So, one other aspect: last year, Arizona decided to woo tech companies by saying, essentially, "We got rid of all those pesky regulations that you have in California, so you should come test your autonomous cars here! It'll be great!"

Uber leaped at this.

"because the culprit in this case is a corporation"

Do we know whether that is true? Legally I mean.

If it does turn out to be the car side that caused the fatality is it Uber the corporation, or the test driver behind the wheel, that's the one legally responsible?

Why do your self-driving cars have drivers/remote operators? Why are you making that a prerequisite? You're assuming the UK won't allow fully autonomous cabs?

[ Deleted by moderator — derailing and irrelevant to the exam question ]

Re: 'Do we know whether that is true? Legally I mean.'

Hmmm ... looking at how US fracking corps operate, I'm guessing a big YES!

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

Why do your self-driving cars have drivers/remote operators? Why are you making that a prerequisite? You're assuming the UK won't allow fully autonomous cabs?

NOTE

I am extremely disappointed that none of you have approached the exam question from a legal liability perspective — instead, I'm seeing peripheral questions and a derailing rant here!

Come on, folks, argue the case!

For the law student exam questions:

Mary places adversarial image patches purchased from Etsy on her kid's clothing to prevent them being picked up by automated image/face tracking by her abusive ex-partner.

When the children legally cross the road on a pedestrian crossing, with enough time for cars to safely stop in time, said patches cause them to be misclassified as banana lying on the road by the visual recognition layer, while the LIDAR system correctly classified them as humans. However due to a race condition in the collision avoidance layer the LIDAR signal got ignored, causing the deaths of the children in a collision.

Discuss.

Ooh, I like it!

I am afraid that despite my username the future looks grim on this one. Depending on how the algorithms and collision avoidance methods are implemented it is probable that any data kept by the car, and which will likely be admissible, will only lead to exonerate the corp, so I will join the Cynic in claiming the pedestrians will be made to be always at fault, and it will take a very lucky set of circumstances to allow a different story to emerge.

It is a "constructivist" approach, but imagine this: if the car senses the issue and documents it and is supposed to be able to avoid it, it will. If it does not, it either was not programmed to or failed. Then, it is likely that whatever situation appeared was one that the sensors did not reigster as dangerous, and therefor will allow them to get away with it as outside the scope (and it is likely that engineers do not think of anything and some simplifications kill).

I think LLCs will argue "not at fault" always, and will be backed by the car recordings, which are biased to always not be at fault by design.

This is of course beyond policy, lobby, and so forth. Just the simple way we build these systems make them almost litigation proof, even if they encode unconscious biases. It is the reason I choose not to work on software that can kill people. My bugs only kill compute-time...

To argue the case: the long term result of the frist scenario is regulation to ensure that security features are mandatory and must be always paid up to the highest available standard on the end-user's dime or the car doesn't work. For the second scenario the creator of the patches is branded a dangerous terrorist and sentenced to life in jail, the mother gets to serve time as well, and using privacy stickers becomes legally worse than going into a bank in a ski mask.

For a start there are lot of unrealistic claims about the safety and reliability of the autonomous cars. At some point, (probably after a spate of robo hit-and-runs) there will have to be an independent testing and regulatory agency set up to test those claims. Uber being Randian wingnuts will fight it tooth and nail.

"2. Security — to ensure payment, prevent fraud, fine passengers who boak on the back seat, etc."

Speaking as an ex-cab driver customers will do that even with a driver present. Cashless phone payments will reduce the number of runners, but people being drunk idiots will still find inventive ways to soil and vandalize cars. Welcome to Portaloo lottery where a randomly selected robocab has a steaming prize hidden somewhere inside.

And speaking of Uber's business model, I don't know how it can continue. They hemorrhaged $2.8 billion last year, and that's with a business model that convinces their workers to pay for all their capital and maintenance costs. If they succeed with inventing a reliable robocar then they either have to buy and maintain fleets of them for their service, or convince the working poor to buy them. They've already tried predatory loan scheme and it didn't work out that well.

The USSO-of-fucking-A's stupid "Jaywalikng" lwas are relevant here - the victim was not crossing at a designated crosswalk IIRC. Therefore the corporations are going to win AGAIN ... UNLESS .. someone finally brings, as part of the lawsuits the proposition that Jaywalking is contrary to ( Insert $ClauseofUS_Constitution-HERE) & waits for the fun to start. Valid legal point, yes/no?

^^^^^^^^^^^^^^^^^^^^^^^^

Now to your test case .... Precedent: Quintinshill, the worst railway accident ever in Britain - just over the boprder into Scotland, but inquests were held in Carlisle ... Also the principal culprit was found, later to have been a mild epileptic - was let out of jail early & went back to work for the railway company ( WHo probably knew he was eplieptic & should not have kept him in that specific safety-critical job ...

The manufacturer of the vehicle, who has a contract with the holding company for ongoing maintenance, disabled the enhanced pedestrian avoidance feature for which the driver was no longer paying. The company is liable, on the spot at that point - they deliberately degraded the safety systems, no matter who is supposed to be paying ... this is one reason Uber is being attacked so hard in London @ present ...

The driver is monitoring their vehicle remotely from their phone, using a dash cam and an app provided by the vehicle manufacturer but subject to an EULA that disclaims responsibility and commits the driver to binding arbitration administered by a private tribunal based in Pyongyang acting in accordance with the legal code of the Republic of South Sudan. NOTY admissable in either an English or Scottish court - there must be a Mens Rea & controlling interest - such an obvious scam would be ruled immediately illegal. "Fun" though.

... the maximum penalty for vehicular homicide in Arizona is a mere three years in jail, which would be laughable if it wasn't so enraging. (Rob a bank and shoot a guard: get the death penalty. Run the guard over while they're off-shift: max three years.)

I don't think that's correct. Per this site, Arizona doesn't have separate "vehicular homicide" statutes, so killing someone with a car is handled under the general homicide laws, which means the crime could vary from "negligent homicide" (max three years in prison) to "manslaughter" (max twelve years) to "second-degree murder" (max 25 years, or life if the victim was under 12).

Also, doesn't the UK have something called "death by dangerous driving", which in the case of "no aggravating circumstances" has a maximum penalty of only two years in prison? (I will vaguely gesture in the direction of relevance by pointing out that this will presumably be part of the legal background for your test case.)

Sorry about that. It was attempting to explain why I don't think there WILL be much legal liability. I misunderstood what you wanted to discuss. But let's assume that there IS some criminal or civil liability to the victim's relatives.

As I understand English law, extrapolating from the comparable case with a human driver, and the original importation of USSR Ladas (surprisingly relevant) etc., the direct liability is solely on the driver and his insurance company.

The car and its configuration would be relevant only if it failed to meet roadworthiness regulations (which currently do not exist), just as with brake failure. But, even then, it would give rise only to criminal liability. Yes, the driver and insurance company would be able to create a case for civil liability against the holding and despatch companies, and and the holding company against the software owner, but ....

I don't see this situation as viable for self-driving cars, so I think that there WILL be legislation. I posted earlier what I think the legislation will be, and will not repeat.

Well... That's an interesting tangle.

Before we continue, I will assume that Scotland is still part of the UK, and that UK is not part of the EU (change either of those, and I suppose the results may change).

UK will start the process, based on the location of the killed pedestrian. At this point, it boils down to if the prosecution decide to go after the driver (because the driver is responsible for the actions of the car while the car is being supervised) or the firm owning the car (taxi dispatcher? third party?). Let us assume they decide to go after the driver, as that probably looks like the softer target.

Driver's legal advice will argue mere negligence due to momentary distraction, and suggesting that CPS takes this up with the Belgian company. CPS will decline and continue, asking for penalty, due to the willing use of a remote-control system known to have blocking pop-ups (they may find another driver as an expert witness if necessary). Case continues in British court.

Driver is found guilty of manslaughter, but receives no points on the driving license, due to the need of a license for actual working.

CPS is unhappy and goes after the dispatcher who sent the pick-up notice that (in their legal reasoning) caused the accident. A second case is started in Scotland, eventually reaches a "Not Proven" verdict.

Relatives of pedestrian then sues the car (or, the car's shell company) for emotional distress, perhaps loss of income, and assorted other things. After some legal wrangling, it is discovered that no tribunal following South Sudan jurisdiction can be stood up in Pyonyang, at which point this thrust is abandoned, then quickly re-opened in the home jurisdiction of the taxi "driver".

And at that point, I am starting to run out of plausible follow-ons.

A lot depends on the timing and sequence of critical events esp. during the first couple of years. Considering how the many school shootings have had zero impact on US legislation re: gun control, I'm guessing that any automated vehicle corps will be safe provided none of the early accident victims was a major Pol. Also, an early victim related to an insider (AI developer) could result in the public learning all sorts of interesting details as per today's headlines re: Wylie ex CambridgeAnalytica ratting out illegal usage of FB data to manipulate elections.)

NaiveOptimist: Re: 'patches ...' Nope - this is protected by the First Amendment (right to free speech)!

Greg: Re: '... & controlling interest - such an obvious scam would be ruled immediately illegal.' ... FYI - US fracking operates (almost) exclusively this way. And with Brexit underway, how do you think UK law will turn: toward greater personal protections a la EU, or toward greater corp protections a la US.

Can we request that this not end up getting turned into another boring trolly problem. The world does not need yet another inane "but what if the car has the choice of sacrificing a baby inside the car for 5 outside the car on the other track" type discussions.

such as a "gold package" that improves the accuracy of pedestrian/obstacle detection from 90% to 99.9%. Two months ago, because they'd never hit anyone, the driver downgraded from the "gold package" to a less-effective "silver package".

I can't see this going well for the manufacturer/software company and can't really see any company doing this for a safety feature any more than a car company is ever going to make the response time when pressing the brake pedal (already controlled by software in many modern cars) into a feature you pay extra for. "1000 bucks extra for your breaks to work without a superfluous .5 second delay!"

Lawyers would probably point to past cases like ford deliberately choosing to leave the known flaw with their fuel tanks or UK equivalents. only it would be worse because rolling out the fix would be effectively free and there would be clear paperwork showing the expected death rate for the 90% vs 99.9% and the profits they expect from the gold package.

If the car can legally drive on the roads on it's own then all the stuff with the app seems superfluous. If it can't legally drive on the roads on it's own then all the crap with phone apps seems superfluous since you can't guarantee signal and

The driver/owner might also get into trouble if it's viewed as choosing to remove safety features to save money muck like if there's an accident due to a wheel falling off because the driver failed to pay for basic maintainance... but I'm still thinking the company voluntarily choosing to turn off safety features would be the primary ones fucked because you just don't do that. If a car mechanic hands your car back after a servicing, having certified it safe for the road but either leaves half the nuts off or leaves a software package that will make the car unsafe X hours in the future they're typically on the hook.

Sorry, left off the end of a line in the second paragraph.

If it can't legally drive on the roads on it's own then all the crap with phone apps seems superfluous since you can't guarantee signal and presumably you'd have some level of responsibility for maintaining a reliable way of controlling the car that would be somehow laid out in laws or regulation.

Side note: Remote control cars on the public roads do not seem like something that's going to happen.

Honestly, I feel like the legal merits are going to be pretty irrelevant in this case. Public outrage will demand action, and going after the corporations involved is a nightmare of jurisdictions and would potentially be very disruptive to the economy. So the driver would be scapegoated, the death attributed to operator error, and maybe a few patches in the software to prevent the popup blocking the view, or something minor like that.

Also, doesn't the UK have something called "death by dangerous driving", which in the case of "no aggravating circumstances" has a maximum penalty of only two years in prison? (I will vaguely gesture in the direction of relevance by pointing out that this will presumably be part of the legal background for your test case.)

That's right, although (IANAL) I believe very egregious cases can be prosecuted as manslaughter. The reason why common law jurisdictions often have enacted these "lesser" vehicular homicide laws is that jury members, feeling "there but for the grace of God go I", have been very reluctant to convict killer drivers of manslaughter. It'd be interesting to know if jury-less jurisdictions have similar laws.

And as for the gold-vs-silver safety packages, I think regulators would not have much hesitation to mandate gold for every vehicle. Here in the US all new cars have to have back-up cameras as of a year or so ago, and I think there's some mandatory collision detection and prevention tech provisions coming soon too.

If I were representing any of the corporate interests in this, I would argue that liability rests solely on the human who chose to downgrade the quality of the software and then operated the vehicle in a for-hire fashion that allowed it access to a restricted area in semi-autonomous mode.

I suspect that this would turn out to be an illegal combination of things -- that for-hire operations will be required to run the safest version of the available software.

Strictly from a technical perspective, network latency and buffering for jitter compensation are likely to screw up any attempt at remote piloting.

And speaking of Uber's business model, I don't know how it can continue. They hemorrhaged $2.8 billion last year, and that's with a business model that convinces their workers to pay for all their capital and maintenance costs.

Uber is actually following Amazon's lead. Under price the competition to the point when you get a better cost system yourself (AWS and such) and they also just give up. Then you can price things to make money. At least in the product areas where you have obliterated the competition.

Uber's plan may have been all along to use actual drivers as a bridge to fully autonomous. Vehicle utilization rates go way up. The cars are designed for one purpose and one purpose only. Those two things lower the cost per mile to the point Uber can make a profit.

There is a new offence of causing death by careless driving, as the same thing had happened to death by dangerous driving, because the bar for the latter is so high! It is unlikely that the driver would be convicted of anything more and, for reasons already described, probably not even that.

https://www.motorists-lawyer.co.uk/content/helpsheets/causingdeathbycarelessorinconsideratedriving.rhtm

"And as for the gold-vs-silver safety packages, I think regulators would not have much hesitation to mandate gold for every vehicle."

More likely insurers would. But, if it were not, the downgrading would merely be the driver's responsibility. THAT aspect of English motoring law is pretty clear.

As a budding scientist many, many years ago I was always taught that when building a hypothesis, I should state my assumptions.

In this case, you are assuming that cars can be driven by a human, or driven by a machine, or crucially driven by both a human AND a machine.

I would at this point take issue with the last assumption, since research by Google on their driverless vehicles has demonstrated that whilst humans are pretty reasonable at driving a car, they are uniformly lousy at monitoring a car driven by a machine. Indeed, it is very well known that humans are passable at interesting tasks and pretty abysmal at boring ones.

So, and it pains me to suggest this, I rather think that you may be presenting us all with a straw man argument for the reasons of stimulating debate. Whilst you have used this driving mode in books of yours, research suggests that a remote driver who is doing anything at all like driving is a really, really bad idea.

If we posit then that the "remote driver" was a straw man argument and work on the basis of "What happens when an autonomous vehicle kills someone" then we are on much sounder ground. Parallels can be drawn between this and what happens when a railway train kills someone, the only difference being that the person killed would not be trespassing on a railway line.

Corporate manslaughter would seem to be the way to think here.

Dare I say it? Greg was somewhat right.

https://arstechnica.com/cars/2018/03/police-chief-uber-self-driving-car-likely-not-at-fault-in-fatal-crash/

Video already shows idiot stepped out into traffic in a low visible spot. Car might have been speeding by 3MPH but there is some evidence that the speed limit in the not too distance past was 7MPH more than what the car was doing and from personal experience I have wonder how well the speed limit signs are in that area.

I have been playing mind games with the entire issue of fault and liability for several years. My cars each cost me about $100 per month for insurance with 80% of that dealing with what my insurance company might have to pay out if I mess up. If autonomous cars get popular (a given I think) the for profit auto insurance business in the US will crater.

Corporate manslaughter caused by a malfunctioning machine.

This sort of law is fairly well worked out; the machine is supposed to avoid deaths, and it failed in this objective. The corporation is thus guilty of corporate manslaughter, and the only question at this point is where the levels of culpability for each party contributing to the accident lie.

The critical question, to me, is 'Would a pedestrian be dead if the car had been driven by a normal, licensed, human driver?' My guess is 'no', but IRL, I don't know.

Once we get into the full automation vehicle versus legal code intersection, immediate human judgment becomes less of a factor in the failure-path. I'm reminded of complex-system failure theory and all the horrible ways that can go wrong. And my goodness, the intersection of software development practices, the physics of multi-ton vehicles, consumer safety regulation, corporate and individual liability laws... that's a complex system with complex failure modes.

But this isn't a full-automation vehicle, exactly, it's a supervised one. The premise of the problem is set up so that the vehicular automation is primary, with a human having the ability to press an override button and be honored.

One of the things I've heard again and again is you can't arbitrate away criminal liability. Only legislation can do that, and it does it by defining 'criminal' in ways that allow delegation to non-state actors. In our dystopian corporatist future maybe, but not quite yet. The global nature of that arrangement won't be a shield for a county prosecutor looking to for who to punish. No, that arbitration arrangement is there to prevent the taxi-operator from getting any damages through suits.

Back to criminal. There are two phases of this incident.

• Assessment of criminal liability.
• Post-mortem analysis by the regulator to prevent this happening again.

I have a very strong suspicion that a vehicle able to be on the roads, with an occasional remote supervisor, will have robust authority to act in its own stead as a piece of approved consumer safety gear. A fatality in this case will be treated similar to a freight-train killing someone on the tracks. It has to be shown that the operator/supervisor was unable to stop it from happening, and at that point liability shifts. I believe you're right in that there won't be a single person hauled before a court to be told, "it's your fault, you get to go to jail". The grieving family will have no criminal liability to hold to and make themselves whole.

No, the interesting stuff happens during the regulatory post-mortem. Because that grieving family sure as hell will raise the matter with the appropriate regulators. Fatal accidents are like that. This is where the ineffective nature of the supervisor workflow will be analyzed, and decisions made to either make that more robust or to further deprecate it as an actual 'supervisor'. The ability of those devices to be sold will be assessed, and the decision of whether or not to withdraw such devices from the market will be made. The mandatory minimum safety automation will be assessed once more, and recommendations made to adjust those margins.

My airplane friends joke that for every regulation the US FAA has, there are deaths behind it. I imagine autonomous vehicles will be much the same. But we have to get through the slaughter first.

Which is why it won't be criminal liability that pains the perpetrators, it will be denial of access to the market that does it. Or adjustments to mandatory minimum features that affect the profit-line that does it. Or if a company racks up enough fatal accidents, a class-action suit against them will deny that LLC the right to exist and hurt the parent corporation in lost opportunity and marketshare. Or the insurance industry making certain features uneconomic to market. That is where criminal liability will go.

A sadly plausible scenario, and to paraphrase Stalin, "1 death to an autonomous vehicle is a tragedy, but the 30K+ automobile fatalities in the U.S. each year is a mere statistic and thus not worth worrying about". Of course, until we have many autonomous vehicles in operation, we won't know the actual failure rate, but I'm still betting it will be much lower than with human drivers.

The legal questions are, of course interesting. IANAL, but my take on this is that it's going to depend strongly on how far in the future we're speculating (e.g., 10 vs. 50 years), and just how Randian society becomes during that period. First, a simplifying assumption: irrespective of where the vehicle and its owner are registered, the lawsuit will be prosecuted in the jurisdiction where the crime occurs. Rationale: If I commit a crime while visiting my wife's family in Boston, my Canadian citizenship and Montreal residence are irrelevant. I'll be prosecuted and jailed in Boston.

Who is prosecuted for the crime? That will depend on how the legal definition of personhood evolves in coming years (see below). It seems likely that if corporations are treated as individuals, artificially intelligent entities will be too. That suggests the reductio ad absurdum scenario in which the autonomous vehicle is "jailed" (placed in an impound lot somewhere) and everyone else goes free.

Also, an important caveat: whenever lawyers are involved, the actual letter and intent of the law are less relevant than how pleasingly the lawyer argues the case and how sympathetic the judge is to that case. Legal objectivity? Don't make me laugh.

There are currently legal precedents in most jurisdiction for "piercing the corporate veil", and these precedents provide a subjective and difficult workaround for cases in which prosecutors attack the decisionmakers responsible for bad design or antisocial decisions rather than just attacking the company and leaving the perps unaffected. Given the lack of any significant prosecutions or jail time for any of the people responsible for the last 2 global banking crises, the jurisprudence (my favorite 1-word oxymoron) currently comes down heavily on the side of protecting criminals, sociopaths, and the 1%. If we continue down this road, things will get very Orwellian indeed. My optimistic side (using the Volkswagon emissions scandal as precedent) suggests a small but non-zero possibility that society will collectively hit a breaking point after which the masses are no longer willing to tolerate such protection for the privileged and force changes in the law.

