


    @ 183 Those food "choices" are being made in the UK as well ... I really can't understand how people can EAT most of the meat purveyed by the supermarkets. I mean the beef is PINK, ugh. And, of course, it means a lot of work, but the home/allotment-grown veg is FRESH, TASTY, and you know what went into it, and food-miles = zero....

    @ 184 They buy for their tongue instead of their health. Erm, that's exactly the WRONG way around. Healthy food SHOULD taste BETTER. In the same way that my pint of Sambrook's PA or Tim Taylor is both cheaper, better-tasting, and better for me than "Fosters" ....(shudder)

    Last few posters ... are missing something, even though it is (almost) being said. It's down to Education and Intelligence, both. The stupid, the gullible and the ill-educated will, by and large eat what the advertisers and the coorporations put in front of them, stuffed with sugars, and some unhealthy fats. The others, won't .... Income seems almost irrelevant, until you get to really low/bad poverty levels.

    Don't forget genetic traits, either - I am still wearing the same waist-size trousers as I was 35 years ago - and I don't stint on food or ale. I admit my food is really healthy and nutritious (see above), but even so! Whereas there are others who only have to look at a potato to put weight on ....


    I think the biggest impact will not be computation, but communication.

    Cheap, powerful processors will enable communications technologies that put us constantly in touch with the world - not just actively (mobile phones have that pretty much covered), but passively - information will seek us out.

    To deal with that torrent of information, those same cheap powerful processors will create new ways for us to interact with it - haptics are my personal favourite.

    This constant connection is possibly a bigger challenge to the concept of privacy than the panopticon - we'll leave ever more detailed data wakes, and sorta kinda get used to the fact our privacy is eroded; right up to the point where it has a negative impact on lots of people. My personal bet is that credit ratings are going to be the cause celebre - once nice middle class people can't get that mortgage approved because the credit rating agency knows their credit card statement includes items from a hairloss clinic, they will man the barricades.

    I also foresee ever-increasing customization and personalization - manufacturing will become more and more knowledge-driven, and even moderately priced items will be personalized. Mass production will not go away - but smart manufacturing and logistics systems - driven by cheap processing - will put mass-produced components together in one-off combinations. Clothes, furniture, TVs, telephones - everything will be cheaply available in "just for you" editions.


    Why are 'stay-at-home' adults required for cooking? I almost always cook fresh, and I'm working a full day. Now I concede, I rarely work much more than 40 hours a week, but that's hardly 'stay-at-home'. I gave up on 80-hour weeks years ago.

    No, I side with Greg here in that it's likely to be education, but with the proviso that their palates are also lacking in education.

    (Tonight's meal is likely to be pumpkin soup, since what herself thought was an enormous globe courgette wasn't.)


    Just a chance to throw out a geek moment: perhaps new technology will actually allow us to make a working 0.25c Alcubierre warp drive (or at least, a model of such in metamaterials (Arxiv ref).

    I can't see the article, but this looks like a classic August science moment.

    Anyway, back to the "how to get people to eat good food" argument. I've had this one in real life too many times to enjoy thrashing it out online.


    We are now, give or take, all telepathic (mobile phones). We won't be getting teleportation in the near future, but several other Tomorrow People-style superpowers might be possible with huge amounts of tiny processing. Invisibility cloaks are limited by processing rather than theory. Objects might all be built to detect intention of nearby people and respond in a way that would be a bit like limited telekinesis.


    We are now, give or take, all telepathic (mobile phones)

    We've been telepathic since the invention of verbal communication, if you like. We don't need mobile phones for that.

    On the other hand, being able to communicate silently, with no external signs of such communication, is possibly further away than ever, given the number of people I see apparently talking to no-one as they walk down the street.


    Don't forget genetic traits, either - I am still wearing the same waist-size trousers as I was 35 years ago - and I don't stint on food or ale.

    Do you smoke? Much? There's seems to be a big correlation between smoking and not gaining much weight as you age.

    Your question contained an invalid presumption, that you need stay-at-home adults to prepare potatoes instead of packaged foods that say "microwaveable" right on the label.

    Ah, so all you're really interested in is pushing a rather odious right-wing talking point rather than actually finding out why people have the eating preferences that they do. You could have save me some effort if you'd just said that up front and I'd never have bothered to reply.

    I've seen this in folks living pay check to pay check. At $30K per year and likely more than $100K per year. Annual earnings aren't the issue. It's a desire to always want more than you have because you are "owed".