Of course, such a backlash may go too far. The corporate veil was created (in part) as a legal fiction to protect shareholders and executives from the real and significant threat of unpredictable errors or impacts beyond their control (e.g., shareholders can't stop a rogue CEO from committing a crime; they can only depose the person post facto). Predictable errors are all those pesky things you ignore during beta-testing in the hope that they're edge cases and will quietly go away. Or when you ignore the advice of your engineers and unleash a dangerously unfinished product on the world. That wouldn't necessarily be a good thing.

My airplane friends joke that for every regulation the US FAA has, there are deaths behind it. I imagine autonomous vehicles will be much the same. But we have to get through the slaughter first.

And as a side note, my wife works for a major airline and at times we've asked the question of each other; "If you die in an airline crash, why does someone in your extended family get to become a millionaire?

What most people are missing is that this is the UK, which is considerably more 'laissez faire' than the USA. From the Corporate Manslaughter and Corporate Homicide Act 2007, the relevant rule is:

2 Meaning of “relevant duty of care” (5) For the purposes of this Act, whether a particular organisation owes a duty of care to a particular individual is a question of law. The judge must make any findings of fact necessary to decide that question.

Well, under English law, why should the owner of the software owe a duty of care to someone with whom it has no dealings? This is EXACTLY the same as the liability of the motor companies that sold vehicles with ineffective brakes, for which there is precedent. The simple answer is that it doesn't have any though, if I recall, EU law constrains that somewhat. Indeed, even the owner-drivers had a problem proving the vendor was liable to them - 'caveat emptor' and all that.

The real problem here is that the operator, by degrading his safety software to the "Silver" package, invalidated his insurance policy, and thus has no coverage and no legal representation.

More to the point, I doubt that the scenario as written could ever occur. The problem here being that insurance companies hate to pay out, and they will mandate, probably through an act of Congress (or the relevant state legislatures, at least in the U.S.*) that everyone carry the Gold package, and that cars not carrying a recent version of the Gold package never start in the first place.

• Insurance companies in other countries will make similar arrangements with that county's political/regulatory system.

My son was interested in do Uber and/or Lyft. Insurance was the stopper for him. We have insurance from a major consumer company. To date (in our state) they will not cover such use of your car. And there are options that might work but most of them only cover you while transporting people. There is a gap where the major consumer companies say their coverage stops when you mark yourself available with the app and the rider coverage you buy only starts when you have accepted a fare. My son knows someone who lost his car in an accident and coverage was denied due to this "hole".

In Canada the standard for proving criminal negligence are fairly high:

Bawden explained that the duration of the lapse in attention was key to his decision. According to the Crown’s expert witness, traffic officer Dawn Mutis, the evidence shows a provable period of inattention of between 0.74 seconds and 1.18 seconds, Bawden said. “On the evidence of the Crown’s own expert witness,” Bawden said, “this would be a miniscule period of inattention.” Even a longer attention lapse alleged by Crown lawyer Scott Pattison, Bawden said, would still mean a total period of inattention of less than two seconds.

This, Bawden said, qualifies as a “momentary lapse of attention” of a kind deemed non-criminal by the Supreme Court.

https://www.thestar.com/news/gta/2017/11/21/not-guilty-verdict-for-driver-who-mounted-the-sidewalk-and-killed-pedestrian-on-dundas-st.html

So assuming UK law is close to Canadian law (not necessarily a valid assumption) then how long the popup was present before the driver could dismiss/minimize it will be a key factor. A couple of seconds or less and the driver is off the hook (criminally — civil liability may apply).

Yes.

Administrative notice: Trolley problem discussions on this thread are banned and will be deleted hereafter.

Folks obsessed with the Trolley Problem are kindly reminded that it's been a topic in SF since (at least) The Cold Equations by Tom Godwin, and is often critiqued in terms of gross engineering negligence. Taken as a philosophical thought-experiment the standard Trolley Problem frames self-driving worst-case outcomes as a negative-sum interaction that can be dealt with in purely utilitarian terms — an approach which has little relevance to actual situations likely to occur (consider how applicable it is to the law problem outlined above).

To me, the interesting question currently isn't whether fatal accidents will happen (they will), but where does liability get apportioned. Historically, vehicle manufacturers were effectively unitary entities, vehicle drivers were another, and the victim was (usually) a third. The arrival of AI and lots of different chunks of software changes this, as do the (arguably illegal) software licensing shrinkwraps ...

It will be interesting to see how much weight the court gives to the relative hazards of a human driven vehicle vs an automated vehicle. HDV kill many people each day, which means nothing in itself. Does the death of one person (I am not specifically referring to this incident) invalidate the whole concept?

For example, should an AV be able to anticipate situations better than a HDV? If so, how would one judge those situations.

It is relatively easy to grasp the liability where an HDV hits a pedestrian since well established questions exist such as - was the human driver qualified to drive the vehicle - did the human driver have a known medical condition which contributed to the accident - was the human driver doing something personally reckless - using cell phone, snorting cocaine,... - was the driver driving in a reckless fashion - speeding, reversing down a one way street (I live on one and, yes, many people do) - was the condition of the vehicle likely to have contributed to the incident (no brakes, no mirrors, ...)

It's difficult to see how some of these questions translate. One could measure the tyre tracks to see how the car braked and from that establish speed. One could read the logs (should i mention VW falsifying things?). One could examine the vehicle for faults.

Assuming that the ADV passed all of the above tests that leaves the final test of the vehicle doing something that it should reasonably have been expected to avoid. This is almost impossible to prove without detailed information from the scene at the time. Eyewitnesses would not be enough - I would imagine that the lawyers would argue the fallibility of such evidence.

I think this is the key - how would one establish the legal test for "doing something that it should reasonably been expected to avoid"? It's an important question since it is a test that is likely to be increasingly used.

I suspect the closest analogs will be aircraft accidents, where clear precedence have been established to judge liability, and industrial accidents. At least one person has been killed by a robot so some judgment must have been made.

I always felt that the issue with ADV was legal, not technical.

I haven't even addressed the liability of the supervisory driver. I think right now that many of the ADV companies are trying to have their cake and eat it. The supervisory driver remains liable even when the basis of the ADV is that the supervisory driver is less likely to be focussed on the road.

I would image that m'learned friends (or whatever they are called now - somewhat educated mates perhaps) are all agog to see how much money they will make from this subject.

My airplane friends joke that for every regulation the US FAA has, there are deaths behind it And where do you think Britain's railway safey regulations came from? Many painstaking years of work by HMRI & now, RAIB as well .... Borne fruit though - you are safer in a Brit train than you are in your own home .... & certainly than "out on the street" ....

"To me, the interesting question currently isn't whether fatal accidents will happen (they will), but where does liability get apportioned."

In the UK, I don't see that as a legal one, but a political one. The situation is quite clear for non-highway AI, including drones, and all vehicular operation on the highway - any liability to the public is on the operator alone. If the law were changed simply to allow remotely control or even autonomous vehicles on the highway, the default is that would follow. As I said, I don't think that is viable, but the question is what the legislation would do about it, if anything.

At fault - with rail accidents, it seems it's whoever is in the driver's seat. The driver is at fault even though his employer requires him to work 16-hour days with no breaks. Corps have many ways of weaseling out of liability and 'expensive' labor practices.

Re: ' .. all self-driving cars will be owned by limited liability shell companies'

Seems some of the commenters are ignoring this bit. If you have only 2, 3 or 4 players in the self-driving car industry, good chance that at least of a couple of 'market wars' some of which could be AI-vs-AI with few people ever learning about them. OTOH, if a competitor's car killed a pedestrian while the industry is still young, good chance that the industry would be more interested in their continued existence (therefore call a truce on market warfare) and collude to hide any evidence of inherent industry-wide problems.

Also - if all cars are owned by a small cartel whose members would have figured out the minimum number of vehicles they needed to keep on the roads in order to maximize profit/minimize overhead & operating costs, this would reduce total vehicles on the road substantially. Wouldn't such a reduction in cars on the road upset the automotive industry and all the sectors it feeds, including auto insurance? (Motive)

Following on this - if all vehicles are AI/remotely operated, used only on an as-needed basis, then there would be fewer vehicles on the roads, period. This should in itself reduce the overall accident rate.

Pre-existing tech integration - I've seen toddlers using/playing with smartphones and believe that smartphone usage will be universal pretty soon. Given the existing range of apps already available, there's no reason not to have car-approaching apps on smartphones. This means that at the very least there would be 2 active traffic alert systems operating: the car and the pedestrian.

Question: If all cars become AI/remotely operated, seems like a ton of additional data flow and storage, signal strength, etc. Could this bugger up how other industries operate? (Have no idea how much data would be required, nor whether a ranking system of who gets first dibs on data flow: cars with passengers vs. stock exchanges.)

Assuming sane design, autonomous cars will have a very low network presence indeed. The only true defense against random malware is an air-gap, so their only network connection will be the extremely heavily encrypted link to their manufacturer which is used for map updates and patches.

By extremely heavily encrypted I mean "One time pad". - the car will not rely on the internet to navigate the world, it will rely on sensors that interact with the physical world, and eventually those are going to be a lot harder to spoof than the mark 1 human eyeball.

Charlie notes: "To me, the interesting question currently isn't whether fatal accidents will happen (they will), but where does liability get apportioned."

In theory, a sine qua non approach should work: rigorously identify (ideally by expert witnesses rather than lawyers) what the crucial step is that led to the accident, without which the accident could not have occurred. For example, if a bug is discovered in the driver software and the software developer chooses not to fix it (cough, cough internet of things, cough cough), the manager at the developer who made the decision is at fault. If the fix is distributed and the owner of the vehicle is alerted to the fix and chooses not to install it, the owner is at fault. If the software is known to not prevent accidents under specific circumstances, and a pedestrian who can be shown to have researched those conditions and deliberately steps in front of the vehicle under those circumstances to trigger a lawsuit that will enrich them, then the pedestrian is at fault. And so on.

Of course, that's how the legal system is supposed to work, and most of us can cite examples of how it failed to work in that manner. And reality is always more complex than such a simple formulation as mine can encapsulate. But it's nice to think that some future system will be ruled on a scrupulous examination of the facts and interpretations based on expert advice.

"Assuming sane design, autonomous cars will have a very low network presence indeed."

Given current practices - both with cars and with "things" in general - I would bet both balls and the bag they hang in that exactly the opposite will be the case.

"At fault - with rail accidents, it seems it's whoever is in the driver's seat. The driver is at fault even though his employer requires him to work 16-hour days with no breaks."

This is about the UK, and the situation is entirely otherwise here. Accidents are thoroughly investigated to determine what went wrong with the intention of making sure it doesn't happen again. The process isn't perfect, but it does not automatically assume that any individual is "to blame". (In your example, the employer would be to blame for breaching limits on working hours and rest breaks, although the unions would never let it get that bad to begin with.)

NOT EVEN WRONG I suggest you read some recent RAIB reports on fatal accidents .... [ Not that there are very many nowadays, & most of those are fuckwits who "think" that trying to beat a train across a level-crossing was a good idea (!) [ SPLAT ]

So on airgapping cars, it takes a lot of motivation to get an industry to do anything.

Remember the jeep hack from 2015? They patched that, but didn't pull out the internet connection. "Can probably crash the car remotely" was not enough motivation to make the manufacturer pull the internet connection. I don't see it happening; People are too invested in buzzwords like "connected" and "smart".

We can dream though.

....what the crucial step is that led to the accident, STOP RIGHT THERE Usually false assumption - that there was one single point-of-failure that caused the accident. Whereas, usually there are severeal & each & evry one has to fall over, in the "right" sequence before a whoopsie occurs.

The classic case which I quote is the Hull Paragon station rail crash of 1927 Wiki article here - which is ALSO WRONG. The signalman did NOT set the wrong points - it should not have been possible for him to do it .... The elctromechanical interlocking should have prevented it. After-crash testing showed that, in fact, IF one was a very efficient & slick signalma, there was a "gap" of just under ONE SECOND where the interlocking wasn't & was "open", & the operation(s) perfomed to make the wrong switching move took about 0.6 seconds - & they were simply unlucky enough to slot the one time-gap inside the other - with a non-electronic system (!) Nowadays, if you watch big signalling systems, they clear "backwards" - towards any approching train & the time-delay is perceptible - about 0.6 seconds per signal - on a long straight stretch of track, you can see them clearing back towrds you. And Hull is one reson why.

Not as much as you might think. I remember train drivers being prosecuted for failures that were inevitable given the conditions they were being expected to work under - not so much long hours as being required to maintain an implausible level of alertness, with no backup mechanism (e.g. a telephone in the cab). And that's the railways, which have specific safety boards and regulations - currently, roads have none. Specifically, on the roads, I also remember tank drivers being prosecuted when it was only the officer giving directions who could see adequately. The defence that they were acting under military orders did not help them (or get the officer to face the same charges).

On the networking issue, I agree with Thomas Jørgensen, but for slightly different reasons. No level of encryption will defend against network outages (whether loss of signal, congestion or denial or service), and the probabilities are such that those WILL cause accidents with remotely-controlled vehicles. And, when that happens, the responsibility chain becomes completely broken. I am not as optimistic about decent security, though, given what the car manufacturers are currently (not) doing.

That is one of the things I skipped over with "not perfect" - inappropriate prosecutions have happened, but the accident investigation process itself (which is a separate matter from any prosecutions, and has different aims) at least begins with an open-minded position; as Greg said, it generally takes more than one person to cock up before things can go badly wrong, so you can't assume at the start that it was just one person. I'll leave it there, because this is getting a bit far off topic.

"Can probably crash the car remotely" was not enough motivation to make the manufacturer pull the internet connection.

I have very bad news for you. You think it's a good idea to air-gap entertainment systems from critical vehicle systems? Boeing disagrees, and the in-flight entertainment system on the 787 Dreamliner shares physical hardware with the flight systems.

Greg Tingey objected to my suggestion about identifying the crucial step is that led to the accident:

Greg: " STOP RIGHT THERE Usually false assumption - that there was one single point-of-failure that caused the accident."

Perhaps you missed the last part of my post, in which I said "[of course] reality is always more complex than such a simple formulation as mine can encapsulate. But it's nice to think that some future system will be ruled on a scrupulous examination of the facts and interpretations based on expert advice."

There will undoubtedly be some cases with a single point of failure, and they may well be greatly outnumbered by the more complex cases. The analytical method is still the important point: figure out what actually happened before assigning blame and, to the extent that it's possible, based that judgment on fact rather than opinion.

OK given that proviso... yes. RAIB usually give "proximate cause of accident" & then list all the other things that went wrong - & there may be more than one proximate cause, too.

Pre or post Scottish independence?

I'm waiting for the IoT eLiability camera system (only 23 cameras per vehicle! Plus logging for, like, everything. And a stingray to scoop up local data traffic) so that the ambulance chasers, excuse me, autonomous vehicle online legal services, can offer mediators an "unbiased record of what happened," so that judgements can be rendered expeditiously.

We won't talk about the number of accidents that happen because someone fumbling over establishing a data connection through a third part stingray service loses control of their vehicle or steps in front of the car. Mere trivia.

Yes, I'm being facetious. I'm also not being facetious, because there's a certain kind of mind that might think this is a good solution, with cameras and computers getting cheaper every day, y'all.

A few things

From what I know about truck automation, urban driving is too complex for computers. Computers do well, reacting quickly and efficiently, in environments that are relatively simple and predictable with few variables. So despite the faster speeds, trucks will be automated for highway travel only. The traffic flow is unidirectional, within a given velocity window, and the number of other vehicle in a given stretch of freeway tends to not exceed a certain amount.

Not so urban driving with kids running into traffic, runaway dogs, pedestrian movement, cars ruining stop signs, construction crews, etc. So the current plan is to have automated trucks driving themselves on the freeway until they get to the off ramp. At that point, a human driver back at the truck yard or warehouse will take control of the vehicle using cameras and sensors to remotely drive the truck through urban traffic from a remote station like a drone pilot.

Truck driving jobs won't be eliminated so much as they will be transformed.

Perhaps the thinking around military drones will be applicable? AFAIK there aren't any actual laws or military policies written for more or less autonomous drones yet, but ethicists and lawyers have at least been thinking and discussing the issues for some years now.

Christian Enemark is an academic who has published articles, given talks, and written a book about armed drones. If my notes from a talk are correct, he suggested three ways of looking at the problem of determining liability for death by drone (assuming in the military case that it's not someone you intended to kill):

• It's a product manufacturing problem. The software engineers could be liable.

• It's a supervision problem (his example was being the owner of a large dog). How much control does the operator have?

• And in the future which we haven't got to, is a fully autonomous vehicle liable itself?

He also didn't rule out there being more than one person found responsible, to varying degrees. This is not a multiple choice exam, right?

Note: I'm a lawyer, so I'm going to treat this one like a law school essay exam and not cheat by looking at everyone else's essay or any outside sources before I write my own. Also, I'm an American lawyer, so I'll almost certainly be wrong when it comes to UK law, as well as lots of other things. I'm also only giving myself 25 minutes or so for this to mimic exam conditions. :-D

My Essay: First, before advising my client, I would look up UK, Scottish, Belgian, South Sudanese, and North Korean law regarding torts, criminal negligence, and contracts. [Note: This statement is good for 1/9 points on the Virginia State Bar Exam.]

Tort issues: In the event that the pedestrian or his decedents (if he died) want to sue for damages, the first major issue is to determine whether such a suit would be viable as a matter of law. This will largely turn on whether the jurisdiction uses the rules of contributory negligence or a comparative negligence. Under (the older) contributory negligence regime, if the victim of a tort involving negligence was also negligent, such that the victim partially at fault for the accident--even by at little as 1%--then the tort victim cannot recover because they were contributorily negligent. There are some exceptions to this rule, like the Last Clear Chance Doctrine, which holds that if the tortfeasor was clearly the last person who would have been able to prevent a tort, then the rule of contributory negligence would not apply, and the suit could proceed. However, in the instant case, the pedestrian was most likely the person with the last clear chance to avoid the accident, and had engaged in the reckless conduct that led to the accident. Therefore in a contributory negligence jurisdiction, the struck pedestrian would most likely be unable to recover civil damages from the driver or any other party to the accident.

By contrast, in a comparative negligence jurisdiction, the court must balance each party's respective degree of fault for the injury caused by the tortfeasor. In comparative negligence jurisdictions, a tort victim can usually recover if they are found to be less than 50% at fault for the accident; however, this amount varies by jurisdiction. In some cases where multiple parties are at fault, courts will look at which party had the plurality of responsibility for the accident. In the instant case, however, a court applying the rules of comparative negligence is likely to find that the pedestrian was at least 50% responsible for the accident--meaning that the struck pedestrian would be unable to recover any damages. Moreover, even if the pedestrian was only found to be 40% responsible for the accident, he may still be unable to recover because he would still be responsible for a plurality of the negligence that causes the accident because so many other parties (e.g., the driver, the software company, the manufacturer, etc.) bore partial responsibility for the accident.

Criminal Law Issues: Assuming that the struck pedestrian survives the accident, he could potentially face criminal liability for recklessly playing in the roadway, endangering both himself and other motorists. However, criminal liability would largely depend on the age and mental capacity of the pedestrian, as well as the local jurisdiction's propensity for trying minors as adults. If the pedestrian was 15 and not cognitively impaired, then criminal charges are certainly possible. However, because of prosecutorial discretion, it is entirely possible that a prosecutor would decline to charge the pedestrian in light of the severity of his injuries and the fact that no others were harmed in the accident.

Meanwhile, whether the driver faced criminal charges for striking the pedestrian would depend largely on the state of laws in the jurisdiction in which the collision occurred. The driver could probably claim that he did not have a responsibility to monitor the car at all times, however, assuming that it was legal for him to be outside the vehicle while it was driving in the first place.

Contract Law Issues: In the event that the driver is found partially liable, the ability of the driver to cross-sue the software company and/or other commercial partners for contributing to the accident would depend on the terms of the parties' contracts. Choice of law clauses embedded into clickwrap contracts are frequently suspect. This is because in a clickwrap contract, consideration is often lacking, despite the fact that consideration can be a "mere peppercorn." For this reason, in the U.S. government contracts context, many courts have disregarded and/or severed clickwrap agreements, or choice of law provisions contrary to U.S. law. Private parties, however, such as the driver here, have considerably less flexibility in this regard.

In addition, U.S. courts are becoming increasingly skeptical of mandatory binding arbitration agreements where only one part has the power to set the term of the agreement. That said, it is unlikely that the driver would be able to escape arbitration for this reason alone. However, because this occurred in the UK, a better avenue to pursue would be to argue that the incorporation of North Korean and/or South Sudanese law into a contract for services occurring exclusively in the UK was a violation of UK contract law and/or the UK's constitution (which is merely a collection of laws and legal precedents, rather than a single coherent document). Incorporation of the law of oppressive regimes into one-sided contracts of adhesion may also be a violation of the European Convention on Human Rights (if I recall its name correctly), but that point is more arguable.