    I really don't know how to put this any more delicately: There's a difference between a single parent of two children making $20 K/year living paycheck to paycheck and a childless married couple making over $250 K/year and living paycheck to paycheck.

    Since you are unable to acknowledge this extremely simple and extremely obvious point, there really isn't any further reason for us to talk. Which is just as well, since this is a massive thread drift.


    More stuff for "the future is here now, just spread unevenly," file:

  • throwable robots as "smart grenades" (although $10-20K a robot shows the expense of fitting a grenade with brains, wheels and a camera):

  • nanostructured glass for "permanent" optical data storage (the word is voxel instead of pixel):

  • And, of course the "smart doorbell of the future," in September's Wired.

  • 202:

    No personal attacks, please. That includes attacking people you don't know and who aren't here.

    Ah, so all you're really interested in is pushing a rather odious right-wing talking point rather than actually finding out why people have the eating preferences that they do. You could have save me some effort if you'd just said that up front and I'd never have bothered to reply.

    Please quote my actual words where I endorse a right wing talking point. I have read plenty of right wing judgment about people eating badly because they are lazy/stupid/not tugging hard enough on the ol' bootstraps. I don't endorse these interpretations. But neither do limited dollars and minutes nor food deserts explain widespread poor American eating habits. Staples don't require much time/equipment/money to eat instead of packaged foods; they're actually less expensive, calorie-for-calorie. The vast majority of American households are not stranded in a food desert. Yet packaged/prepared food consumption is high and has been rising for a long time.

    Apart from the partial explanation provided by food deserts and people who literally don't have access to cooking equipment, there are a few other factors that seem plausible major contributors. One is advertising for packaged and fast foods targeted at children. Another is taste preference formation: if the advertising achieves its aims, children will regularly be exposed to and form an early preference for relatively bland, fatty/sweet/salty foods such that they'll continue seeking those experiences later. The consumption of (e.g.) fried, battered chicken nuggets dipped in syrupy-sweet sauce is not some culturally invariant biological imperative that fast food companies just ran with. Recent immigrants and visitors often find such fare repulsive. It takes effort to shape tastes toward what's profitable for food product manufacturers, but once that's done buyers often pass over a bag of potatoes for a bag of fries, even if the former are in the very same store and less expensive.


    Agreed. I'd add a few other points causing issues with adopting healthy diets:

    --bitter is generally an acquired taste, and it's not being acquired so often these days. I was surprised to find my niece only drinks coffee when it's drowned in cream, sugar, and flavoring in one of those calorie bombs from a major coffee chain. People who can't stand bitter miss on many of the most nutritious greens, red wine, dark chocolate, and so on.

    --re: immigrants. Meat with every meal can be a symbol of how far they've come from their homeland (this is from observing close friends). Asking them to eat something more like their old "peasant" food may raise issues, especially if they are deeply invested in being "successful Americans." This is not meant as a slur on anyone, simply an observation. It's hard to find out that the American Dream has hidden traps, especially if you've truly sacrificed to get here.

    --similarly, greasy, calorie laden food may be what you share with friends. Choosing between being antisocial/healthy eating and social acceptance/bad diet is a nasty problem.

    --Food quantity is often a symbol of generosity. A kale stir-fry on potatoes is healthier than five pounds of deep-fried cheap meat with alcohol, but the meat is often a better symbol of generosity. Hosting feasts is an ancient form of showing off one's status and ability as a provider. It's also bad if it happens every day, and worse if it's subtly encouraged by mass marketing. Healthy food may even be seen as a criticism, as it implies that the person receiving it needs their vitamins, or some such.

    Obviously, all these forces can be overcome, but without a good basis in education (ideally from parents), it's hard to make the switch.


    Apart from the partial explanation provided by food deserts and people who literally don't have access to cooking equipment, there are a few other factors that seem plausible major contributors. One is advertising for packaged and fast foods targeted at children. Another is taste preference formation: if the advertising achieves its aims, children will regularly be exposed to and form an early preference for relatively bland, fatty/sweet/salty foods such that they'll continue seeking those experiences later.

    Not to agree with the guy with the insanely long handle but people can just not go to McD's or it's clones every time their kids want something. We didn't. For us, once a week would have been a frequent visit. We went maybe once or twice a month when they were young. Many of our friends didn't either. And for the most part our kids grew up not over weight. We also encouranged them into sports. Summer swimming at the local pool. Baseball, soccer, lacrosse, etc... And let them pick out any cereal they wanted to eat as long it had less than 10 grams per serving. (More to make them learn to read the labels than any nutritional reason. And learn they did.)