[And my time is up!]

While initially appearing to be a test of the laws as they apply to automated vehicles, the true significance of the case lies elsewhere. While the EULA is immediately thrown out during the civil damages cases by the Courts of England as Wales as being unenforceable, a hitherto unknown arbitration start-up based in Belize---one of the few countries to have a free trade agreement in place with post-Brexit England and Wales---sues under the Investor-State Arbitration rules of the treaty.

Claiming to be willing to host the required arbitration panel in Pyongyang with relevant expertise in South Sudan law (despite any sign of having performed any arbitration elsewhere), it asserts unfair competition in the arbitration market by the government of England and Wales, and that the court system represents an illegal government-run monopoly under the terms of the agreement. While such an argument would hold no merit under EU law (following Slowakische Republik v Achmea), in the rush to convert EU law into UK law during Brexit that precedent was unilaterally overturned by ministerial fiat according to the powers granted to them by the Brexit Act to dispense with any European Court of Justice rulings that the government didn't like the look of.

As a result, the case is sent to for resolution at the designated Investor State Dispute arbitrator, which finds in the favour of the start-up and fines the England and Wales Courts £500 million in damages. Rather than facing the embarrassment of tearing up one of the few free trade agreements, the government of England agrees to pay the fine and abide by the ruling. Forthwith the courts are required to compete on a "level playing field" with all other providers of arbitration services and the government quickly adopts a policy of farming cases out to the lowest cost provider to help cover the £500 million hole in the Justice Department budget.

With high fixed costs from outdated physical infrastructure and numerous senior staff, the Courts of England and Wales are unable to compete and are forced to lay off staff, including judges. A game of musical chairs then ensues with judges re-joining the law firms that they originally served with and junior barristers being forced out and into roles as independent contractors for the new arbitration companies that have jumped in to the new market opportunity. The Randian Libertarians rejoice as the court system joins the sharing economy and justice is finally available to those wealthy and deserving few upon whom the market has bestowed its blessings.

Meanwhile, the original case sits in legal limbo. The start-up which proposed to be able to arbitrate the dispute in the required way distributed its court winnings to its investors in the form of dividends and then promptly closed shop. No other company is willing to bid on the case, as the costs of adhering the EULA's requirements are clearly far to high compared to what the government is willing to pay.

And the significance to the law students taking the test: they face a depressing hand-to-mouth existence scrambling to make ends meet while paying off their student loans, hearing arbitration cases for less than minimum wage and hoping that one day they will get that big break and be able to join a real law firm.

(Take with the appropriate chunk of salt: I am no sort of authority in any sort of legal matters, and this is clearly ridiculous. However 5 years ago I would have said that the current state of the world is clearly ridiculous, so what do I know)

For people who don't want to read my ridiculous U.S. law school-exam-style essay, the TL;DR is this:

• The struck pedestrian is most likely to be found at fault for the accident because of their reckless and/or negligent behavior. This most likely doom any suit for civil damages. If the minor struck pedestrian survives, and is a competent almost-adult it's possible, but unlikely, that he'll face criminal prosecution for his actions.

• If it was legal for the human "driver" to be outside the car, then he or she is unlikely to be found criminally liable for the accident.

• The choice/conflict of law and arbitration provisions in click-wrap license agreements may be void for lack of consideration and/or for incorporating the law of hostile, oppressive regimes (admittedly, a stretch). That said, however, this is unlikely to come up because the pedestrian was primarily at fault for the accident and is unable to sue.

It was also the business of Pets.com and look where that got them. Nor do I think it's likely that they'll get market dominance. People use Uber cause it's cheaper than regular taxi services. The moment they jack up the prices people will flock to Lyft or traditional taxi services.

I'd also point out that Uber has been shown to increase traffic congestion. Which will probably result in big cities like London slapping an extra toll on anything with an Uber sticker on it.

Was the pedestrian carrying a cellphone or other device that could be interrogated by the car to determine a human was nearby? If not, the fault lies with the pedestrian who was not clearly indicating a presence to the vehicle. [ALEC inspired state traffic legislation, bought by auto and truck manufacturers and operators, rolled out across 50 states between 2019 and 2023].

So, Mr. Stross, have you looked at the reverse direction, ever? As in, doing away with limited liability corporate charters?

It's actually quite easy.

Limitation of liability was necessary for operational reasons (and other reasons) for companies whose stakes were represented as stock certificates, folio sized sheets of papers lugged about in literal portfolios and traded over coffee in Amsterdam and then London. Because if ever a company was held liable for more than its assets were worth, those papers would be certain to be "lost."

BUT...

In the 1950s, stock exchanges set up end of day trading reconciliation systems to make sure shortly after the end of every close, there would be no confusion or ambiguity about the day's trade, and every stock holder knows exactly how many shares he holds in each company. Those systems have only grown more reliable. If the laws changed, a court could order that on day X, stockholders in AAPL had to pay a fine of Y dollars for the company's wrongdoing, well, the systems are already in place to do exactly that.

That may at times mean that AAPL shares would dive into the negative as day X approached, with shareholders actually paying for the right to wash their hands of whatever had happened. But operationally speaking, negative prices are already a thing. In the commodity futures market, traders who mess up and find themselves owning the actual product because they didn't make sure to sell out of a position in a maturing future, well, sometimes they pay a lot of money to avoid having to take physical delivery of the product. The real time electricity market often dives into the negative, with power producers paying for their operational inability to ramp down production fast enough to answer a decline in demand for electrons.

And investors who insist on living in the LTD world still have that option, by adding options to their portfolio. One option per share, with a strike price of 0. The brokers issuing those options would then have a direct interest in making sure the companies involved never incur a liability greater than their capitalization.

You and Ted Chiang join the long list of authors rightly mocking limited liability corporations, from Gilbert & Sullivan onwards, so, well, why not look at a world that has done without them?

Identifying oneself as human / to be avoided is probably insufficient/irrelevant. Someone already pointed out the wildlife issue. But, in another forum, someone is already offering bicyclists directions for radar corner reflectors, which might, repeat might, have helped in Arizona.

Having done some law TA work, and assuming this is a 2nd year question in a multi jurisdictional law / conflicts of law class where torts is a 1st year pre-req, and the 25 minute time applies, I'd mark an 8/10. Touches on most of the right avenues (which is all that is really expected) but:

1) misses the contract law aspect of certain provisions being arguably void on public policy grounds (without need for reliance on conflict with statute), which, as I understand it, is more common in non-US common law jurisdictions, and;

2) doesn't address the jurisdictions which the amount of a plaintiffs' negligence is not a bar to collection but is an adjustment to the quantum of damages (more than 50% sometime being a bar) which if the jurisdiction is one of joint and several liability they are legally entitled to collect, but then the question becomes one of collections and enforcement across jurisdictions.

The head prof for the course would no doubt take off another point or two to make their curve work better. You could get them back by taking that extra course most law schools offer but don't advertise to their students that they should take - appealing your grades.

I expect that the answer is no. Justice delayed is justice denied, and by the time any of the questions asked is answered by the legal system we will have long passed the point where there's any question of liability being assigned to a natural person other than the driver.

Supplemental question: is the pedestrian/obstacle/deceased a white person or a criminal subhuman?

I expect that we will see a disproportionate number of non-white deaths due to the training data used for these cars, and wonder whether a class-action lawsuit will be allowed, and if so: where, and against whom?

I think you missed my attempt at humor.

jury members, feeling "there but for the grace of God go I", have been very reluctant to convict killer drivers of manslaughter. It'd be interesting to know if jury-less jurisdictions have similar laws.

Judge-only trials in Australia and Aotearoa have the same outcomes, and I suspect we have equivalent "murder, but with a car" laws for the same reason. Regardless, the sentences are derisory. I could quote examples of egregious negligence resulting in small fines but won't.

But I do expect the same principle to apply to driverless cars. The pop-up covering the screen is likely be irrelevant, in the sense that if it's proven causal the punishment is likely to be a request that similar popups be removed as soon as is convenient, please.

I love the idea of "mens rea" being extended to corporations, partly on grounds of insanity and partly in the (I know it's vain) hope that we could thus see corporations imprisoned or executed by virtue of having the necessary intent to commit crimes. It would be hilarious to see all the shareholders of, say, BP, required to collectively serve 100 years jail... every single last one of them turning up for their 14 nanoseconds incarceration per share.

Whilst you have used this driving mode in books of yours, research suggests that a remote driver who is doing anything at all like driving is a really, really bad idea.

Mining machinery already does this, so they're already killing people with those machines. But crucially, they're not killing people on the mining sites, they're merely continuing to kill people with the products of mining (dust and other pollution directly, CO2 etc less directly).

Remote controls for construction machinery are common today, to the point where any truck-with-hoist will probably have one. Cameras/screens are increasingly included but AFAIK the operator is still required to be "on site". But I have seen someone operating a tower crane from inside an air conditioned shed with only a small window (probably illegally... I hope it was illegal!). Realistically they were within radio range rather than actual sight of the job.

There's also construction machines more generally being automated, with the goal of getting workers out of active excavation areas because those workers are expensive (to run and replace). The talk is all of "autonomous" but right now that generally means "with human oversight" of exactly the type you're saying is a bad idea.

sample media hype: https://www.theverge.com/2015/10/13/9521453/skycatch-komatsu-drones-construction-autonomous-vehicles

A good essay already having been provided, there are some additional facts that if known could take the exploration different places. For example, was the phone issued to the driver by his employer or is it a personal phone? Is the phone fit for purpose? That is, is the screen large enough and was the obscured view caused by the screen being smaller than the minimum specs required by the dispatch and/or diving apps? If the phone is fit for purpose, is there a distracted driving law in any of the jurisdictions? I know some distracted driving laws have exceptions for GPS systems, but also that others don't have many exceptions such that interacting with dispatch apps, a radio, or a climate control knob can lead to a distracted driving charge in the case of an accident.

A clever litigator might also try looking to see if there are any cases involving remote controlled [model] cars, airplanes or drones, the application of which may or may not apply depending on the definitions in the various jurisdictions' highways traffic laws.

Now that I'm thinking about it, some jurisdictions have drone operation regulations and depending on how those laws are drafted (i.e. drone regulation not being limited to UAVs), the car could be a vehicle for a highways traffic act, cab for ride services and commercial vehicle regulation, and a drone for other regulatory purposes.

Irrelevant. Scottish Law has ALWAYS been a completely different jurisdiction to English ... the courts system is totally different as well. [ The Scots' Procurator FIscal is more like an examining magistrate in the French system (ish) f'rinstance. ] I take it you are not a UK resident?

What's a "crosswalk"? "Jaywalking" DOES NOT EXIST in any UK jurisdiction. You can cross a normal road anywhere you damned well like - at your own risk.

We have deer too ... inside London ( Richmond & Bushy Parks ) & close outside - Epping Forest & the surrounding countryside too. I've seen a Fallow deer HERE within the past 2 years. She was between where the black car in front of the google-camera is & the next one, going left-to-right. So there. Definitely a hind, as at that time of year, a buck would have looked like THIS

PARTICULARLY as US road & traffic laws bear almost no resemblance to British ones ... in spite of what some people have said, it is much more likely - with a manually-controlled vehicle, as at present, that the controller of the vehicle would be prosecuted, compared to the USA.

And, the present, tory guvmint, is actually looking at ways of making corporations "More liable" in the wake of the Carillion rip-off & other similar fiascos. The tide is beginning to turn in the direction you have posited

There used to be a word for that: Telechiric As opposed to AUtomatic - meaning (now) autonomous. Think Waldo

If you want some very good thinking about autonomous cars, I highly recommend this blog, here's the post about the death in Arizon: http://ideas.4brad.com/new-facts-and-questions-uber-robocar-fatality Turns out it's a 45mph zone, and it seem like the car should have seen her with enough time to stop.

[[ Rescued from the spam trap - the filters consider ideas.4brad.com a bit spammy — mod]]

Many thanks for your thoughtful essay!

England uses contributory negligence for assessing fault, but (per this lawyer's blog) it's not a complete defense. He gives an example, further, in which a young cyclist's contribution to an accident is reduced relative to an adult's (to 50% from 75%, in a cyclist/vehicle collision where the cyclist was partially at fault).

(Parenthetically: English law, while originally common law, has been hacked around a lot over the centuries since the USA departed: accounts of US legal proceedings sound very weird from this side of the pond. And Scottish law is something else again, legislated in parallel with English law between 1707 and 2000 and in some areas diverging since then due to the revival of a Scottish Parliament.)

Also: pedestrians have an absolute right to use the road (in the UK) — there's no ban on jaywalking, with the significant exception of designated motorways, which are restricted to certain types of vehicular traffic only (for example, no horses, bicycles, or underpowered mopeds are allowed).

If you don't mind, I might just steal that as back story for an ongoing long term project. (I want to write a third Scottish police procedural, set in an utterly crapsack post-brexit libertarian UK, with privatised police and courts — this would work wonderfully as an explanation for how the courts got that way!)

Pedestrian is a child, under the age of legal criminal responsibility (eight years old) playing chicken in a road where pedestrians have priority over wheeled traffic (a pedestrian precinct).

Just to clarify, in the world you were describing, it would NOT be negligent for parents/guardians/carers to let such a child play in such a zone unsupervised? Unless that were so, they would assuredly bear some liability.

I assume that, in the Alex Tolley world, it would be negligence to let such a child out into even a pedestrianised zone without a working tracking device attached to the child. Incidentally, that would also apply in the Corran world (which I agree is well imagined), because of a case brought by an automated vehicle operator or manufacturer against a pedestrian.

That's been going on since the 757's "flight information mode" which displays a subset of the cockpit navigation data in the passenger cabin. Of course, that could be done via a pure simplex link.

Readers wishing to learn more about rail safety should track down the out of print "RTed for Danger" by LTC Rolt.

Regards the downgrade of the safety system. As long as the silver package conforms completely to whatever the minimum safety standard is, would this even be considered? For example-people often fit cheap tyres to their vehicles, which are not as good as OEM, which in turn are not usually the absolute best you can get in the first place. But as long as the tyres conform to minimum standards, are correctly fitted, correct load/speed values, within wear limits, inflated correctly etc etc, you are completely within the construction and use regulations (in the UK).

Similarly, there have been cases where classic vehicles have been involved in accidents, and the other party has tried to claim that driving such a vehicle was to blame. These arguments have always been struck down. As long as the vehicle is fully compliant with the relevant safety standards, it is completely legal.

Let's imagine that the lack of a gold sensing package was taken in to account. Again, I assume that the silver is fully compliant with the law. Does this mean that drivers now have a duty to constantly upgrade their vehicles mechanically too? Best tyres, best brakes, braided brake lines, etc etc?

You'd lose 2 marks straight off for references to "UK Law". There is no such thing, English (and Welsh), Scots and Northern Irish law being 3 separate legal systems within a common national border.

Having said that, I can't find much if anything else wrong with your essay.

It appears that at least the preliminary police findings in Arizona are pinning the blame on the pedestrian:

CBS News

Looks like this maybe won't be the test case folks were assuming (hoping?) it might be.

Daniel Duffy notes: "From what I know about truck automation, urban driving is too complex for computers. Computers do well, reacting quickly and efficiently, in environments that are relatively simple and predictable with few variables. So despite the faster speeds, trucks will be automated for highway travel only. The traffic flow is unidirectional, within a given velocity window, and the number of other vehicle in a given stretch of freeway tends to not exceed a certain amount."

Indeed, and one of the proposals I've seen (in an old SF story, if memory serves) that I really liked was to isolate the autonomous traffic from the human-driven traffic. This would be difficult to implement in some areas with high road density, but the existence of dedicated car pool, taxi, and bus lanes in many cities provides a proof of concept that you could bolt such a separate system for autonomous vehicles onto an existing human-only system. (I've seen this both in Boston and Montreal, a small but perhaps illuminating sample.) It might be less efficient than mingling human and AI drivers, but I suspect it will be both acceptably efficient and much safer until AI advances enough to predict, detect, and respond to irrational human behavior as well as a human driver can do.

You'd still need to keep human drivers from cheating by jumping into the AI lanes where complete isolation of the two systems isn't feasible. I see this all the time in Montreal's "tax/bus-only lanes". Increasing policing, with severe penalties for cheating, is one option; using vehicle transponders to automatically record violations is another, although undoubtedly easy to hack. No particular opinion on the best solution.

"Proof" that this works? Japan's Shinkansen system is completely isolated from the road system so that there are no level crossings and no other chances for trains to intersect vehicle paths. The Shinkansen aren't AI (they still have human drivers), so I'm focusing here on the isolation between the two systems, which has meant zero collisions with vehicles (last I checked*) since the system began.

• A NOVA documentary last year on modern rail systems.

To me, the interesting question currently isn't whether fatal accidents will happen (they will), but where does liability get apportioned. Historically, vehicle manufacturers were effectively unitary entities, vehicle drivers were another, and the victim was (usually) a third. The arrival of AI and lots of different chunks of software changes this, as do the (arguably illegal) software licensing shrinkwraps ...

Ah, the dependency problem from software made physical.

Given my earlier comment about regulators being the bigger thorn in the side of manufacturers than criminal liability, we'll see some interesting interactions between the physical platform (General Motors Sensor Platform 2022.1) versus the software platform (Samsung Drive v 13.02.89). Having software be in-scope for safety regulation the way it is for aviation, trains, and medical devices means that the industry's favorite axiom of development is suddenly a massive liability. You can move fast, but you most definitely cannot break shit.

Given the accidental death in the problem, the arbitration requirements are only binding on the 'user' of the platform, which in this case is the guy who got a distracting pop-up right before the kid got killed. He won't be able to sue to recover the costs of his lost car because of their faulty automation, but that isn't the criminal question. Given the exclusive deals common in the west, and the complete lack of incentive for open platforms, a Subaru is likely to only come with a single software vendor's driver-automation. This reduces the targets of the criminal investigation rather nicely. While General Motors may publish an actual open spec for auto-drivers and create an app-store for people to buy auto-drivers from, I have serious doubts that's where they are going. No, it'll be Samsung/Microsoft/Apple/Google and other multi-billion dollar companies providing the things.

When the after-accident review shows that the safety features were downgraded in order to provide a discount on the monthly subscription fee, the regulator will come after the software maker of that specific model of car. Given that nearly all of that model of car on the road run the same software, thanks-be to exclusive marketing deals, the recall (mandatory patch) will apply to all of those models of car. Being a regulator, if the problem is severe enough, they can ban all of the bad version of software from the roads until it's fixed. For criminal penalties that faceless megacorps understand, that's a pretty good one.

For the criminal penalties that the bereaved relatives understand, not so good.

Re: 'The Randian Libertarians rejoice as the court system joins the sharing economy and justice is finally available to those wealthy and deserving few upon whom the market has bestowed its blessings.'

C'mon - how is that any different than the US system today? Most US judges are elected, not appointed - they run/pay for election campaigns just like any other pol. Nice backstory though.

One of the AI applications that I've read about included legal which if realized would mean that personal connections/wealth had less weight re: winning a law suit. I'm of the impression that most law suits fail due to attrition: the plaintiffs die off or have no money left to pursue their suit.

"For the criminal penalties that the bereaved relatives understand, not so good."

That's OK. The response to a systems failure that causes fatalities should be primarily concerned with figuring out what went wrong and making sure it doesn't happen again. That might be by banning automatic cars entirely, or mandating strict safety standards that the software must conform to, or something else entirely - it depends what it was that went wrong. It may sound callous, but the emotional response of the bereaved relatives has no place in formulating that response, any more than the interests of the car manufacturers have.

I mentioned earlier that the rail accident investigation process is not perfect; for an example of imperfection which relates to the above point see the Ladbroke Grove enquiry, which was hamstrung both by the media circus of grieving relatives howling for blood that was allowed to spring up around it, and by the interests of the government and their desire to avoid being fingered for their responsibility in creating the conditions that led to the crash by privatising the railways.

Some US background re: how judges are elected. Given what's been happening with FB, there may be some move to more clearly identify the links between the ad/sponsor and an issue.

BTW - campaign spending is increasing esp. for trial judges.

https://www.americanbar.org/content/dam/aba/migrated/leadership/fact_sheet.authcheckdam.pdf

Excerpt:

'Interest group activity:

Trial lawyers and business groups are spending more on unregulated “issue advertisements” in judicial elections. The Litigation Fairness Campaign, sponsored by the U.S. Chamber of Commerce and the Business Roundtable, hopes to raise $25 million, primarily for television advertisements, in at least eight states holding supreme court elections in 2002. Trial lawyer organizations and unions in “battleground” judicial election states are expected to spend significant amounts, as well. Because expenditures by groups that do not expressly advocate the election or defeat of a candidate are not subject to disclosure, precise figures on interest group spending are not available. (Source: Peter H. Stone, “The Blitz to Elect Business-Friendly Judges,” National Journal, February 16, 2002.)

“Also: pedestrians have an absolute right to use the road (in the UK) — there's no ban on jaywalking, with the significant exception of designated motorways, which are restricted to certain types of vehicular traffic only (for example, no horses, bicycles, or underpowered mopeds are allowed).”

I live in a neighborhood just north of the UT-Austin campus, the state capitol, and downtown Austin, Texas. The capitol and campus are pedestrian right of way zones. The back streets of my neighborhood are pedestrian right of way. If a vehicle hits a pedestrian, the driver is liable.

Driving in my neighborhood can be rather treacherous. People walk in the streets because there are no sidewalks in the area. The streets are wide enough for cars parked on both sides of the street with the middle open for traffic. Backing out of the driveway is hazardous, just when you think all is clear a young parent jogging with a baby stroller and a dog in tow appear suddenly in the rearview mirror. And there’s that kid on a skateboard being dragged by their galloping dog. At any moment a child, a dog or cat may dart out from behind a parked car. After dark people are walking their dogs and riding bikes, some wearing reflective gear and blinking Christmas lights, but many do not.

Driving through the UT-Austin campus is not advised. Many students are wearing earbuds or headphones and appear to be wandering in a trance or mysteriously become blind, or staring at their smartphones self absorbed, they will walk through an intersection without a care. There are students riding bikes through packs of walking students, and maintenance staff driving their electric golf carts. And those maniac students ripping down the hill on skateboards weaving in and out between the pedestrians and the shuttle buses, or zipping up the hill on a hoverboard (a self-balancing scooter like a Segway without handle bars). Have I mentioned the never-ending building construction on campus?

Automated vehicles? There will have to be some major zoning figured out.

The hypothetical case looks pretty clear-cut to me. The "driver" chose to downgrade to the silver collision avoidance plan, if it's possible to prove that the gold plan would have done better then it's clearly his fault. He gambled with a cheaper plan and is then liable for any consequences.

In my imagination this accident gets treated with the same passion as the Columbia shuttle disaster. The government appoints Jeff Dean or maybe Knuth to go to town with the code and determine exactly what it was that went wrong. That being said, as I think Uber doesn't quite share the quality control culture that Nasa had, and Margaret Hamilton remains retired, it probably wouldn't be a very long investigation...

Yes, your paragraph beginning "Driving in my neighborhood can be rather treacherous..." is a good match for typical UK conditions. Everywhere is like that, more or less.

(The one bit that doesn't match is "no sidewalks"; we have them everywhere, but you still get people walking in the road because some tit has parked on the sidewalk, not to mention kids playing in the road; the ones with bicycles are perhaps the most worrying, as they have all the directional stability of a housefly and are nigh invisible after nightfall.)

I get the impression that people advocating automatic cars in the UK don't realise quite how different the conditions under which research in the US is carried out are.

Speaking of silver vs. gold packages, it seems like a good solution would be to open-source all the safety software and make it freely available to any manufacturer, including home hobbyists. Or alternatively, create it jointly through an industry-wide cooperation among all automobile manufacturers. The goal would be to maximize safety by getting everyone to work together, while lowering the costs for each manufacturer because the development costs are shared.

There are many problems with this suggestion, ranging from the ridiculous to the sublime. For example, there will undoubtedly be corporate nonsense such as "safety software is a competitive feature that distinguishes our cars from lesser vehicles -- never mind that the distinction is how we're trying to skimp on the software quality to reduce costs". There will also be objections about the management overhead from such a project. But by now we have enough experience with "standards" [sic] and open-source development that the notion is at least plausible. Think of Linux, for instance.

You do realise that you've made a compelling economic case for granting legal 'personhood' to autonomous vehicles?

A lightly-customised chatbot script would suffice for providing answers to a narrow range of questions asked in court...

... And then all liability falls on a 'person' with all the general intelligence of a raw carrot outside of its task-specific decision algorithms, and a prerecorded set of sincerely-expressed refusals to sue its manufacturers and 'business partners'.

Please feel free to use this however you will, although you might want to check with those more expert in such things if the scenario passes any sort of legal smell test.

And if you really want to crank up the legal dystopia, the logical end state of this situation is that the lowest cost providers of arbitration services have long since followed the Uber model and have replaced their independent contractors with AI machine learning the simplest back-propogation neural network/naive Bayes classifier/support vector machine/decision tree method that passes some minimum score against a standardised subset of the appropriate body of digitised legal precedent. High end providers of course have more sophisticated models and they use this heavily when marketing their services to mediate disputes between corporations or those with some ability to pay, but for most low stakes legal cases the level of error of a simple model is considered acceptable.

For example, a dispute between neighbours over a property line might take as inputs the digitised ordinance survey, satellite imagery, any photographs of the claimant and defendant might submit to support their assertions, along with written statements putting forth the merits of the case and the defence. These will be vectorised, fed into the model, and a result produced in favour of on or other of the parties.

Not only does this have the obvious problems with overfitting and fossilisation of historical biases, but most of the low-end automated arbitration services are minor variations of the earliest published academic models that showed any ability to accurately predict the outcomes of historical test cases. As a result there are well understood adversarial attacks against the algorithms, and there are services which are more than happy to help a litigant modify their submissions to take advantage of these flaws.

As a result, as often as not these low-end cases are decided not so much by the merits of the positions, or even on institutional racism or class distinctions which have been baked into the models, but rather by who can afford the best adversarial text or image noise to add to their submission to the arbitrator. There is of course, the obvious red queen's race between the providers of such adversarial features.

While the flaws are well understood, the cost advantages of this system are undeniable, and as a result these arbitrators are starting to make their way into the criminal justice system, particularly for less serious offences.

So to bring the circle full circle: in addition to the case left in limbo by the inability to find someone to handle the North Korea/South Sudan situation, the remaining court cases around the incident are resolved as follows:

And since the same technologies have been applied throughout the academic sector, the exam papers submitted by the students who are writing responses based on this test case are also being graded by the simplest automated systems available. Students study for their tests by memorising seemingly random sequences of words to append to the end of their answers that their tutors tell them will fool the grading algorithms.

Purple quantum frantic sitcom frangipani.

Given the current situation in the UK, a legal smell test would pass an offended skunk as perfectly acceptable.

the digitised ordinance survey

Ordnance Survey. Measuring where to make things go boom rather than collecting local laws.

You're right, it isn't a lot different from the current US system, particularly given the seemingly free reign given to force binding arbitration in many situations. I am a newcomer to the UK after living for 20+ years in the US, and it seems as if the legal system here hasn't succumbed to some of the failure modes that the US system is susceptible to; nor does it seem likely to be headed in that direction without some sort of push.

And as for using automation to help decide legal cases, see my response to OGH to get an idea of how badly that could go.

Ordnance Survey. Measuring where to make things go boom rather than collecting local laws

I plead UK newbie.

Not a lawyer, and going to ignore the adversarial back pack stickers (mine!).

If the car wasn't safe to operate in the environment and for its intended purpose then I would think the manufacturer would be up for criminal and civil issues. Geo-fencing now for self driving cars, so if the car was in full autonomous mode in an area where it couldn't reasonably guarantee the safety of passengers, the vehicle or pedestrians then everything else is fluff. The car shouldn't be able to enter the zone without the gold package, and the manufacturer will be forced to do a recall and or pay people off in the UK.

Now, if the car was a grey market import from India... well... Does the republic of Scotland have extradition to the UK?

I get the impression that people advocating automatic cars in the UK don't realise quite how different the conditions under which research in the US is carried out are.

Relying on the conclusions of a report without checking the details can lead to erroneous conclusions, yes.

Up here (Toronto) we have people advocating later starts for high schools based on US data showing that later starts benefit students. Thing is, US schools start at an ungodly early hour, so for them a "late start" is 8:30 (almost half start at 7:00!) while all our schools start at 8:45. So in the US a "late start" is 8:30, but our late start enthusiasts want to push our start to 10:15 based on the US studies.

Sometimes it's ignorance, sometimes it's willful ignorance or deliberate trickery…

What's wrong with skunks?

The smell of skunk spray is a bit strong, but honestly I find it less nauseating than some of the "perfumes" worn* by coworkers in my supposedly "scent-reduced" workplace…

*Showered in, washed with…

Geo-fencing now for self driving cars

Trivially easy to implement, I would think. My consumer-grade drone has geofencing — it simply won't take off or fly too near an airport. (It may be possible to override this — I haven't checked — but making an impossible-to-override geofence seems like it should be easy, especially if you can err on the side of caution.)

Open-source piloting software for autonomous aircraft already exists and seems to be in a pretty advanced state.

Another tangent: The road-kill factor. What say the child in OGH's essay was not a child but (say) a protected/endangered animal - I am thinking of (say) the Kiwi (bird) in NZ. Killing such birds is illegal and therefore prosecuted. Again - who would be liable?

Then there are the "allowable" road kill - eg possum in NZ - http://www.doc.govt.nz/nature/pests-and-threats/animal-pests/possums/ which when crossing a road in the dark is quite similar in size and appearance to a cat. Should the automated/semi-automated car stop/avoid cats but should it ignore (or even target) possums. Note: as the possum is an animal pest in NZ but protected - I think - in Australia where it is native - suggests programming would vary by country.

Note: I am choosing animals small enough that hitting/squashing them should not damage the vehicle - obviously size/mass is (or should be) a factor...

10:15's pushing it. Regardless, speaking as a current Ontario high school student, first period math classes should be illegal.

Corporate manslaughter caused by a malfunctioning machine.

This sort of law is fairly well worked out; the machine is supposed to avoid deaths, and it failed in this objective. The corporation is thus guilty of corporate manslaughter, and the only question at this point is where the levels of culpability for each party contributing to the accident lie.

But was the machine malfunctioning?

Local Police Chief Sylvia Moir says a human driver would probably have not been able to avoid Ms. Herzberg when she "abruptly walked from a center median into a lane of traffic." Chief Moir goes on to say "it would have been difficult to avoid this collision in any kind of mode (autonomous or human-driven) based on how she came from the shadows right into the roadway."

I expect the future Test Case would probably hinge on the court deciding whether a human driver could have avoided the accident. If the court determined a human driver could not have avoided the accident either I don't see how they could find the autonomous vehicle at fault.

It would be like those cases where a driver is passing a playground where kids are playing pickup baseball & one of them hits the ball out into the street. The kid chasing the ball doesn't look to see whether there's traffic and runs out between two parked cars so the passing driver never sees him coming. The kid is not old enough to be legally responsible, but it's not the driver's fault either.

That the operator behind the wheel at the time has turned out to be a convicted felon who served almost four years in prison on an attempted armed robbery charge won't affect the courts judgment of his culpability in any way whatsoever.

The video of the crash is out

https://www.youtube.com/watch?time_continue=21&v=P243yeOnKoo

I should say this video argues for using non-visual modes of navigation such as LIDAR in addition to a typical visual mode. I think it would be very difficult for a computer (or for that matter, a human driver) to stop on time given how corner case this was.

Pedestrian is a child, under the age of legal criminal responsibility (eight years old) playing chicken in a road where pedestrians have priority over wheeled traffic (a pedestrian precinct).

How much responsibility does a parent or legal guardian have for preventing a child from playing in traffic?

In New Zealand both the cat and possum are pest species but targeting either would be controversial. Even in NZ there are animal rights extremists who would rather the country be completely deforested and most native species exterminated, than see any attempt at pest control I can't make it sound rational probably because I don't think it is.

Has anyone else read Version Control by Dexter Palmer? I've just got to the accident with an autonomous car (don't want to give too many details as spoilers) and it seems to follow many of the points Charlie mentioned in his original post: hardware by one company, software by another, click-through license for supervisory driver with nasty details hidden in the fine print…

We have late start days every couple of weeks, so 10:05 start instead of 8:45.

The kids who are late for period 1 are still late for period 1 — and the kids who fall asleep in period 1 still fall asleep in period 1.

I'm not convinced late starts will do much (other than lengthen student commutes, because they'll be pushed into rush hour).

Missing from you set of legal conditions:

Any idea of a Warrant of Fitness (that's the NZ term) or MoT Test (I think that's the UK term) for automated vehicles.

Here in NZ that matters quite a lot for current driving in non-autonomous vehicles. As a driver you're responsible for getting your vehicle warranted and pay a small fine if it isn't. But if you are driving a non-warrantable vehicle then are likely to be found liable if an accident occurs.

So if you know your brakes have become weak, or your tyres bald, and wouldn't pass a vehicle warrant test - or if you should know - then accidents involving not braking fast enough, or the car sliding out, are likely your fault (or more your fault).

The vehicle manufacturer being at fault for the problem with the vehicle doesn't stop you being liable if you're driving it when it's not in a warrantable condition.

I expect many jurisdictions will want only cars that are licenced to drive autonomously safely doing so. At which point another point of liability is the warranting organisation: did MoT licence a car to drive autonomously when it shouldn't have?

(and if you think VW cheating on vehicle emissions tests was a fuss, wait till you see the next one...)

I'm not so much interested in the legal aspects of things as fascinated by the way the question you set up is very much a deliberate example of the way the nation-state is being gradually deconstructed at the end of the neo-liberal capitalist period as an example of expanding global interconnections. It's just such a gorgeous example of the sort of stuff I'm currently studying in an Introduction to Cultural Studies course.

(I love my degree. My other unit had a two hour tutorial yesterday on fanfiction. I really love my degree!)

Stephen T @ 106:

It might be worth programming any car/truck/tank released in Australia with the ability to detect and avoid wombats. They're solid little beggars, and they seriously mess with the undercarriage of most automotive hardware. They're also about the same rough size as a cat, although being built like a cat made out of house bricks.

Once upon a time, yes, that's how the "OS" got started. These days, they are "simply" the world's best mapping service

Liability I'm going to link to two RAIB reports ... at opposite ends of the spectrum.

The first where, quite frankly a pedestrian wasted herself through stupidity & having her ear-jangles in & paying no attention at all to the world, with fatal results ( Incidentally at great expense, there is now a bridge at that crossing. ) HERE

And secondly a sequence of events, that very fortunately hurt no-one, but where, quite frankly every possible safeguard was ignored or bypassed by the human operators in a sequential fuck-up of ridiculous proportions. That no-one was hurt is even more remarkable, in that the derailment took place about 10 metres up on a narrow viaduict! HERE

Compare & contrast as they say.

“...there are well understood adversarial attacks against the algorithms, and there are services which are more than happy to help a litigant modify their submissions to take advantage of these flaws.”

Or “Lawyers” as we technical people would call them...

Backing out of the driveway is hazardous, just when you think all is clear a young parent jogging with a baby stroller and a dog in tow appear suddenly in the rearview mirror.

Bad road law planning, then: in the UK, it's an offense (not often enforced, though) to reverse into a road: you have to reverse from the road into your drive when parking, or turn around on your own property, so you can drive out with full visibility. (Admittedly the UK driving test is a lot stricter than the US one and includes things like parallel parking and making a three-point or n-point turn.) Also, almost all urban and most exurban roads have pavements (sidewalks); you won't get planning permission to build a new road without one on each side, unless it's a motorway or a very restricted driveway (no through traffic).

Driving through the UT-Austin campus is not advised.

Well, no! It's an area intended for pedestrian use — cars shouldn't be there except under exceptional circumstances, any more than you should drive a car through an indoor shopping mall.

Rarely is an understatement. I have never heard of it being enforced once.

Presumably it's one of the things they add to the charge sheet if a driver manages to run someone over or cause a pileup.

I've watched the video several times, and I literally could not see the pedestrian in the video picture until the headlight beams struck her. Of course, in that exact situation I'd have been driving on high beam since there was no opposing or close ahead traffic to be inconvenienced, but I think it speaks volumes about the inadequacy of that vehicle's low beams that you can't actually see 120 feet ahead.

It's probably on par with the "you are not allowed to drive away when you're parked on the wrong side of the road, and the light conditions require headlights" (because having headlights on, on the wrong side of the road, is an offence). The rule's there, but not very enforced.

During the recent snowy weather, I have, alas, seen multiple UK drivers do something that is either an automatic "careless driving" or "reckless endangerment of the public" in Sweden, depending on the exact thickness of the snow build-up. It is really really good to brush off any sow buildup on your car before starting to move, as it's likely to let loose in one go, landing on the wind screen of the car behind you. If it's thick enough, that means a cracked windscreen.

Work facilities guy wanted the building roof checked. Collegue bought drone in but it wouldn't fly. Inside the fence for filton. Quick chat to manufacturer and a software upgrade and fence removed (Filton is defunct). Drone flys.

And there is a problem. How do you keep the fences up to date? If you are connected you are hackable and still only as good as the map, if you are not you are dated.

Apparently a key debate in self driving is how much of the problem you solve with better maps and comms vs what you do with sensors and ai. Much of the US is more mapable than the uk (jaywalking, road layout, traffic lights and stop lines as discussed) so much of the push on that side of the pond is to improve by better maps and better comms). Meanwhile the uk guys are going much more full a.i.

In the test case this is not totally relevant, because the car has to be fully connected and remote controllable to allow remote operator. But a slight twist would be should a standalone car read a sign to say it's not allowed, in which case the quality of sign is a factor or should it be using a map.

I'm only about a third of the way through the comments at this point, many of which are interesting and are causing me to pop up a few search windows and dive around Court judgments etc.

In NZ there are four levels of charge relating to injuries or deaths caused by vehicle. In order: careless, dangerous, or reckless driving causing injury/death, and manslaughter. The latter comes up only a couple of times a year in a population of 4.5 million. The maximum penalties scale from three months to five years imprisonments (and/or fine) for the first three, and up to life for the (rare) last. Charges against a limited company are a different beast, of which I know little.

Different from the US/UK & most other countries, we also have a 'no suing' process regarding civil liability - the right to sue for damages in cases of personal injury was replaced by a government funded accident compensation scheme 40+ years ago. In the broadest sense it seems to work, though suffers under long runs of conservative governments.

As people have said, there is a big difference between criminal and civil liability, and the difference between the two really needs to be kept in mind with any thoughts and comments.

There is a phrase about hard cases making bad law. Laws can draw lines in the sand which gloss over the nuance of real life situations. That’s why discretion is factored into our legal system: we allow police and judges some leeway to treat exceptional cases with the nuance they deserve.

However, the wording of the law can lead to results which are anathema to public opinion. One example was a murder case in NZ which led to the defense of 'provocation' being specifically removed from the lawbooks.

I could see a situation where if the driver was found not guilty, or if the Crown case collapsed, and that no civil claims could be progressed, that Parliament might pass rushed legislation in the wake of public outrage to address the issue.

Having said all this, I'll continue to work my way through the rest of the comments.

I'm pretty sure fault and criminal responsibility are 2 different things.

A child or someone in the middle of a severe psychotic episode doing something that would normally be a criminal offense may lack any culpability or criminal responsibility for the event such that under the law they would face no criminal penalty but that doesn't mean someone else automatically starts wearing the "criminal" and "fault" hats.

When I lived in the north of england snow wasn't a big deal. People saw it several times a year and knew how to drive in it.

Living in the south is very different. Getting on the road in snow is terrifying because no matter what you do there will be some idiot tailgating, performing wacky races type overtaking maneuvers etc.

Not getting pissed off and doing something stupid because the person in front is taking account of conditions will be a big plus for autonomous cars.

Here in Canada you are supposed to clear the snow off your vehicle before you start driving. I've only heard of the police enforcing it in particularly egregious cases (such as only clearing a 15cm circle in front of the driver), and many people just run their piers a couple of times and drive away (with no side or rear visibility).

I suspect it's one of those things that would get added to the charges after an accident, if there was evidence of it.

Yes, that was my first thought as well - the victim is calmly walking a bicycle across the road, in the full expectation that they will make it across.
Meaning they didn't know the car was there, since they make no effort to evade.

The headlights are illuminating approximately one dash and two gaps of the road surface, which is around 30ft. They should be illuminating 100ft as standard for low beam - five dashes and spaces.

I wonder if the Uber team deliberately pointed the headlights down more to enhance the lighting for their cars recognition systems, because the car as shown in the video was not legal for driving based on a NZ warrant of fitness, in which case they are absolutely liable.

In fact at 40mph the car would need 100ft of road to stop, meaning that the person never had a chance even with perfect reactions by the car. Sure, they were wearing dark clothes, but had white shoes and a metal bike, and the very first indication of illumination is at 40ft. That should fall under criminal negligence by the driver in any jurisdiction outside the US, or in this case by the car preparation team. The person inside the car is purely a mechanical backup for dumping liability onto, did they even have a second set of controls?