    My point is advertising is just that advertising. People can ignore it. Or they can be slaves to keeping up with the latest whatever. Their choice.

    And I noticed that most of the kids that grew up with ours and are on the plus size of the population had parents who "gave in" a lot or were themselves on the large size.

    Sorry but I don't blame McD's for all the diet issues of the country. Any more than I blame social liberals for destroying the Leave it to Beaver world we used to live in without any social problems what so ever.

    People make choices. Choices have consequences.


    Good points about the social constraints on meals. In my social circles serving light meatless meals would never be a slur on host or guests. That goes back to my earlier comment about the structural advantages I had even when I was, strictly by the numbers, living near the federal poverty threshold.

    If you cook at home, you can of course also prepare lots of meat, onion rings, waffles swimming in syrup, etc. Home cooking is not a panacea. But if you never cook, the the false dichotomy I highlighted becomes real: you pretty much have to choose between good (or at least "better"), expensive prepared foods and inexpensive junk.



    This is not the "junk food is killing us all" discussion.

    Discussions of food preparation are derailing.

    Those of you who are into that discussion get one more comment on the subject. Then I'll have to assume you're deliberately derailing the "What am I missing" thread and ban you.



    But I wonder what diets will look like in the future. I just hope we're not headed for a future as depicted in WALL-E. Only slightly worse will be you swallow a tiny bot or bots that allow you to eat as much junk as you want but it cleans it from your system before your body absorbs the sugars and other crap. And a deluxe model that transforms whatever you eat into useful molecules. Of course it will be priced at the equivalent of $300 per day, $1000 per day for the deluxe, and there will then be calls to subsidize it for the poor untill a "generic" shows up. All accompanied by a few 1000 lawsuits.

    I personally think that much of the diet issue in the USA would resolve itself with the elimination of corn and sugar subsidies. When / if this happens the future for food will definitely change. I suspect that a lot of innovation in the future of diets is being surpressed by these subsidies.


    A possible weird development: Machine Precognition, named in the tradition of machine vision, machine learning, etc. In the general case, predicting the future is absurdly difficult. Predicting the next several seconds in spatially limited domains is a much more heavily constrained problem, possibly constrained enough to yield useful results ranging somewhere from the merely better-than-guessing to truly eerie foreknowledge. We've already seen a foretaste of this with casino-fleecers hiding computers on their person so that they can run physics models of the roulette wheel and beat chance after the ball's in motion. But humans actually "decide" to do things before they're conscious of it. What if you had enough portable sensors, computation, and training data to predict what people are going to do before it happens, at least in limited ways and in the short term, without fMRI or other bulky/obvious diagnostic equipment? You might also add background information to make the model more specific: predict this man's near-future actions based on pupil measurements, breathing, subsurface blood flow, AND his past Twitter posts and credit score...

    You don't have to limit the modeling to humans, either. Combine physics models and sensor fusion so that you can 1) take a picture* of the immediate physical environment 2) evolve the scenario several seconds forward in time, according to physical laws and public facts about objects ("a Toyota Camry traveling at that speed takes about X meters to stop on wet pavement of approximately this texture") and their interaction with people. You could also add a step 3, scenario exploration, like "run the simulation again assuming that Person 3 in the scene is going to be punched at time t + 1 by Person 1."

    I don't know if it can be done, but there is a certain intuitive appeal to its possibility. Machine learning, given enough data, can be eerily good at predictions and correlations. Humans generally aren't as spontaneous or unpredictable as they think they are. All organisms and machines are constrained by physics in important ways, it's just that humans can't read the physics of every object in view at a glance the way they can with (e.g.) a single baseball's trajectory. There's going to be a wealth of portable sensors, not just logging data for later retrieval but providing input to predictive models. Even future super-phones may not be powerful enough to do it all alone but they could handle a lot of low-latency data preprocessing before sending data back to an oracle in the cloud.

    As for applications, of course there's enforcing security and safety, defeating security measures, gaining an edge in playing sports or betting on them, predicting others' moves in political or business negotiations, and probably a lot of much more specific or ambitious things.

    *Not necessarily just an optical image, of course; the "picture" is understood to include all relevant sensor data.


    how many transistors does it take to model a single synapse?

    Because if it were one (which I doubt, I suspect the number is much higher than one), but giving the computer the benefit of the doubt, it would still take 2,000 of your 40 billion transistor cpu's to model a normal human adult brain.