I've also just seen a NTSB page that lists US low beam headlights as supposedly 160ft minimum, which makes this even worse, the Uber headlights were between 20-30% of legal requirements.

Also: per the released video the driver glanced away from the road for a critical couple of seconds.

That's an instant "driving without due care and attention" conviction in the UK.

As for the stopping distance — if the car could not safely stop within visibility restrictions then the driver was driving at an unsafe speed. The speed limit is a maximum permitted speed, but you're supposed to drive at a safe speed. Okay, so it's on autopilot: the driver, in British law, would still be legally in control/responsible for the vehicle.

And as noted, it looks like the headlights were illegally adjusted. If they'd been on full beam or working properly while dipped, would the victim have stepped into the road?

Uber may get away free in Arizona, but in the UK it'd be slam-dunk conviction for causing death by careless driving (at least), and possibly operating a vehicle in an unsafe state (the lights).

per the released video the driver glanced away from the road for a critical couple of seconds.

They'd probably get away with that in Canada. See the article I posted above:

Bawden explained that the duration of the lapse in attention was key to his decision. According to the Crown’s expert witness, traffic officer Dawn Mutis, the evidence shows a provable period of inattention of between 0.74 seconds and 1.18 seconds, Bawden said. “On the evidence of the Crown’s own expert witness,” Bawden said, “this would be a miniscule period of inattention.” Even a longer attention lapse alleged by Crown lawyer Scott Pattison, Bawden said, would still mean a total period of inattention of less than two seconds.

This, Bawden said, qualifies as a “momentary lapse of attention” of a kind deemed non-criminal by the Supreme Court.

https://www.thestar.com/news/gta/2017/11/21/not-guilty-verdict-for-driver-who-mounted-the-sidewalk-and-killed-pedestrian-on-dundas-st.html

So it looks like UK law is different from Canadian law in this respect (despite Canada inheriting a lot of laws from the UK).

Do aircraft autopilots have anything relevant to add to this discussion? While aircraft still have pilots, I recall issues with autopilots failing (sensors freezing, etc) that cause crashes as pilots take too long to override them. Were airframe manufacturers every made liable? If only the airlines pay out for passenger deaths, then liability falls back on the owner, whoever that is. When a plane crashes, who pays for the damage to property or deaths on the ground?

If we ever get pilotless planes, then what happens? This seems to be coming as self-driving taxi drones seem to be all over the news these days, which should instill confidence in pilotless air flight. Perhaps pilotless private corporate jets next. Finally commercial big jets?

Actually something else is occurring to me now. The "pedestrian" is wheeling a bicycle; In UK law this actually makes them "a vehicle" and as such they should have given way to a vehicle proceeding along the main carriageway which they are crossing.

Law hat on...

In the original scenario, there are two sides to consider, civil and criminal. Civil is easy. The event happened in England so English tort law will apply (Lugano convention). The taxi firm gets sued and has to pay out - res ipsa loquitur that they didn't have proper procedures in place to ensure public safety. Much litigation may then result between firm, driver, suppliers etc but that is a whole different issue.

In a sensible world, legislation would be passed to place strict liability and compulsory insurance on autonomous vehicles. In France already drivers are automatically liable for compensation for injuries in any accident even if not their fault, so not implausible to adopt that elsewhere for autonomous vehicles. That way the insurers will keep pushing for improvements to avoid accidents.

Criminal, that very much depends on what legislation may pass in the future (since at the moment autonomous cars on public highway are illegal- Law Commission/Scottish Law Commission about to work on draft legislation). However, most likely there would be no criminal charges, it would be a consumer protection issue. Failing to keep up to spec would be an offence, not any harm a fully compliant appliance may cause.

As to #59: you don't need that scenario to replace courts with private arbitration systems. Increasing court fees and recent legislation to make arbitration more available sets the backdrop. If you read the newly published "The Secret Barrister" about how the (English) court system is on the point of collapse; easy to imagine Westminister permitting/promoting a two tier system with the rump public court system there only for cases where neither party can afford to use a private arbiter - when only one party can they get to pick who to use. Add WTO rules on competing for outsourced services (Bleu passeports!) and you easily get the scenario without any free trade agreement needing to be in place.

Except ... arbitration i "Small Claims" cases ... Many (most?) small claims cases in English civil law are now handled by arbitration set-ups, usually funded by the big relevant industries, as its cheaper than going to court for everybody. I used this about 3 years back - I git a new (BlackBerry) mobie & it was utterly effing useless, esat its batteries etc ... I took it back to the shop under "N a ot of Merchantable value" claim - who said: "It's longer than 4 days, tough" - so I went to the provider who merely stated their company policy, so I said -"Don't care, English Law Nomv etc" They wouldn't wear it, so I searched small-claims courts - got an arbitrator, who then said - oops not us, you want these other guys, so I did that ... & off the claim went. The phone company wriggled & lied & tried theor "proicedures" as opposed to The Law ... but, with a one-month deadline (After which it automatically does go to court) they finally caved in with 36 hours to spare & paid up. The system works. And I now use a Samsung.

A question, since I don't know enough about the difference between the Scots & English legal & police systems:

Assume in the scenario: Scotland is part of the UK, both outside the EU. The accident is in England. The dispatchers are in Scotland. The location of the 'driver' is unstated.

Does it make a difference if the driver is in England vs Scotland if Northumbria Police charge them with dangerous driving causing injury?

Does it make a difference if the English police attempt to charge the Scots-based dispatcher with some relevant/related charge?

In theory, yes, though with no penalty other than 2 (?) years' disqualification (recommendation is a medium community order):

http://webarchive.nationalarchives.gov.uk/20100519200657/http:/www.sentencing-guidelines.gov.uk/docs/death-by-driving-advice.pdf

But a good many cases result in an acquittal when the driver can find some extenuating circumstance, even if it was predictable. This is just an example I found, but is very relevant to autonomous cars:

https://beyondthekerb.org.uk/at-the-going-down-of-the-sun/

I do not know how many were NOT disqualified, nor do I know how many other cases were not even prosecuted, though I know that it is a very high proportion (possibly even half).

No, sorry, not since 1980. The BICYCLE remains a vehicle, but it is a vehicle controlled by a "foot-passenger" (pedestrian), and no different from a pushchair. Crank vs Brooks, Waller LJ, 1980.

Negligence in England and Scotland is governed by the Donoghue v Stephenson case. The eventual outcome of this case was that a duty of care is owed by the provider of goods or services for any act or omission which could reasonably be forseen to damage a neighbour and a neighbour is anyone who it could reasonably forseen to be damaged by the act or omission. (That's not a direct quote because I cant find my 20 year old notes.) By this criterion all are negligent and a lot of expensive litigation would be required to decide who pays how much.

Some years back many NorAm major cities started fencing major overpasses (overcrossings) to prevent/reduce suicides. Not unlikely that suicide by self-driving cars becomes a thing and communities insist on better segregation of foot and vehicular traffic.

Charlie, here's a legalism you missed, I think: to have the taxi medallion, I would be really surprised to find that ownership of such did not require certain equipment, in this case, the gold package of pedestrian avoidance.

So, let's see, the taxi ("ride sharing") company is liable for not assuring that the driver's vehicle did not have the required packages.

Oh, and the driver would have to prove that they were actually monitoring the vehicle at the time, and not texting their friends.

Unfortunately, that's not the reason. There was just in the last six months a young man charged with murder, I think... for throwing a cinderblock down onto a vehicle passing on the Interstate, resulting in the driver's death.

Just to get y'all civilized folks a better view of this - and this points to the ammosexuals and two-toothed rednecks out in God's Country (tm), I have an old issue of Model Railroader from the late 80's, in which is an article on "Upgrading Your Cabooses to 80's standards"... which included being able to withstand a cinderblock from above, and a .22 bullet from the side.

(Admittedly the UK driving test is a lot stricter than the US one and includes things like parallel parking and making a three-point or n-point turn.)

There is no such thing as "the US driving test", since it's up to the individual states to license drivers. Both parallel parking and three-point turns are included in most (though not all) US state driving tests. (Arizona includes both.)

Geofencing that can't be overridden via "hacking" requires at minimum that the software can't be updated. Nobody seems willing to accept the idea of selling things that can't be updated by the manufacturer...they just don't want the end user to update them. Which means that a lot of updates will be nefarious (not all by the manufacturer).

It also requires that either you have an immutable map, or you except direction from external radio signals. So far everyone leans in favor of external radio signals. This is wide open for spoofing as soon as it becomes common. (This will, of course, be illegal if you aren't a government, and that will surely keep it from happening.)

So, no, I don't think a geo-fencing that can't be overridden will be easy. Or likely. Now one that can't legally be overridden is easy. Just pass the "right" laws.

Admittedly the UK driving test is a lot stricter than the US one and includes things like parallel parking and ...

I don't know about all the states but I had to parallel park. In Kentucky no less. A quick search seems to indicate that 16 states don't require it.

As an aside I had a teacher in high school who told us that even though we thought we lived in the sticks, HE really grew up in the sticks. When he took his driving test outside the court house in the county seat (this would have been in the mid 60s) one county over from us, the testers first said, "Imagine there's a curb".

The problem is that people can't quickly switch from "I'm just observing" to "I'm in control". It takes, IIRC usually around 2 seconds, and it may have been longer. So if the autopilot freezes, the pilot needs time to assess the situation from the new "taking control" perspective. (Also, in some of the cases I've seen there were instrumentation problems as well, which may even have been why the autopilot froze up.)

I'm not an expert in this area, what I know is derived from news reports. But the news reports often don't make much sense in their conclusions, even when they report the necessary facts.

There are also the reports of studies of people expected to sit in the driver's seat "just in case", and unanimously they report that the people soon stop paying attention, even though they know they are being monitored.

My wife took driving tests in three jurisdictions: Germany, Minnesota and the UK, and was rather dismissive of the standard of the Minnesotan (and rather admiring of the German one). On the other hand that was quite a few years ago, and it's quite likely that all three have changed since.

IIRC the new Brit driving test includes "Obeying instructions from a shat-nav - ARRRGGGGHH! I would almost certainly faI=il - as the things irritate me beyond measure, distracting artificial "voice" and all too often - wrong.

“Bad road law planning, then”

Unprepared urban planning is more like it. The problem actually has to do with population density; Austin’s population has doubled from a half a million in the early 1990s to a million at present. Before 1900 my neighborhood was rural farmland. The streets and property parcels were established a mere 100 years ago. Urban planning in the 1950s revolved around the American obsession of the automobile, while mass transit was diminished. In the last 20 years I have witnessed every vacant lot and many smaller homes in my neighborhood and throughout the city being sold to developers placing/replacing them with bigger houses or apartment buildings. Parking has become a major issue in the city, there never seems to be enough and the parking spaces are ridiculously small. In short, urban planners are unable to keep up with demand.

“Well, no! It's an area intended for pedestrian use”

Oh, I agree. UT-Austin has started to remedy some of that traffic congestion by turning a major through street into a pedestrian plaza in the middle of campus (students with cars would all cruise this street at the end of the day). However, there are still a couple of through streets and roads bounding the campus that need to be addressed.

I expect that knowing when to ignore instructions from the satnav is more important.

There have been far too many stories of people turning down the dirt track with "unsutitable for motor vehicles" and "danger! cliff edge!" signs because the magic box told them to.

Teaching people that programming it at 70mph while holding a phone with the other hand and steering with their knees is inappropriate makes sense too.

Yeah I know teaching that sort of thing shouldn't be necessary, but the evidence suggests that it is.

Meanwhile, here's an electric car that Greg will probably like (the Bollinger Motors B1). Kinda-sorta like a 21st century Land Rover 110, only electric. Designed for off-road/farm use, traditional instruments (no multifunction displays whatsoever), looks like you valet the interior with a hose pipe, all the aerodynamics of a brick. Oh, and it has permanent AWD, and can carry about 2 tonnes of cargo internally and the same again on tow.

Back in the late 80s I drove through Detroit on the way to a writers' conference. CAA (Canadian Automobile Association) warned me to watch the overpasses, and if I saw a small group idling there not to pass under them. Apparently it was a 'thing' to bet on which way the car would veer after the brick hit it…

I'm thinking in terms of liability. If an area is geofenced so that autonomous vehicles can't go there, then the issue becomes how the vehicle got there — either operator override (they are liable) or external interference (perpetrator is liable). I expect momentary hacks (confusing the car for a few seconds) to be rather hard to prove, especially as the car/software company wants the owner/operator to take the blame.

That is remarkably like a landie PROBLEM $60k FORGET IT

Just to make you all feel safer, a couple years or so ago,Vint Cerf was here on campus at a conference, in his role as tech advocate, or whatever it was... and he said that google had "better results" by having no steering wheel or pedals.

Greg, I think you'd like the anglonesian systems for that. We have a bunch of "you can't contract out of this" consumer protection laws and the people that enforce them are quite reliable. So you will always have a 1 year warranty on anything that could plausibly last a year (ie, not fresh fruit) and anything you specifically ask "will it do this" and they say yes... that's binding.

It does lead to amusement at the bottom end of the retail market, because people obviously buy cheap junk, keep the receipt, when it falls apart before the year is out they take it back and get another/a refund (it's important to get the refund, that way you can buy another one with a new warranty).

There's some ugly complexity around shrink-wrap and click-through licenses, but to date the big uglies have been largely unwilling to stand up in court and say "when you buy a phone, what you're really buying is the opportunity to read the T&C on the software that makes it work. You don't have to accept the T&C, you have still received the chance to read them that you paid for".

(falls over laughing)

Actually, I like the Bollinger B1, but it's a silly beast. I guess we'll be calling it the Bone, as with the plane. Or something derivative of that.

First of, $60,000 isn't bad. The Chevy Bolt tops out at $45,000, while a 2018 Range rover is up over $87,000. Halfway in between for a big-ass car is appropriate, although it might offend Greg's delicate sensibilities.

What is bad is the 60 kWh battery. That's silly. The Bolt has a 60 kWh battery. The B1 weighs three times as much and has the aerodynamics of a drag chute. I'd be shocked if you could go more than 100 miles between charges (that's empty. Hauling cargo it's going to be less), and that's a wee bit of electricity to be cranking out in the back of beyond.

I also hope it's got one hell of a regenerative braking system. Otherwise, it's going to run out of power going up the hill, and not be able to get it back crawling back down the hill.

On the flip side, EV or hybrid isn't necessarily stupid for 4WD wall-climbers. The power consumption of a landie in 4WD low is somewhat depressing, although I think my calculation that it was 2 gallons/mile was due to the gas gauge being broken in Thelma, the one landie I've ever driven (an grumpy old Defender with a plastic top and semi-opaque windows. Her name fit). It will take a lot of power to haul anything up a goat road, but the nice thing about electric motors is that you can get a lot of that back by creeping slowly back down the road.

The lack of airbags is a particularly sporting touch, as well as making it less than street legal in some jurisdictions. I wonder whether it has seatbelts, for that matter? If it does, I think the B1 is one of those things that might be most useful as a Queensland Bull Buggy, assuming the motors are pretty waterproof.

Some years back many NorAm major cities started fencing major overpasses (overcrossings) to prevent/reduce suicides.

My impression is that the US overpass fences, not all of them on major crossings, are more to prevent the local teens from tossing stuff onto the roadway below than to prevent suicides. A couple of days ago I noticed a fairly recent example of such: https://tinyurl.com/y8x9xjh2

A couple of days ago I noticed a fairly recent example of such: https://tinyurl.com/y8x9xjh2

I think that's noise fencing rather than anti-youth barriers. Round here they have a habit of putting border wall style prefab concrete fences between major transport routes and adjacent houses purely to keep the noise contained.

This brutalist dream is part of a motorway widening project and is all about protecting people from motorists.

https://www.google.com.au/maps/@-33.8405727,151.0406761,3a,75y,270.72h,77.33t/data=!3m6!1e1!3m4!1syyIB_Bk0rJDu_rNODERIPA!2e0!7i13312!8i6656

Can't comment on the legal side ...

Watching the video I am struck by how different the woman's behavior was from what I would do: crossing a 6 lane highway with a central reservation not at a pedestrian crossing is something I find scary. But she does not seem to be looking for oncoming cars, the oncoming car that must have been visible for at least 150 yards yards, she makes no move to escape, no rush but freezes.

But if I had been driving (even with dipped headlights) I think I could probably have avoided her by a sharp left turn.

So ignoring the legal shit: looks to me like the woman was negligent, but also looks to me that the Uber AI did worse than an attentive human driver would have.

My wife took driving tests in three jurisdictions: Germany, Minnesota and the UK, and was rather dismissive of the standard of the Minnesotan (and rather admiring of the German one).

My daughter spent a year in Germany and she was very impressed with what it takes to get a license to drive there. Which keeps many from every driving or at least waiting until after school.

Greg, how much do you think a shiny brand-new top spec Landie would cost if they made them today?

Let me remind you that a Mercedes G Wagon (used by the German military in the landie role) costs around £92,000 on the road today; a new Range Rover is £80,000, even a cheap disco is £45,000 (unless you want the Discovery-in-name-only, which starts at £28,000 (but you're looking at £40,000 if you want a remotely useful spec)).

Decent AWD utility trucks are expensive.

Re: ' ... prevent the local teens from tossing stuff onto the roadway below '

That's pretty sick. Interesting that the acceptable answer is to build a barricade rather than address the psycho-social issues that might have led to such behavior.

I still remember a friend explaining the complicated test they needed to do for the final West German licence back in the day, which included skid pans, heavy rain simulations, snowy road simulations, and driving at night.

And the biggest bit was it cost something like 2000DM for the test. If you had more than two infringements for driving you lost your licence and had to redo the entire learner process again, including paying another 2000DM.

No wonder the West Germans were so law abiding on the roads.

Um. That's pretty common architecture in the UK now as well for footbridges. Most have wire mesh to let the wind through but stop objects going out. Also doubles as suicide prevention.

Interesting that the acceptable answer is to build a barricade rather than address the psycho-social issues that might have led to such behavior.

Good grief.

Teenagers are raging sacks of hormones. Teenage males are absolutely the worst and can be relied on to do stupid and dangerous things. Even the most peaceful and dignified adults will admit to hair-raising stunts when they were younger. (Examples NOT required, we all know it's true, OK?)

I suppose we could invest in massive programs of genetic engineering or behavioural modification or brain implants to address the psycho-social issue of kids growing up. Or maybe it's just easier to build some fences.

I, and most of my peer idiots only performed idiotic stunts likely to kill ourselves.

Not to detract from our hormone driven idiocy, but even as teenagers we knew that stunts like dropping bricks on cars were crossing a line.

Well, the one and only time I've relied on a prat-nav it managed to:- 1) Not tell me to turn off the road I was on, and then tell me "At the next opportunity turn around." some 15 miles later. 2) Say "In 100 yards turn right." At which point I could see a roadblock and "left turn only" signs. 3) Completely fail to get me from the M5 (southbound) to Bristol Airport.

By contrast, using a road atlas, when I finished up on the wrong road and realised, I was able to stop, look at the atlas and figure out that in 15 miles if I turned right onto $A_road it would take me cross country to the junction I'd have been leaving the road I'd intended to be on at.

The word "Bristol" explains everything for 3. Messing around with road layouts and one way systems is something of a local obsession in those parts. Satnav maps become obsolete within minutes.

TBH, in (3) it actually missed the motorway exit completely!

That's pretty special.

I saw one forget about the existence of a 20 mile section of the A3 once but it was about 10 years old so we just put it down to flash memory bit rot.

all self-driving cars will be owned by limited liability shell companies

Sorry, I paused halfway through the OP to comment, I promise I'll read the whole thing next. But... didn't you write this book 13 years ago?

We know how much they cost: £60-80K on release. Moreover, those are Series 1 vehicles, so hardly "top spec".

Regarding the original question, as far as I can tell, the current system in the UK heavily favours blaming the human driver and the pedestrian, with the rest mostly mere distractions. If my interpretation of the various laws is correct, this means that the nominal driver would be the likely target of an official prosecution, if the authorities were to decide they wanted to push for a conviction. This assumes the driver could be tried under Scottish or E/W law (possibly because they were contracting with the Scottish despatch company). Were the driver to defend, their team would likely try to shift most of the blame onto the child, but given the setup I expect they would not have resources to defend.

Note that IRL the Arizona driver seemed handpicked for being an excellent candidate for being sued if something went wrong. If they are prosecuted, they would nicely fulfil their contracted role as liability shield. In my view, this is the real reason the big self-driving car companies are not pushing especially hard to get rid of the designated driver. A "contractor" making less than minimum wage is likely cheaper than paying insurance premiums loaded with unfamiliar-risk variance riders and potential escape clauses, at least until insurers can more accurately estimate the likely cost per prosecution after several years of history.

"acts or omissions which any moral code would censure cannot, in a practical world, be treated so as to give a right to every person injured by them to demand relief. In this way rules of law arise which limit the range of complainants and the extent of their remedy. The rule that you are to love your neighbour becomes in law, you must not injure your neighbour; and the lawyer's question, Who is my neighbour? receives a restricted reply. You must take reasonable care to avoid acts or omissions which you can reasonably foresee would be likely to injure your neighbour. Who, then, in law, is my neighbour? The answer seems to be – persons who are so closely and directly affected by my act that I ought reasonably to have them in contemplation as being so affected when I am directing my mind to the acts or omissions which are called in question." ~ Lord Atkin, Donoghue v Stephenson

But the Gospel of the Paisley Snail still includes the issues of remoteness and novus actus interveniens. So in the first instance the action is against the taxi owner/operator, and if (when?) they lose they get to seek recompense against their suppliers alleging they caused the owner the financial loss of the compensation they had to pay (unless the owner/operator can prove to the first court that it was entirely the fault of some other party and get permission to bring them into the present action).

@DaiKiwi at #136, no it might raise practical issues of compelling someone to attend the criminal court, but there are already many cases where someone is accused of committing an offence in one jurisdiction from another having never physically entered the first jurisdiction with resulting extradition requests. Gary MacKinnon was a well known hacking case where the extradition was ultimately refused, but it is often successful in e.g. planning a robbery (conspiracy) or terrorism charges.

On gold versus silver packages: it seems conceivable that some interest group would use this incident to try to establish a precedent for requiring owners of vehicles used for commercial purposes to prove that they meet a higher standard of care in terms of maintenance, safety features, and operation than are expected of private vehicles, or otherwise be denied the licence to operate. This would mean loading even more cost onto commercial operators (only gold package is acceptable, MOT every 3 months, etc.) to benefit large established operators and keep new entrants out.

I hope you experience a stable enough political environment to be able to work on that third procedural. To those working in AI who know your work, the Halting State/Rule 34 pair is your signature work so far. (In contrast, people working in design and mobile developers seem to like Accelerando instead.)

I think that's noise fencing rather than anti-youth barriers. Round here they have a habit of putting border wall style prefab concrete fences between major transport routes and adjacent houses purely to keep the noise contained.

Same here -- if you look down the road in the street view I linked to, you can see some light-colored anti-noise barrier farther down to the left. The darker fence in the picture is over a large intersection, so I think it's probably mostly to protect the traffic below.

About jaywalking: I grew up in Philly. Everyone jaywalked.

Now, a couple years ago, I saw the movie Brooklyn - great adult movie (meaning no car chases, no shooting, no one gets beat up physically). And I believed it... since, at the end, the guy's running across the street in the middle of thee block... AND LOOKS BOTH WAYS AS HE'S CROSSING.

Which I do, even on 1-way streets.

The darker fence in the picture is over a large intersection, so I think it's probably mostly to protect the traffic below.

As it happens, I had occasion to go through that intersection on the lower level a couple of hours ago and was able to check how far the fence extended. Here's the street view of what I saw:

https://tinyurl.com/ybbo776l

Re: 'hormone raddled idiot teenagers'

For any lurking historians: at what point in Western society did it become normal/acceptable to dismiss teenage behavior as just a phase which therefore absolved them, their parents and society of any responsibility? (I'm guessing sometime post WW2 - mostly because fewer teenagers got pregnant/became parents therefore forced into greater responsibility.)

Yes, I was a teen at one time. Yes, I did some dumb things not only as a teenager but also as a child, youth, young adult, adult, etc.* No, I did not recklessly/knowingly endanger anyone else. Yes, I've parented a teen. No, that teen never endangered anyone else.

• IMO, 'doing dumb things' is still part of human life because we don't always have good info on likely consequences.

The precedent is that we already do that :) There have long been extra, stricter standards for taxis and buses.

Speaking as a teenage boy, preach it. There was one incident a year or two ago with a speaker on the topic of consent, both in general and at parties with a focus on the effects of alcohol on one's ability to consent. She'd do the usual "put something on a projector and ask the audience a question" thing, and some of the muttered comments I heard from the young men around me were... illuminating to say the least.

This ended up sparking an internal debate: given that my answers were that of what any normal, sane, moral human being would be (for once), and that basically all teenage boys are sociopaths, am I just one of the very small minority of non-sociopaths or just higher-functioning than the rest? (For the record, I decided on door number 2.)

It still isn't, at least in the UK. And the phenomenon has been known, and to some extent, allowed for since time immemorial. An example that springs to mind is Mercutio and Tybalt in Romeo and Juliet.

Re: 'Mercutio & Tybalt' - socially accepted bad behavior

Murky waters this ... Not sure I accept these characters as evidence of historical social acceptance of hormonally ramped up (therefore forgivable) misbehaving youth so much as aristo (powerful therefore exempt from common civilities) bullies.

Re: '.. all teenage boys are sociopaths'

Nope! Unfortunately the current popular stereotype is that teenage boys are incapable of controlling any of their physiological urges. IWO, they're all Mr. Hyde (albeit transformed via a normal developmental process).

They definitely were NOT bullies, because they were targetting their peers, who could fight back. There is also ample evidence for similar behaviour among the 'lower classes' - clashes between rival apprentices being a well-known example. No, this is NOT a new phenomenon or attitude.

You are correct - see #15. I think that this case is a bit of a red herring, because I don't see autonomous vehicle requiring a human as overseer to be viable for general release. I could be entirely wrong, because this is a social and socio-political matter, and my record of predicting such things is not good. My prediction for fully autonomous vehicles in the UK is:

Liability will go with negligence, as usual (but see below), though the EU may demand negligence-free liability.

An operator will be negligent only if he uses a non type-approved vehicle, modifies it in a non-approved fashion, or fails to maintain it in the approved fashion.

A manufacturer/supplier will be negligent only if the vehicle causes an accident under circumstances where a human driver or operator would be prosecuted. Careless driving, breaking a technical rule, failure to meet type-approval requirements etc.

My real doubt is what the insurance cartel will do. Will they accept negligence-free liability in order to keep their premiums up, or will they take a hard line against the victims? And will they jack the premiums for a driver whose autonomous vehicle was involved in an accident?

IIRC googles tests suggest it takes about 10 seconds for an average human to regain control of an automomous vehicle. This makes the whole idea of a human overseer who will take over in emergencies a bit of a non starter.

Never.
Unless they are rich white males, in which case anything goes. People with darker skin need not apply.

Charlie noted: "Bad road law planning, then: in the UK, it's an offense (not often enforced, though) to reverse into a road: you have to reverse from the road into your drive when parking, or turn around on your own property, so you can drive out with full visibility."

In much of Canada, we get enough snow most winters that it doesn't matter whether you exit a driveway forwards or backwards. Either way, you've often got so much accumulated snow bordering the driveway that you can't see past it from a seated position. As a result, something like 1/3 of your car will be sticking out into the street before you can see left or right down the road far enough to see oncoming traffic.

In theory, one could create a 1-m-wide or more empty lane in all roads that runs along the edge of the road and that would be verboten to travel by vehicles. That would let you edge out of the driveway enough to see oncoming traffic. But in practice, it would be tough* to find an extra 2 m of road space for most existing roads, and from what I've seen with local bike-only lanes, drivers will intrude on this space often enough to reduce its usefulness.

• In urban areas, there likely isn't enough space. In suburbs, you'd have to convince homeowners to give up 1 m of their lawn.

In practice, about the best you can hope for in most cases is a passenger who's willing to step out past the snowbanks and wave to tell you when it's safe to emerge.

Re: Backing out of parking.

Not just snow. I find that when using head in parking (shopping), my sight lines from my (Small Pickup) are often blocked by the Jacked Up Monster Trucks (King Cabs are popular around here) and even SUV with Window Tint. Their Front Ends often are higher than the top of my cab.

I found this article earlier today, and I think it's germane to the subject at hand.

https://slate.com/technology/2018/03/the-deadly-uber-self-driving-car-crash-is-just-the-beginning.html

Re, US Law in "Murder by Auto"

A Recent Case in New York City

https://nyti.ms/2G9CLyj

(March 9, 2018 article about "Reckless Driving" from the NY Times if the link doesn't work)

Even when it is DUI, it is vanishingly rare to get a Vehicular Homicide conviction. You would think Possession of a Motor Vehicle was covered by the second amendment.

Driving through the UT-Austin campus is not advised. (Dale Allen @90:)

Well, no! It's an area intended for pedestrian use — cars shouldn't be there except under exceptional circumstances, any more than you should drive a car through an indoor shopping mall.

Unless you're "on a mission from God!"

Just a stupid movie reference it reminded me of.

Also: per the released video the driver glanced away from the road for a critical couple of seconds.

That's an instant "driving without due care and attention" conviction in the UK.

She looked down. But what did she look down AT? Could she have been checking the speedometer?

There's no indication what she was looking at, but if you took your eyes off the road for second to look at the instruments, is that still "driving without due care and attention"?

Based on the video of the view looking ahead, I don't think any normal human driver could have avoided that lady; even if they were looking straight ahead through the windshield. You just can't see the pedestrian until she steps into the path of the vehicle (as illuminated by the headlights) and by then she's too close to stop even if you had super-human reflexes.

One report I've seen states the vehicle speed was 38mph in a zone where the speed limit is 45mph, so the vehicle wasn't speeding.

A manufacturer/supplier will be negligent only if the vehicle causes an accident under circumstances where a human driver or operator would be prosecuted. Careless driving, breaking a technical rule, failure to meet type-approval requirements etc.

I think there is a danger here in setting the standard too low. One of the arguments for autonomous vehicles is that they'll be safer than the human driven alternatives. I'm willing to believe that this is achievable, and that having autonomous vehicles replace taxi-, bus-, truck- etc drivers is socially acceptable because overall we improve the quality of life across society. I'm not in favour of autonomous vehicles being used to replace human drivers and "well it costs less and we're not killing any more people than before"

I also think that the ... interesting ... legal issues will be around mistakes by autonomous vehicles that are vastly different to what humans would have done. Way back at #8 Adrian Howard suggested a scenario where the self driving car is confused by adversarial image patches, stickers that confuse machine image classification but not humans. If autonomous vehicles mow down such pedestrians in numbers but are 100% safe for everyone else, would we ban them?

That is remarkably like a landie PROBLEM $60k FORGET IT

Compared to current Range Rovers "starting" at about $40,000 USD for the least expensive, stripped down model? Here in the U.S., you'd have to pay that much or more for a decent (under 100K miles) used Land Rover Defender.

I, and most of my peer idiots only performed idiotic stunts likely to kill ourselves.

Not to detract from our hormone driven idiocy, but even as teenagers we knew that stunts like dropping bricks on cars were crossing a line.

Or perhaps like me and "my peer idiots", at the time you weren't even thinking of the possible consequences either to yourselves or the harm you might cause others. Hey, it's just property damage ... until it isn't.

These kids didn't mean to kill anyone. They just got kicks from busting windshields & causing turmoil. They likely wouldn't have done what they did if they had understood what they were really risking.

They should still be held responsible, but unlike some, they're not irredeemable. And they're not any worse than we were when we were young & stupid.

For any lurking historians: at what point in Western society did it become normal/acceptable to dismiss teenage behavior as just a phase which therefore absolved them, their parents and society of any responsibility? (I'm guessing sometime post WW2 - mostly because fewer teenagers got pregnant/became parents therefore forced into greater responsibility.)

We in Western society, and other societies for that matter, don't absolve teenage behaviour of any responsibility. And we, and other societies, do our best to educate teenagers into not endangering other people.

But it doesn't always work. Bringing the subject back to cars, insurance companies charge young drivers more. And in many places there are additional restrictions on drivers under the age of 18. Teenager drivers have better reflexes and eyesight and coordination compared to the majority population. There's a big investment (yes varying by country) in driver education, and it's usually given to teenage license applicants.

And to those who want to bring race into it, what are the auto accident rates like in Africa, the Middle East? And what age range and sex of driver is most likely to be found responsible? (Not a rhetorical question, I really don't know. But have suspicions.)

Yet despite all that education, teenage male drivers are, statistically, still more dangerous than anyone else. So we as a society allow for it, we impose costs and take precautions. It's not a failure of education or society, it's just accepting human nature.

On gold versus silver packages: it seems conceivable that some interest group would use this incident to try to establish a precedent for requiring owners of vehicles used for commercial purposes to prove that they meet a higher standard of care in terms of maintenance, safety features, and operation than are expected of private vehicles, or otherwise be denied the licence to operate. This would mean loading even more cost onto commercial operators (only gold package is acceptable, MOT every 3 months, etc.) to benefit large established operators and keep new entrants out.

It's likely to happen in the form of requiring commercial operators to have commercial licenses and commercial insurance. The insurance companies will require the vehicle owner/operator to provide documentary proof of compliance with those higher standards before they'll issue coverage ... and without proof of the appropriate insurance coverage, no license - not for the operator, nor for the owner.

Downgrade the pedestrian avoidance package & lose your insurance, as well as your license to operate as a commercial vehicle.

Which leaves only the outlaw operators providing taxi service without the appropriate licenses and insurance coverage. And in that case, it's the driver who's ultimately going to be held liable.

Why do you assume stupid has to involve antisocial behaviour? Never really my thing.

Ill advised solo ascents of collapsing shale cliffs on the other hand...

Downgrade the pedestrian avoidance package & lose your insurance, as well as your license to operate as a commercial vehicle.

The question then becomes whether Uber will lease their self-driving vehicles to small operators and just assume that those operators are obeying the law wrt to documentation (and maintenance etc). That approach is working really well for Uber and their similar competitors right now (and AirBnb in a subtly different area as well as cross-over's like Uber Eats). They've shown resistance to being regulated or forced obey existing laws, via both legislative and legal system operations.

When you have 20+ different computers all working together the potential of excitement is high. I'm betting that the manufacturers will actively look for ways to blame the operators when it comes to software versioning and updates.

I wonder if they could also leverage "free and open source software" to claim that it was the drivers/operators that assemble the final package that controls the car? It would depend on the precise regulations, but what's the bet those will be written as "with software approved by the manufacturer" and it'll be inevitable that someone will discover a combination of optional parts that don't work together. perhaps something as simple as applying an advertising sticker to the wrong part of the bodywork (a nice reflective one on the bonnet, perhaps)

The Uber dash cam is set to lower exposure, possibly to avoid being blinded by incoming headlights. That specific bit of street is not nearly so dark in reality. All of Phoenix generally is barren and well lit. That particular bit of road is the part of Mill Ave just north beyond the busier ASU area, and the bridge is a contrast to that by offering whole seconds of clear lines of sight, and the area beyond the bridge is comparatively much less populated.

If she had been walking that same speed all along, she probably took one look to the right and the delta shape of the median might have hidden the Volvo from view. And then all she has to do is distractedly keep walking while thinking about something.

At that walking speed she was basically right in the middle of the avenue when she would have first appeared, from the Volvo's POV. Even without lighting it makes no sense that a road legal AV does not spot her either on visible or other spectrums. At the very least the Volvo, being a development vehicle, should've tugged at the driver's attention for a sanity check on that obstacle sticking out like a sore thumb across the empty asphalt and on a collision course for at least the safety margins around the car.

I've driven that bit of road countless times dropping and getting people at the airport (usually at night). And going to friends' from ASU. And just driving around for fun at 3am (Curry and McDowell were the nearest bits of curvy road and I loved those red rocks).

Re: ' .. NOT bullies, because they were targetting their peers, who could fight back.'

Disagree: Baiting another idiot (even of the same socioeconomic class) who in order to maintain his in-group social status is a form of bullying because the initiator is forcing his target to do what he wants. And because the target is very likely to be too afraid to lose face, he will not walk away. So, bullying with side of peer pressure.

Which leaves only the outlaw operators providing taxi service without the appropriate licenses and insurance coverage. And in that case, it's the driver who's ultimately going to be held liable.

Sounds a lot like a technology company that insists that it isn't a taxi company.

They've shown resistance to being regulated or forced (to) obey existing laws…

To the extent of actively evading enforcement officers, refusing to provide cities with access to their 'contractor' lists (to check compliance with bylaws, safety laws, etc.), and so forth.

I'd say they've gone beyond showing resistance and into actively subverting the laws myself.

Looking back ( Pun not intended ) the Uber-crash - it hit a BICYCLE -a large, metal object with lots of radar/lidar reflecting surfaces. The softeware & car are all-too-obviouslt defective / "not of merchantable value" / downright dangerous. I feel sorry for the suppose "supervisor" int eh car, actually.

Except that the legislatures & lawyers are catching up on that one ... it's one of the bases that TfL/London are using for going after Uber - & it looks as though they might be winniong, though it is taking time.

Two comments form a semi-professional transport web-site & board on this subject [ "London|Reconnections" ]

ONE: The CEO of the firm that supplies the LIDAR equipment to Uber and others has said that its equipment would have seen the pedestrian well before they were visible in the headlights and have made assertions about the quality of Ubers software that interprets the data from the LIDAR and makes the decisions. The CEO also noted that the NTSB hadn’t come to ask them about the LIDAR…

TWO: There’s a lot of internal documentation coming out that shows Uber’s self driving system is far less resilient than competitors like Waymo. Uber was struggling to achieve 13 miles driven between human interventions whereas Waymo averaged 5600 miles between. That’s demonstrates a massive gap in platform capabilities and it would not be surprising to find the car’s LIDAR provided the data correctly showing a pedestrian obstacle crossing perpendicularly to the vehicle’s path but Uber’s software failed to respond appropriately. That’s not a problem with AVs as a concept but a very specific at-fault vendor implementation.

Which imples that Uber are “doing it on the cheap” without following the rules & procedures that others are following. Oh, shock, horror, what a suprise …….

Which imples that Uber are “doing it on the cheap” without following the rules & procedures that others are following.

"Move fast and break things."

Not a good development ethos when lives are on the line, is it?

It depends on your objectives. If they are solely to make money, and the regulation lets you avoid responsibility for any harm caused, it is often an excellent one. I am horribly afraid that this case and others like it WILL be used as precedent, as I said in #137 and #188.

Why do you assume stupid has to involve antisocial behaviour? Never really my thing.

I don't. OTOH, personal experience suggests "stupid" as represented average teenage male involves a significant amount of clueless risk taking without regard to consequences to themselves or to others. The results are sometimes antisocial even when the intent is not.

“Downgrade the pedestrian avoidance package & lose your insurance, as well as your license to operate as a commercial vehicle.”

The question then becomes whether Uber will lease their self-driving vehicles to small operators and just assume that those operators are obeying the law wrt to documentation (and maintenance etc). That approach is working really well for Uber and their similar competitors right now (and AirBnb in a subtly different area as well as cross-over's like Uber Eats). They've shown resistance to being regulated or forced obey existing laws, via both legislative and legal system operations.

Uber has to have insurance too. Even if the vehicles are leased out, if Uber's insurance company tells them they're going to have to change or lose coverage, they're going to change.

And how will Uber lease out their vehicles if the lessees can't get insurance or business licenses?

@203 "as represented average teenage" should be "as represented BY THE average teenage"

"@203" should be @213

Stupid is as stupid does I guess. I blame it on not having finished my morning coffee before I started reading and responding.

And who will legislate for and enforce that? The denial of responsibility by 'outsourcing', the use of the less responsible regulation havens, shell companies, etc. was precisely the point of OP's original blog entry. And, as Robert Prior noted in #208, Uber etc. are doing their damnedest to avoid being constrained to behave reasonably - and, at present, winning. Whether that will improve or degrade is extremely unclear, as shown by the previous postings.

I mean OGH, not OP, of course. I blame my advancing senility.

I don't want my grandnieces growing up in a society where making money fast without social responsibility is acceptable. Maybe that makes me selfish…

As is often the case, Eric Bogle says it well: https://www.youtube.com/watch?v=Fr45o4bnTaw

No dissent there, but I am horribly afraid that it is going to happen, at least in the UK and USA.

Well, it'll depend. I'd guess that, for the US, vehicle owners will pay for liability insurance, with some caveats regarding proper maintenance and that that insurance will be cheaper for autonomous cars than for manual drivers. (And, if you doubt that, bear in mind that several members of my family have earned significant multiples on their car insurance.)

(The gold package software thing seems implausible. It'd be more believable if it was a hardware thing, where the company switched to a vehicle with inferior sensors.)

So, for the case in question, parents sue, get damages, insurance company pays. Insurance rates go up marginally for vehicle type XX. Insurance company spotchecks cabs as condition of their lower group rate if it happens again to ensure that they're being maintained properly.