    If you are short of money: why is there a case of beer in your house?

    I don't like alcohol and it won't go with my meds.

    is your PPV bill higher than my entire cable bill?

    I had to think what that was, but I don't have a PPV bill.

    does everyone in your house have smart phones?

    Nobody here has a smart phone.

    do you have the turbo option on your internet package?


    do you have 4 TVs in your house?

    I have one, but sometimes I watch shows on the computer.

    do you have a really nice ride?

    I have a 25-year-old minivan.

    do you go to first run movies several times a week?

    No, I put good movies on the bottom of my Netflix queue which is five years away.

    are your kids wearing Air Jordans?

    I don't have kids, but if those are expensive, they probably wouldn't have them.


    When I was growing up, we didn't have McDonalds, we had A&Ws and went a couple of months. We did have picnics, and my father made me use plastic plates, cups, and silverware. Now, I'd do that myself, but having to wash even plastic stuff drove me nuts back then.

    Hmmm, the closest A&W now is about 45 minutes away.


    But if you never cook, the the false dichotomy I highlighted becomes real: you pretty much have to choose between good (or at least "better"), expensive prepared foods and inexpensive junk.

    You should look at frozen microwavable meals. I use a lot of those, since I can't cook anymore. Many of them have changed to be quite healthy. Healthy enough that I have to eat two a day to get near the required minimum calories of 1000 and the required 50 grams of protein. They aren't all that expensive, particularly if you buy on sale. A lot are then about $2. I am usually up to boiling -- so rice and noodles -- but that's pretty much it.


    I think the concerns about state control and oversight in a ubiquitous computing environment are valid but miss a central aspect of technological history. Repurposing and the impossibility of keeping secrets. Like a digital copy of a song, once a copy is in the world it is permanently out there - all the more so in a world of ubiquitous computing.

    So the military creates a swarm of hornets that can attack an area. One disaffected soldier/techie leaks some of the code to the world and everyone who wants to can make them - if not a swarm then a single wasp-like weapon platform. A nightmare for the Secret Service.

    Historically new technology has been very appealing to the rich and powerful. And it has worked well at keeping them in power, for awhile. And then it has worked well at supplanting them.

    Armour was cutting edge military technology, controlled and used by elites only until guns came along. Various elites thought they could use/capitalize on guns while still keeping their existing social order. It worked for awhile, but sooner or later someone realizes that they can point the gun at Lord Fauntleroy.

    Technological advances are appealing to the powerful, but unintended consequences are inevitable - good news for the rest of us (mostly). Also good fodder for stories.

    216: 211 thro 215 inc.

    Marilee, I think that mostly proves that you're not the stereotypical "idle poor" that the quotes you've addressed are supposed to apply to.


    Take a look at Hannu Rajaniemi's "Deus Ex Homina" (see "Year's Best," 2006), which took place in a world where the source code for recursively self-improving AI was leaked.


    Coming late to the party but this news article caught my attention;

    Apparently there's a growing trend for people to inconspicuously take pictures of other commuters with their smartphones before uploading the picture to various websites there fashion sense and looks are rated. One website,, encourages people to take pictures of people they find attractive.

    Lifelogging would really boost trends like this, as weird as it sounds it might become the norm to find your picture uploaded to dozens of blogs, rating sites and other places on the net every day.



    IBM talked about that at last week. Here it is on EETIMES:

    A search for 'IBM demos cognitive computer chips' will probably bring up various other sites.

    'The eventual goal is to create a brain-like 10 billion neuron, 100 trillion synapse cognitive computer with comparable size and power consumption to the human brain.'

    256 neurons = 4mm2 ='a few million transistors' on a 45nm process.

    These chips have 'learned to beat humans at pong'




    Thanks Graham.

    Looks like you have to teach that IBM array, too. It might take 20 years to program it.


    Here is a more interesting question: what would a secret agency do if they discovered technology that allowed them to process information much faster than the rest of the world?


    They would win WW2.


    Watching the weather channel talk about how the east coast is about to be hammered....

    Will we ever get much better weather prediction more than a few days out or do the computational requirements go up as a power function that computers will never catch?

    Will we be able to dissipate tornado and hurricane energy in a way that's less destructive than the natural event?

    Both of these will require huge computational increases and likely changes to the way we approach the problem.

    And the folks not in the US don't understand how bad this current storm might be, the NC coast may change in non trivial ways and Philidelphia, NYC, Boston, and other points may get to see how well they handle 80+ MPH winds continuous for hours.


    Sorry but I didn't finish the thought. If we are all life blogging or have monitoring "everywhere" can that be fed into a system of better weather prediction and/or control. If course coverage would be very uneven if tied to people especially if current trends towards urban life continue. And it doesn't help much with data collection above the ground surface.


    I think that beyond a certain limit, it's physically impossible to predict what the weather will be like, even witht he best of all possible computers. Very small differences in initial conditions lead to very large differences in later weathers, and as it's not just impractical but impossible to measure the position, velocity etc of every atom in the atmosphere and the world's surface, so it's impossible to say with any accuracy what the weather will be like in, say, 3 months time. We may be close to the limit of predictive accuracy at the moment.


    We may be close to the limit of predictive accuracy at the moment.

    Yes. I wonder if our current 7 day / 15 day forecasts are about as good as we'll get. Maybe better within 7 to 15 days but no way to extend the horizon.


    In the summer, I read three forecasts at least once a day. I'm not supposed to be out when it's 90F or higher and/or Code Orange or more polluted. It's interesting to see how they change. For example, for a while, this week was all 80s, but then yesterday was supposed to be in the 90s. What changed? I don't know, but I slept through the day anyway.

    I don't think that you can accurately forecast a hurricane or typhoon sooner than a few days. Even then, there's a cone where it might go. At this point, it looks like the Chesapeake Bay and Potomac River are in High, and Manassas, where I live, is in Moderate. I'll still bring the chair and table in off the porch and I've already filled all my empty tupperware with water. I also have windup flashlights/radios plus a windup phone charger.


    Yes, the definition of a chaotic system is one whose uncertainty (potential error in computing a prediction) increases exponentially. So to get the same accuracy at a one month forecast as we currently do in a five day forecast (about 60% accuracy out at the end, IIRC), we'd need to multiply the accuracy of the data (and of the models we're computing) by a factor of 1 million. I don't think that's likely until we have nanosensors on every square centimeter of Earth's surface, land and sea, and enough computing power to back that up. It's way too late at night for me to attempt to calculate how many zettaflops that would be, but I'm betting we're talking about turning a large part of the Earth's surface into processors and I/O circuitry. Won't matter much what the weather's going to be if we've used up all the room for predicting it.

    I don't think the assumed scaling is reasonable, and a bundle of other effects cut in way before the scale Charlie talks about.

    You're assuming that the circuits have to continue to work the same way, only smaller; this isn't a safe assumption at all. Below some minimum size which we'll be hitting in the next few Moore's Law generations we'll have to switch to using something other than electron currents for logic and stored charge for memory. The number of electrons involved in the flipping of a state or the storage of one bit will be so small that quantum effects will cause unacceptabe numbers of errors unless a large percentage of the devices are committed to error detection and correction. The most likely replacement at this point is electron spin ("spintronics", a rather ugly word, but we seem to be stuck with it), but there may be other possibilities later (high-temperature Josephson junctions or topological braids or other exotic devices).

    230: 223, #225 et seq ref weather prediction. Living in the UK, and having both basic qualifications for and actual experience as an amateur forecaster (eg, being able to actually read synoptic charts) I don't see us ever getting reliable forecasts much past 5 days except in places where stable weather systems and a body of historical records allow us to say that there's a less than 5% chance of a new weather system forming over $period. It's not uncommon to be able to see weather systems forming off the American seaboard, and tracking across the Atlantic over the course of 4 days.

    Reflective coatings don't work against lasers. The coating reflects a portion of the laser's energy but not all of it, so the first pulse of the laser damages the coating and allows the rest of the pulses to burn through. Mass is very effective against lasers though, particularly ablative carbon compounds.

    Some people have said that this future looks like an animistic magical world. That's one possibility. The other is Star Trek: TOS. Everything is easy to operate using very simple interfaces. Objects which appear to be under human control can in fact operate for long periods without human supervision; Humans provide strategic direction and outside-the-box decision making (The Enterprise could carry out its mission for half a decade without personnel). Visually simple tools that provide precise short-range control of various EM frequencies for apparently magical effects (Surgical protoplasers, energy weapons, etc). Tiny translators, so that everyone appears to be speaking the same language. Blank white walls in your quarters that can be set to any interactive display you want (OK, never actually shown on the show but stated in a lot of novels).

    From a human perspective - Everything WORKS, and it's easy to operate, and you're never outside the reach of your tools (Unless the writer needs to inject a bit of drama into the situation).