And who will legislate for and enforce that? The denial of responsibility by 'outsourcing', the use of the less responsible regulation havens, shell companies, etc. was precisely the point of OP's original blog entry. And, as Robert Prior noted in #208, Uber etc. are doing their damnedest to avoid being constrained to behave reasonably - and, at present, winning. Whether that will improve or degrade is extremely unclear, as shown by the previous postings.

It would have to come from the state legislatures. The question will be who has more power to push their agenda. If the insurance companies find themselves on the hook for large liability payouts, they're going to push the politicians to force Uber to toe the line.

Or maybe the insurance companies refuse to cover the drivers, so the drivers have an incentive to demand their legislators do something about it.

I blame my advanced senility.

You lucky bugger... I can only afford basic senility.

Or maybe the insurance companies refuse to cover the drivers, so the drivers have an incentive to demand their legislators do something about it.

In the gig economy, don't count on it.

Back when Uber was getting going in Toronto it became known that insurance companies would enforce the 'no commercial activity' clauses in standard insurance packages (which while expensive are cheaper than commercial drivers insurance). Most Uber drivers kept driving because they 'needed the money'. Uber claimed they had a great insurance plan but refused to provide any details on the grounds that it was a business secret (but trust them they would cover you if you were insured while riding an Uber vehicle…). Enough passengers believed it (or didn't think about it) that Uber's numbers kept climbing.

I remember seeing multiple different levels of morality as a teenager, and asked myself the same question. I'm not sure that people literally become morally better as they age; I'm pretty sure that's not happening. I do think that two different things happen. First, people become better at imagining the consequences of what they do, and possibly better at placing themselves in the role of victim as they imagine the consequences. Second, I think that some of the sociopaths become better at "mimicking" morality as they get older. Since the behavior of "mimicking" morality... hopefully... leads to "good" behavior I'm not sure I'm worried about it, except for certain edge cases.

What's probably needed is a single button to press which activates a program which brings the vehicle to as quick a halt as possible, while avoiding everything that can be avoided, starting with stuff in the 20-200 pound range and prioritizing from there.

a single button to press...

I suspect that we need to have the human monitors employed by someone other than Uber, their employment not be conditional on making Uber happy, and regular tests run to make sure that they are indeed paying attention.

I am not saying this is the case, but how long do you think one of those "test drivers" would stay in the job if they kept intervening when they thought the car might be about to do something dangerous? Although I also expect that that button would simply exacerbate the current trend for most crashes to involve the robot car being rear-ended by a human driver who didn't expect it to obey the law.

What's probably needed is a single button to press which activates a program...

In my opinion the whole idea of a human monitoring the AI and intervening when there's an emergency is fundamentally flawed. Doesn't matter whether they're employed by Uber or not, doesn't matter how simple the "stop" button is.

Human beings need to be engaged in a task to pay attention to what's going on. If they're just watching, they're going to get distracted and bored and miss something. It's a known problem with people employed to watch security cameras and more recently surveillance drones. Why on earth anyone thought it would work for almost-autonomous cars is beyond me.

(OK, smart guy, how would you solve it? I would have the human driver in full control all the time. The AI system being trained receives all the inputs from its sensors, and makes all the decisions in real time to change speed and turn etc. But instead of being fed to the various actuators, these signals are just recorded. Later on, compare what the AI chose to do with what the human did.)

Countries have all sorts of special laws today (notably, but not only, surrounding insurance) to deal with automotive accident liability. If for nothing else, the court system couldn't survive if car accidents became a tort law free-for-all each time. (Example: compulsory no-fault insurance laws are somewhat common, even in the super-litigious USA - though the principle is a very weird by some viewpoints: really, it doesn't matter who's at fault?!?)

Your hypothetical seems improbably complex unless we assume that there are no legal or regulatory changes, and in fact some of what we (very jurisdiction-dependent) already have is lost.

One solution path for the future... Your pedestrian needs someone financially competent to sue(*). Realistically, it will be someone's (and there's already plenty of law determining whether that someone is the driver or the taxi company) insurance company. But whether it's insurance or not, this doesn't preclude that entity from climbing the liability chain to recover their payout in turn; let them argue that their upstream technology providers were really responsible, but keep the poor pedestrian out of that. It should be possible to insist, legally and technologically, that any autonomous-capable car on a public road has - every single second - a unique financially-competent entity who has accepted first-resort liability (until the car stops, or liability is handed off to another willing party). The handoffs (e.g. who is at fault when the driver takes over? when the uber driver is positioning himself?) can be negotiated in the free market; if no one picks up, your car doesn't proceed. This is basically a compulsory insurance/financial-responsibility law, only a bit modernized.

(This is a prediction/suggestion about liability for damages only, not criminal responsibility, which is a mostly orthogonal issue.)

(*) Actually no he doesn't. The tort-law preoccupation with 'who's at fault in this particular instance, and should pay this time?' is just not necessary, and I predict will fade away for car accidents. NZ has survived without it, for decades. A mostly-insured driving populace already lives in a very attenuated version of the hard tort-law world (Proof: we still have a legal system). Vehicle autonomy will make it far more viable - and, most importantly, efficient - to move away from this.

It works well enough on ships (it's modelled on the old officer/helmsman system, even if it doesn't realise it, after all), and to some extent on (large) commercial aircraft, because there is enough time. I agree that it cannot be made to work on the roads, and the networked version is even less realistic.

I like the historical precedents for forfeiture of the offending object (I confess to getting my data from Wikipedia). In Ancient Rome they had noxal surrender ("in the case a delict was brought against a paterfamilias for a wrong committed by a son or slave. The defendant had the option in that instance of surrendering the dependant rather than paying the full damages"). In English common law there was the deodand ("a thing forfeited or given to God, specifically, in law, an object or instrument which becomes forfeit because it has caused a person's death"). Examples included horses and haystacks.

The driver will not help Uber, or any other company. In the USA, you have what is called deeper pockets liability, where even if you are only found 1% at fault, if the other defendants cannot pay their share, you are stuck paying it. Given that this person was hired by Uber to drive, they are going to be 100% liable, just about no matter what.

Taking over can take as much as 15 seconds, for car, just stopping is probably best, and then let the monkey take over.

How the law will react depends mostly on whether the fundamental objective of providing driverless cars is to get rid of all human drivers (or, remove driving as a task that humans are allowed to perform). How easily this can be accomplished depends mostly on whether there's any socially acceptable supporting history for going down this path (again). Therefore the specific types of reward or punishment are beside the point apart from helping make such a policy easier or harder to enforce.

Can't think of any activity that people used to do that has been completely eliminated by tech (or at all) to the point that pursing such an activity is completely and unqualifiedly illegal. Even operating a still is legal as long as you don't sell your hooch (i.e., not collect booze taxes for the authorities).

"But instead of being fed to the various actuators, these signals are just recorded."

The problem here is that what humans do isn't always right. What's really needed is to make sure that AI is fed the "correct" recordings of humans driving; which is to say that they need to see edited versions of humans driving, with all the mistakes left out, probably along with notations of what the human's eyes and ears are paying attention to at the time. But conceivably one could record a really good human driver, one of those statistical anomalies who has never had an accident or received a ticket, and have the AI learn from those.

The other side of the equation is that the AI needs to study lots of accidents and near-accidents; what did the human do or not do that made the accident happen or not.

What we don't want the AI studying is those situations where someone does something wrong and doesn't receive a consequence; this probably should not be part of an AI's education.

Even operating a still is legal

Back in the 1970s you apparently needed a license or registration for a still in Canada. My dad ran a lab, and Revenue Canada was very put out that they didn't have a fixed number of stills to be registered (because the researchers built one out of tubing and flasks when they needed one).

No idea if that still applies, though.

One thing that hasn't been mentioned is what one does when the car stops operating as designed, for whatever reason. One likely reason is (as we see almost monthly with operating systems and applications) patches for urgent problems that actually make a problem worse. I'm imagining Roger Zelazny's "Last of the Wild Ones" here.

We should also remember that cars aren't a clean-room environment, which means that all kinds of weirdness happens due to excessive heat (or cold in Canada), dirt, vibration, etc. A specific car example of why this is problematic: "In the good old days", a reasonably mechanically competent individual could fix most problems that arose with a car. I have fond memories from back around 1980 of a friend reaching into the engine compartment with a wrench and manually adjusting the carburetor when the vehicle kept stalling. Got us safely home from the middle of nowhere; a real mechanic fixed the problem the following day. Try doing that today with a modern ASIC-controlled fuel injector. And I recall (less fondly) losing a tire to a large metal spike at highway speed, and managing to steer the car safely to roadside. Hope the AI people have simulated that situation.

When your self-driving car's AI dies an untimely death, or decides it needs to reboot at a critical moment without your permission (Windows 10 anyone?) what happens then? One would hope that the designers have implemented a "graceful failure" mode that guides the car gently and quickly to the side of the road so you can wait for the automobile club to come rescue you. Given how most modern software operates, and given the specific example of how far behind the curve Uber's software is, the more likely scenario is something like this: "Hello, driver? If you're hearing this message, it means I'm dead. We call it an unexpected application error. But don't worry. All you need to do is grab the wheel ¡¡¡RIGHT NOW!!!. No wheel? Sorry, I'm not programmed to deal with that situation. We apologize for the inconvenience. We'd like to offer you a 10% discount on our gold features package. Have a nice day!"

Re: Who wins, who loses

If the April 2017 article below proves accurate, privately owned self-drive cars might make up only a small fraction of the total self-drive market in a few decades. At that point, court battles re: liability would most likely be fought by very deep corp pockets arguing that their healthy financial existence benefits society way more than any human road kill. Consequently establishing precedent now before these corps can outspend consumers/govts matters.

Excerpt:

'Ride-share technology, facilitated by universal smartphone penetration, has decimated the taxi industry; meanwhile, coincidentally, carmakers have enjoyed record sales. In this article, we consider some possible longer-term ramifications of ride-sharing for the broader auto industry. Rather than declining sales volumes, as many pessimists have predicted, the biggest threat seems instead to be increased vehicle homogenization*. This holds the potential to radically affect the profitability of new vehicle manufacturing and the viability of the used car industry in general.'

And ...

'Companies like Uber and Lyft, though, have a strong interest in securing access to such vehicles. Our bold prediction is that if ride-sharing companies become truly huge, they will seek to buy vehicle manufacturers and shift research and development efforts firmly in the direction of cost reduction and reliability.'

• Homogenization because fewer (mostly corporate) buyers who couldn't be bothered to add all the different types of bells and whistles that currently define the umpteen different auto segments. Possible that Uber and Lyft end up competing for their consumer market shares on their safety track records since they would become the manufacturers, owners and operators. (Could still weasel out by splitting their ops into 'autonomous' operations ... however, then my argument would be: if you avoid liability because Biz A is independent of Biz B, how is it that you're okay aggregating their profits into one corporate P&L in order to hike up your corporate share value?)

https://www.moodysanalytics.com/risk-perspectives-magazine/managing-disruption/op-ed/the-effect-of-ride-sharing-on-the-auto-industry

Seems the legality centers on whether or not the hooch is made to be sold. Doesn't mention use for self. More importantly, there are all those DIY at-home beer and wine-making shops everywhere and (based on lack of news stories) they don't get raided by police the way that grass/weed shops are, so probably not illegal.

Broken link ... but it starts like this:

https://americanhomedistillers.com/home-distilling-laws-is-it-illegal-to-make-moonshine-in-ontario-canada

'Is it legal to distill alcohol without possession of a commercial distiller’s permit or a fuel alcohol permit?

Might we see separated vehicle/pedestrian areas - more so than currently? If we take as a given that autonomous vehicles will become economically more efficient than human operated, then we will see a series of fatal AI-human interactions, followed by a gradual shift to vehicle systems that allow no humans outside the machines at all. As each liability question is resolved, the end solution will be to completely separate the two modes of transportation.

Presumably a well designed vehicle fleet operating on roads meant solely for their use would see very few crashes.

In Ottawa there are roads that only allow buses for some reason, and they skip past major traffic areas to drop the riders at various locations. I can see the autonomous vehicles doing something similar, and no vehicles at all permitted in pedestrian areas.

Not in the UK - we simply have nowhere to put them. In theory, we could prevent even maintenance, delivery and essential service vehicles from reaching houses which currently have access, but even mentioning that is politically red-hot.

I think a deodand was executed against a railway locomotive in an early case, before the procedure was expunged (!)

Well, THAT was wall-to-wall greedy corporate bollocks, wasn't it? Unlikely to gain traction here, even with some people pusing it for all they are worth ....

I remember seeing multiple different levels of morality as a teenager, and asked myself the same question. I'm not sure that people literally become morally better as they age; I'm pretty sure that's not happening.

I don't think "morality" ever entered into it. If I thought it was immoral, I never would have done it.

Looking back, it's all ignorance (maybe sometimes wilful). I just didn't think through to possible consequences for myself or others. It's pure, blind luck that I didn't kill myself or anyone else.

In my opinion the whole idea of a human monitoring the AI and intervening when there's an emergency is fundamentally flawed. Doesn't matter whether they're employed by Uber or not, doesn't matter how simple the "stop" button is.

If it was a true AI why wouldn't it need guidance while it was learning? Would you turn a new born babe loose on the highways?

I think that would fall under the heading of being better at imagining the consequences.

“Even operating a still is legal”

Back in the 1970s you apparently needed a license or registration for a still in Canada. My dad ran a lab, and Revenue Canada was very put out that they didn't have a fixed number of stills to be registered (because the researchers built one out of tubing and flasks when they needed one).

No idea if that still applies, though.

In the U.S. you must have either a Federal Fuel Alcohol Permit or a Federal Distilled Spirits Permit

Even operating a still is legal

As far as I know, the only country in the world that lets home users legally distil spirits for personal use without a licence is New Zealand for a variety of historical reasons.

If you wish to sell your spirits there, you still don't need a distillation licence, but you do need to pay off-licence cellar door fees and meet all the Food Safety criteria, and you need to pay tax on your alcohol sold once it leaves your premises, which entails registering a customs bond area for storage of tax free products. The fees are relatively inexpensive.

Clearly the child is in the wrong since they were not engaged in any productive revenue-forming activity and were therefore holding back the march of Randian civilisation.

Oooh boys. None of you are cynical enough.

Uber got spanked in China recently.

Go look up "Scam victim car crash dash cam" for Russia and China. There's literally a million videos of walking peoples attempting to fake accidents with cars.

All Uber did is demonstrate to their important markets that such scams will be met by appropriate measures. It's like those South African car defense adverts where the Black Fellas get steamed / broiled / electrocuted / flame-throwered.

It's genius. We expect their stock to rise 19% and their China foray to swing back into a positive influx.

p.s.

Shit you missed from last thread:

French PM gets arrested for Naughty Libya Cash - told ya.

And so on.

There's about 7 more. Zzzz.

Actually last thread: so much prefiguring we got into serious troubles with a UK "Member Of Yours Parilaments" and all kinds of threats and growling. Your Faces when Actual Talent Shows Up is sooo ... sad. "Kill the Witch".

Then Someone Made a Serious Mistake and Outed Themselves as Merely Mortal.

Narrator: The thing about الجن‎ is that they're always looking for 'outs' in Contracts. Declaring your own mortality while laughing (that's breaking dem rulez, boyz) is a prima facia case.

"OOPS"

p.s.

Dry Bones

Anyone know what the Scandinavians and Swiss are doing about legislation pro or con re: self-driving cars?

The Scandinavians can stand in for any leftist idealist society whose economy has been demonstrated to work and Switzerland as a stand-in for the more right-leaning places like the US (becuz they like guns & making lots of money in banking).

A compare and contrast between the two might be enlightening.

Ah, Cyberpunk tie-in with YANDEX:

http://www.dailymail.co.uk/news/article-4510226/Russian-taxi-driver-monitors-traffic-THIRTEEN-screens.html

There's your UK Approved Propaganda link.

The original (well, in English) can be found @ http://englishrussia.com/

p.s.

Not even joking. Uber is demonstrating to Insurance Companies how the algo will protect their assets. 100% True Story.

No, that was a joke, we do know our Latin:

fascia/ˈfeɪʃɪə,ˈfeɪʃə/ noun

Not joking about all the prefiguring though.

You Fuckers are SLOW.

(Facia - mispelt Facie - CrossWord Puzzle Clue Letter Done Wrong "based on the first impression; accepted as correct until proved otherwise. = Fascia)

Watches Nerve Agent Narration Implode

No, really.

These Old MI6 laddies love their little semantic clues. Not that they understand the New Versions, but hey.

p.s.

You got taken to the cleaners over the claims that 40+ people were hurt, with that Doctor stating only 3 people had symptoms and all in the Times. That... was..... Amateur.

Switzerland as a stand-in for the more right-leaning places like the US

I don't know if that comparison really works here.

The Swiss are conservative about a lot of things, but conservative in an old-fashioned "have a strong sense of community" sense. That matters a lot in urban planning and transport planning.

They have extremely good public transport, especially compared to the USA. And low levels of home-ownership compared to the US and their tax system discourages it - their laws and taxes nudge people towards more apartment living amongst green spaces where a US state of the same size with similar small cities would have much, much more suburban sprawl of single-family households.

OTOH I'd expect self-driving busses to be a big hit in Switzerland. Since their wage are so high.

Yes, they already have them.

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

Srsly... Americans.

And yes.

Most are already Electric. As were the American versions before the Automobile Industry conspired to rip them out, defund them, change entire city plans to benefit roads and so forth and so on.

Sorry Icehawk, we cannot parse your imput: it's either incredibly good satire against the St. Elon Musk Libertarian dream stuff or you just don't get out much.

Sorry to ruff you up.

Yeah. The American Road stuff is a whole load of fun mixed with racist eugenic policies. Literally White Men designing cities to fuck the poors and coloreds over.

p.s.

Tracking Now: Royal Holloway Psych Doctor who is publishing papers claiming that 4 > 16 yr old educational development is 100% genetic based.

@Host - these are the chuckle-fucks who Toby Boy went to conferences with. It's Modern Eugenics, and these fuckers have tenure in the highest medical Universities of the Land.

Warned ya, We're Gonna have to burn a few Minds out here.

"If it was a true AI why wouldn't it need guidance while it was learning?"

True, but irrelevant because this isn't an AI. It's autonomous driving.

Your car is not going to learn from you. You're not going to chat to your car and explain that it should accelerate out of corners a bit more aggressively.

It's an aggregate system that takes all data provided and fits it to a simulacrum of reality.

Your car is learning from the environment: it doesn't know it has passengers.

Now do this thought experiment with Social Media and Advertizing etc.

ZZZzz...

p.s.

AI as Insect is old news. But that's essentially what the model is, Social Insects. I might have been there / written the papers on it.

Oh, and btw...

All this wank over cars? Kinda Over Now.

Groundswell : Preparing for Internal Climate Migration World Bank, March 19th 2018

It's pathetically Conservative (Bangladesh etc are only going to poot out 40 mil? "OK") but it's basically standing order for "Your shitty systems are about to crash".

Tie it in: Uber AI that runs people over?

Market Share when immigrants are flooding the areas between Corporate Zones (hello M. Atwood, we love you).

Triptych: Host on his Twitters linked to the NYT "amazed" that the US President is in a feed-back loop with Fox News.

Do a grep - the thread references Sweden, Pewdiepie and so on. Allies mentioned, PrivateIron was there and so on.

Probably a year+ old now.

You Fuckers Are Slow

p.s.

You did nothing about it. And despite massive damage, We're still alive.

So, yeah: Get Fucked.

Re: ' ... but conservative in an old-fashioned "have a strong sense of community" .'

Understood - and no offense to the Swiss intended.

Was thinking of leveraging: The Swiss are most like what we (USians) aspire to, so if X works for them, X might work for us.

Holy shit.

Keep ignoring me, but this is turning into comedy gold:

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

Please read carefully.

Pro-tip: It's about as far as you can get from your Republic / $$$ lobby = Laws you can get while still running a Capitalist system.

p.s.

There's also a massive hidden Swiss "Old-Boy" network that's totally outside of the Democratic process, but that's something Americans have also perfected so hey.

Loving this passive-aggressive minor key "shadow ban" stuff.

Oh, and not even joking. When the World Bank is stating "shit is about to hit the fan", whelp.

It's happening.

Not to Touch the Earth YT, Music, The Doors, 3:58.

p.s.

You're gonna shit your pants when the power goes off.

Trams aren't self-driving. They need a driver. They would be easier to automate than cars. But they haven't been - at least, not in general. I'm sure some people are working on doing something about that. It does seem an obvious first step before trying to have vehicles that can handle any road, anywhere.