    Ok, I've skimmed over the last half of the comments - excuse any repetition (But a lot comments seem to center around cheap computing power, not ubiquitous computing)

    brainstorming: Sticking with the question of ridiculuosly cheap, smart CPUS/Computers - right now, smartphones are gadgets that carry a lot of identification details, and offer personalized services/apps. Most people only have one. The cheaper tiny computers get, the easier you can have account separation by having separate physical devices for that function as business cards/augmented reality/phone/target for advertising/interface to XXXcontacts site/games

    Bitcoin account management in a physical coin (You can look up the value of the coin on the coin and on bitcoin servers, where all transsactions by this coin are stored ... but not how often the coin itself changed hands.)

    CNC tools everywhere - can opener, screwdriver, and so on with a smart component

    Metro ticket that tells you the qickest way to dest, incl. delay

    postal stamps that do bookkeeping and routing (for trasnferal between posting systems that use different payment schemes) - you load some digital money onto your parcel and give it to the first carrier, from there on tha parcel books it's own tours based on how much money it has, speed s of the carrieres, customs etc.

    With computers everywhere, you need less network/internet: the first example (where does my egg come from?) is just a bookkeeping/database issue. In germany, theoretically you can tale the number printed on an egg and look up where it's from and so on. With microcomputers, this could work without a central database.

    Data separation between different devices, eg storing your passphrases/access codes that you need with you phone in an earring, and you tans in a tooth

    Generally everything that's a ticket, bill of cash, coin, token, stamp can be a pseudonymous interface to the world.

    This is'nt very concise, sorry, but actuall I'm pleased to say that some of them are probably not boring.


    Against lasers, wood is rather effective. It chars to carbon then gets very difficult to cut unless there is a flow of oxygen to burn it away. I once had a long discussion on sci.military.moderated on the potential ineffectiveness of anti-artillery lasers using just such cheap countermeasures


    Cutting edge is about to become 12nm, so conservatively at least a factor of 10 in areal density (more if Intel's 3D structure is used) if modern tech is available. A human brain complexity neural computer within this decade looks very feasible.


    It may be possible to make active spectacles for those of us for whom varifocals are at present the best substitute for younger eyes. That combined with augmented reality and pattern recognition (which would highlight recognised items in the visual field) would help us find whatever we lost.

    Car keys, house keys, all that kind of thing that is easy to lose (if we still have them in around 10 years).


    If the processors can work fast enough - and if so it will probably be in bursts, after which the cores in use will need to cool, it may be possible to take photographs by analysing the incident phase, frequency, and amplitude of EM waves in the visible spectrum (or wider) to store image information from which the whole scene can be rendered, in focus, down to the pixel size of the sensor.

    That would make glass lenses obsolete.

    Something like that would make a great sensor element for the active spectacles.

    BUT there would be some tricky software to write. Maybe I need to work on my C skills?


    There's already glasses like that, but very expensive. I can't remember the name and am not using good search terms.


    "genocide of an order not previously envisioned"

    ... except by Heinlein in his novel Sixth Column.


    I just finished reading Ben Hammersley's speech to the IAAC. He absolutely nailed something for me in that speech and I think it's interesting.

    Charlie's talk touched on how much more data could be collected, stored, and streamed from each of us. Ben's speech touched on how society feels that this exchange should be reciprocal. If we suppose that the prediction of wireless bandwidth holds, we have the capacity for not just our own data to go out, but for other data to come in.

    Is every store not just tracking how many people go in and out, like they do today, but also making that data available? Can the store and my personal devices work me into a queue (completely virtual in the sense that I'm not actually in line), so that I arrive at the store about the time that a sales person can assist me?

    Is my phone letting me know as I'm walking from work to the train that the police are present at a scuffle along my usual route, a crowd is forming, and I should walk around it? Does it buzz when I'm leaving the bar at night to let me know I'm entering a higher crime neighbourhood or headed the wrong direction?

    Are the buses in my city tagged so I can follow them on Google Maps to see if they're behind schedule? Do they include data about the number of passengers, seats available, etc so I know if the bus will be crowded or not?

    What data is my municipality sharing with my personal network? What is my provincial/state gov't or my federal/national gov't sharing? Forget lifeblogging me, are my services lifeblogged and available for sharing and remixing?



    About this Entry

    This page contains a single entry by Charlie Stross published on August 20, 2011 3:21 PM.

    Brief announcement was the previous entry in this blog.

    Why I'm not on Google Plus is the next entry in this blog.

    Find recent content on the main index or look in the archives to find all content.

    Search this blog