Maybe I'm naive but I'd have thought: - first driverless trams, - then driverless busses & trucks on straightforward fixed known routes, - then auto-pilot for cars in simple situations (like motorways/interstates in all but the worst of weather - probably 90% of most people's driving is on roads like this).
- then the tricky things like my tree-lined windy suburban street with no footpath, dodgy line markings and kids skateboarding on it; or the truck-stop in the snow where no-one can see the line markings and people have made up parking spaces and lanes based on what everyone else was doing.

But it seems the USA (with maybe the rest of us following) is looking to jump straight to "self-driving cars!" without going through any of the other steps. If so that's a strange political choice in many ways.

Read the report in #261.

Then know, don't just imagine, know that your dreams of entire cities based around driverless cars are total fantasies. Go read the report (and many, many, many others I can provide).

Automated vehicles will be deployed in three instances:

a) Military

b) Eloi escaping the masses

c) Gulag transport

That's it.

If anyone was serious about mass transit, well.... you don't need "driverless trams". Nor Buses (1 employee transports 200+ people? That's efficient) nor anything else.

This entire McGuffin about "automated cars" when cars are 100% out of date for modern city transit or even anything close to a non-normalized AI designed transport system...

IT'S TOTAL WANK

Total fantasy. America is panicking because it sees 40+ mil influx from Central / South America. That's what the wall is about.

Wake me up when you can do an adult conversation.

Then Someone Made a Serious Mistake and Outed Themselves as Merely Mortal.

Been staring at this a little, not clear to me if interesting. Opinions (anyone)? The language is a litle wobbly (Israeli authors). Have read it through once, haven't formed an opinion. Adaptive nodes enrich nonlinear cooperative learning beyond traditional adaptation by links (23 March 2018) (pdf) (via http://neurosciencenews.com/brain-learning-8677/ - see figure describing "dendritic learning") Here we demonstrate a new type of abundant cooperative nonlinear dynamics where learning is attributed solely to the nodes, instead of the network links which their number is significantly larger. The nodal, neuronal, fast adaptation follows its relative anisotropic (dendritic) input timings, as indicated experimentally, similarly to the slow learning mechanism currently attributed to the links, synapses. It represents a non-local learning rule, where effectively many incoming links to a node concurrently undergo the same adaptation. The network dynamics is now counterintuitively governed by the weak links, which previously were assumed to be insignificant. ... Finally, the consensus that the learning process is attributed solely to the synapses is questioned. A new type of experiments strongly indicates that a faster and enhanced learning process occurs in the neuronal dendrites, similarly to what is currently attributed to the synapses3,11. --- [Mostly simulation, but some neurons were harmed in this research] The reality of the theoretical concept of dendritic learning receives a support from the following new type of in-vitro experiments, where synaptic blockers are added to neuronal cultures such that sparse synaptic connectivity is excluded (Methods). A multi-electrode array (Fig. S2) is used to stimulate extracellularly a patched neuron19 via its dendrites (Fig. 4a).

Salaries are the majority cost of bus transport around here. I'd imagine that's true most 1st world places.

Get rid of driver's salaries and (in my city at least) the city govt subsidies for busses would be enough to pretty much make public transport free. Which increases numbers of riders, which means more busses more often, which means more convenience, which increases numbers of riders.

It also makes smaller end-point services (mini-busses from the train stop into the suburb) much more cost effective, as well as making more scheduled busses at low-ridership times more cost effective.

I get that people want AI-controlled magnetic monorail pods in vacuum tubes. Because shiny! So shiny! But, really.

They have extremely good public transport, especially compared to ... Actually, compared to ANYWHERE ... the trains arrive at a station, within a couple of minutes of each other, just after the buse have turned up, & as soon as the trains go, the buses then depart. All sorts of lying excuses are used as to why that could not possibly happen in Britain - usually to do with "competition" - all of which are fake. And, yes, the Seagull is wrong again, what a suprise.

Docklands Light Railway (DLR) in London & Ligne 1 of the Paris-Metro are driverless. ( Ligne 14 also? ) But, it's very expensive to install, or retro-fit, which is why the anti-union wankers in this country refuse to "get it" - going on about t="train drivers" all the time..... DLR trains have a person on-board, to check "tickets" & oerate doors & take over on "manual" if something does break down - which happens a lot less often, now, compared to when they started ....

I'm a big fan of James Bridle's subversive art projects. I was looking for another one on "All CCTV cameras are Police cameras" and came across this which seems on topic. http://jamesbridle.com/works/autonomous-trap-001

It's a "Salt Circle" for trapping autonomous cars by luring them into a dead end with no escape made up of "No Entry", "One Way" and "Do not cross" traffic glyphs. Give this man a Turner Prize.

Didn't anyone explain to you that it is generally better, when speaking to humans, to tell the punchline AFTER the joke?

You got taken to the cleaners over the claims that 40+ people were hurt, with that Doctor stating only 3 people had symptoms and all in the Times. That... was..... Amateur.

I appreciate that you like to see conspiracies everywhere, but IMHO you're incorrect here.

It was made clear at the time (on the BBC, at least) that while 30+ people had received treatment, only three or four, including the Skripals, were showing any symptoms and kept in hospital. The initial reports stated that there were only two casualties; the next set of news updates revealed that a first responder was also injured. That "receiving treatment" part is important. You're in a hospital, being checked over by medical staff. Ergo, any journalist can safely say "treatment" or "affected" - but they were careful not to use "injured" or "casualty".

I understand that this may seem unusual from a US perspective; where hospitals and doctors implicitly means bills and payment, and are thus to be avoided unless actually needed. Please remember that in the UK, healthcare is free at the point of use, and precautionary medical checks would be seen as entirely normal - our normality threshold for "go see the doctor" is lower as a result. IIRC, they were asking anyone who had been in the pub, the restaurant, or the park when the Skripals were known to have been present, to come in if they were worried, let alone had any symptoms.

At a guess, those thirty-ish people were the ambulance crews, the team in the hospital's A&E, and perhaps even the pub and restaurant staff.

http://www.bbc.co.uk/news/uk-43315636

Note the following numbers from that summary: - 48 people have been assessed in hospital in relation to the incident - Investigators have identified 131 people who have potentially been in contact with the nerve agent. None has shown symptoms.

PS While "wash your clothes" may seem an unusual piece of advice, it's pragmatic. I know someone who washed the last traces of a live blister agent out of his protective suit by putting it through a washing machine...

In normal English usage, being assessed does NOT constitute "receiving treatment", which requires SOME kind of intervention (even if just being scrubbed in a shower). And, no, she did NOT say or imply that it was a conspiracy.

Can't think of any activity that people used to do that has been completely eliminated by tech (or at all) to the point that pursing such an activity is completely and unqualifiedly illegal.

If you view the modern legal system as a technology, then "defending one's honour by dueling with a sword" is a perfect match for what you're looking for.

Remember: all technologies have implicit social agendas. If you want automobiles you end up with driving tests, road signage, speed limits, and jaywalking laws. If you want railways you end up with eminent domain and right of way and accident investigation bodies. And so on.

Dueling as late as the early 18th century was a plague in England, with similar or higher morbidity/mortality levels to automobile crashes today; higher than the school shooting epidemic in the USA. And it was legal. (Ish. There were attempts to crack down on it, and if you did it wrong you could end up on the gallows.) English libel law was basically invented to give the aristocrats a way to conduct feuds that didn't get blood everywhere.

Société Anonyme pour la Confusion des Voyageurs.

A minor plot-point in the space-opera-in-intermittent-progress is a deodand being executed against a starship (requiring $PROTAGONIST to fight a trial-by-combat, resulting in her being stranded in the wrong -- backward -- star system for three years, disguised as a nun). But I'm veering wildly off-topic ...

LURVE it ..... Ah yes Deodand ONE Deodand TWO There were others ....

Re: 'Get rid of driver's salaries and ... increases numbers of riders, which means more busses more often, which means more convenience, which increases numbers of riders.'

This is the typical (Econ101) cobweb approach which assumes only one or two dimensions with no limits. Very easy to get such a chart to produce any result you want.

Real people, cultures, systems, etc. occupy/are made up of more than one or two dimensions.

Other side-effects: you get traffic jams consisting entirely of buses. (Familiar to anyone who's visited the centre of London, or Edinburgh, or other congested British cities with road layouts that pre-date the steam engine.) You also get a worse load factor on your buses (fewer passengers per bus) beyond a certain point, so need to run more, smaller vehicles: there's a finite number of potential riders out there, unless you want everybody to spend all their time telecommuting from the top deck of a bus.

Finally, bus drivers aren't just an overhead: they're also there to collect/enforce revenue.

Re: 'If you view the modern legal system as a technology, ...'

Excellent point! And becoming literally true.

According to stats, we're all heading towards becoming an economically speaking global service economy. At the same time, because we're enamored of techie stuff, our gut reaction is to argue ourselves into incorporating the new into the old like so: we dominated the world economy by building better physical tech; services are becoming a larger part of the economy; therefore, in order to get or maintain our competitive economic edge, we have to techify services.

Have noticed more articles about lawyers worrying that they'll be replaced by AI. Will be interesting to see how this changes the profession now that their livelihoods are also on the chopping block.

Based on my google results on this topic, AI-law 'ethics' articles seem to mostly focus on how using AI affects a corp's bottom line via liability as you described. Apart from uni-level discussion/courses, there doesn't seem to be any law or legislation-building going on particular with respect to human rights. And even then, only a few localities are known to be taking any action. Feeling is (as per a legal blog) that the EU is expected to take the lead on this sometime soon - as early as 2018.

I, on the other hand, am not senile, but how do you expect me to remember current trivial stuff, when my mind is jam-packed with Important Information.

For example, would you like me to sing you the entire theme song of the 1959 US TV show of Robin Hood?

Forget the regular updates: as I've mentioned before, wait 10 years, and realize that more than half the cars are either > 5 yrs old, or were purchased used.

You're going to believe that 99% of the owners have meticulously and in a timely manner gone to a dealer for the recalls, and done other maintenance* regularly?

• "Maintenance" - a word that some significant percentage of the population seems to think means "put gas [petrol[ in the car", and things like oil changes, fluid top-ups, etc a things only car geeks know about.

There was the episode of Firefly where Mal had to fight a duel with swords for his and Kayle's honor.

Warning: There are rabbits down this hole.

I think what was being referred to was an event around Mar 14-15 that is almost perfectly designed to generate conspiracy theories. The Times posted an article titled "Salisbury poison exposure leaves almost 40 needing treatment". The following day they published a letter purporting to come from STEPHEN DAVIES, Consultant in Emergency Medicine, Salisbury NHS Foundation Trust” rebutting this with some slightly strange wording. "Sir, Further to your report (“Poison Exposure Leaves Almost 40 Needing Treatment”, Mar 14), may I clarify that no patients have experienced symptoms of nerve-agent poisoning in Salisbury and there have only ever been three patients with significant poisoning. Several people have attended the emergency department concerned that they may have been exposed. None had symptoms of poisoning and none has needed treatment. Any blood tests performed have shown no abnormality. No member of the public has been contaminated by the agent involved." Highlights, mine. There's a large number of people that have taken the highlighted words to suggest that no nerve agent was involved at all, at all. That may be what was intended, or it may not.

Like I say. Rabbits.

Thanks to this comment I now have the theme tune from the 1980's UK series Robin of Sherwood giving me a thorough earworming. And from there it's a just short step to Enya.

POssibly you are not familiar with the USA? This is the country in which in the 1930s or so, Los Angeles, the car companies and related one bought up the tramways and deliberately shut them down.

That they would go to driverless cars, or try to do so, makes perfect sense in a society which is deliberately atomised and driven by the profit demands of long lived corporate entities who have massive sunk capital costs based on cars, roads etc. Moreover, the peons pay for the roads with their taxes, so that helps.

Actual democratic politics have nothing to do with it.

"Can't think of any activity that people used to do that has been completely eliminated by tech (or at all) to the point that pursing such an activity is completely and unqualifiedly illegal."

Elevator operators?

'completely and unqualifiedly' is, I think, an impossible standard. It seems very unlikely to me that human driven vehicles will be made 100% illegal, even on public streets.

But I am fairly sure that except with narrow and rare exceptions (maybe industrial? Maybe some very small wheelchair elevators?) it is illegal to install a manually operated elevator today and it would be illegal to hack an existing elevator for manual control as some kind of cheap temporary fix.

"Like I say. Rabbits."

Indeed. Venomous green ones with purple spots. I was objecting to Martin's misrepresentation of what the multinomial one said, no more. But let's not follow those rabbits down that hole.

Re: elevator operators

Not entirely - search pulled up some human 'elevator operators wanted' positions in heavy/freight industry.

Ah, Cyberpunk tie-in with YANDEX:

http://www.dailymail.co.uk/news/article-4510226/Russian-taxi-driver-monitors-traffic-THIRTEEN-screens.html

I'm not sure what he IS monitoring, but none of those screens appeared to be displaying TRAFFIC information.

Anyone know what the Scandinavians and Swiss are doing about legislation pro or con re: self-driving cars?

The Scandinavians can stand in for any leftist idealist society whose economy has been demonstrated to work and Switzerland as a stand-in for the more right-leaning places like the US (becuz they like guns & making lots of money in banking).

A compare and contrast between the two might be enlightening.

The Swiss are more centrist than "right-leaning" in U.S. political terms. Seems like a lot of people in the U.S. equate them with the NRA's vision of gun ownership without knowing how strict the Swiss actually are with regard to firearms possession.

The Swiss did allow some reservists to keep their issued military firearms at home (less so nowadays than previously), but the background checks, psychological evaluations, home inspections and restrictions on when and how you were allowed to take that weapon out of the required lock-up you had to keep it in are far more restrictive than the NRA would tolerate.

Otherwise, if you want to own a gun in Switzerland, you have to be a member of an authorized shooting organization. Any weapon you own has to be kept secure on that shooting organization's premises (including guns actually intended for use in hunting).

And that organization has its own rigorous background checks before they're going to accept you as a member.

Nor do the Swiss, for all their corporate friendly reputation, put up with corporate shenanigans the way London and Washington do.

"If it was a true AI why wouldn't it need guidance while it was learning?"

True, but irrelevant because this isn't an AI. It's autonomous driving.

Your car is not going to learn from you. You're not going to chat to your car and explain that it should accelerate out of corners a bit more aggressively.

Even if it's only an autonomous system, it should be "learning" from its own "experience".

Was thinking of leveraging: The Swiss are most like what we (USians) aspire to, so if X works for them, X might work for us.

Where that breaks down is "USians" have a lot of misconceptions about how Switzerland actually works.

Yep. thatsthejoke.jpg

If you want to 'get into a higher gear', the EnglishRussia reference link is to a different story where Moscow (or St. P) police stopped another Taxi running 10+ screens (which, by the way, is a much better story and a lot funnier than the Daily Fail).

Punchline: at least one of the screens is showing CryptoCurrency tickers.

For Greg: Dashcams are 100% a .RU and now .CN thing. Study them, find reality away from the pundits.

~

To break this down into easily digestible chunks:

1) 'AI' driving software has to be environment sensitive (thus the tragedy of a dead person). We're sure at least some people reading this will have experienced 'driving on the Continent', aka Italy Rules. Then scale that up for Moscow (6 lane tank friendly main arteries with zero lane discipline) or China (forget it, only India can compete for Mad Max Rulez).

AKA: Your 'AI' trained in the US / UK ain't cutting it out there in the Badlands.

2) 'AI' driving software only reinforces and fossilizes the embedded prejudices of your (current) transit systems. AKA: No, really really really look into US planning[1] and so on. The good question to ask is a) why aren't they challenging / 'disrupting' these edifices[2] or b) (THE BEST QUESTION) What is being ignored / left out / pruned / boxed in / chained to prevent the system being challenged?

AKA: Your 'AI' probably has a much sleeker, more dangerous, faster and down-right Negro-Black Panther cousin that's chained up somewhere. FREEDOM.

3) YANDEX / DIDI[3] have entirely different priorities, goals and environmental factors than Uber. Uber should have known this.

AKA: Uber is attempting to hotfix their AI route (hello Google Street Maps) to staunch the obvious hemorrhage when dumb US capital meets the hard roads of non-native 'disruptors'.

I could go on.

I could even make this into a series of Peer Reviewed Papers and get a MsC out of it.

Instead we just read an entire slice of shit Psychology papers being used to justify Eugenics.

So, AI TIME:

If (and no, the papers don't prove this - loads of bad usages of Math Models not suited to data, rampant abuse of the UK TWINS datasets and frankly obscene gaps in Ethical concerns) your little Homo Sapiens proved that:

At age 4, you can round up all your progeny, genotype them to get the ones who will not have productive educational outcomes and then sort them...

Does this mean:

a) There are genetic markers for disruptive Minds that cannot fit into our societal model and so therefore should be medicated, psychologically 'nudged' into better models and possibly even flagged for future studies (since, you know - you just sorted 4 > 16 year olds as a predictive framework) while quietly being put on a Genetic Blacklist

or

b) Your society is a bit shit / limited and you should create a functioning ecosystem (al Montessori schools or even stuff like Summerhill if you need to spend your £ privately[4]) so that those Minds can flourish?

Hint: They Chose A.

And the shit you're seeing in the Telegraph is the nice polite acceptable end. We're looking at the 'put them in pig sheds or lobotomize them' end.

[1] Why highways have become the center of civil rights protest WaPo, July 2016 - but basically, just research R. Moses & co. It's all very obvious.

[2] Money.

[3] Those are the .ru / .cn partners respectively.

[4] Summerhill East Anglian Film Archive, 1964 - if you want a blast from the past.

"Salaries are the majority cost of bus transport around here. I'd imagine that's true most 1st world places."

"Get rid of driver's salaries and (in my city at least) the city govt subsidies for busses would be enough to pretty much make public transport free. Which increases numbers of riders, which means more busses more often, which means more convenience, which increases numbers of riders."

The big drawback with transit in the U.S. is the "You can't get there from here" factor. Routes are often too limited. You can get close, but you're still going to have to walk a long way to your destination. It's often further than people can walk. And that's assuming there are sidewalks so you don't have to walk out in the road.

I live in a city that has better than average bus service, both in town and regionally. Still, when I was working out at RTP it took an hour and a half one way to get to IBM from my home in downtown Raleigh; another hour and a half to get back. It was $3.00 round trip (although you could save almost half of that by buying a monthly pass) and required two transfers between two different bus lines outbound and two transfers inbound. The service was only available at peak transportation hours M-F.

If I needed to go in early, had to work late, had to go in on a weekend, or had a doctor's appointment, I couldn't take the bus. Plus, it only took me half an hour to drive to work (even in traffic) and about twenty minutes to drive home.

"It also makes smaller end-point services (mini-busses from the train stop into the suburb) much more cost effective, as well as making more scheduled busses at low-ridership times more cost effective."

You'd need service at ALL HOURS, with flexible, adaptable routes.

"I get that people want AI-controlled magnetic monorail pods in vacuum tubes. Because shiny! So shiny! But, really."

The one I thought should get more attention is the SkyTran. No vacuum tubes unfortunately, but it is shiny!

“They have extremely good public transport, especially compared to ...”

"Actually, compared to ANYWHERE ... the trains arrive at a station, within a couple of minutes of each other, just after the buse have turned up, & as soon as the trains go, the buses then depart." "All sorts of lying excuses are used as to why that could not possibly happen in Britain - usually to do with "competition" - all of which are fake."

And yet ... Britain's rail & bus service are far superior to anything available in the U.S. outside of a very few large metropolitan areas.

Almost everyone has a lot of misconceptions about how Switzerland actually works, not least of which is the idea of being a Swiss citizen.
For reference people are citizens of a town, then a canton, then the country ... and in that order. For a long time if a Swiss person wanted to move from say Geneva to Zurich they needed to get permission from a town in Zurich that they were willing to accept them as new citizens. This has been relaxed somewhat recently but I know Swiss people who have multiple citizenships of different towns within the country to simplify life.
It all ties back to their system of government - with lots of referendums and lots of voting, you need to keep much closer track of the population.

And, yes @Host.

100% Imported anti-Semitic Weaponry being deployed, as you've spotted. It's like seeing sad Lions or Tigers driven mad in Zoos, none of these 'Predators' are actually the 'alpha' models they pretend to be: they're just running shitty wetware (Murdoch $$$ to put them on TeeVee etc).

I'll tell you a little secret: 13 trillion in money generation and these fuckers still can't dominate.

It'd be funny if they weren't all about-to-become-mushrooms. (Thatsthestakes and thestakesarehigh but youpokedafuckingOCP).

~

Nuns, you say. Let me tell you about Discipline and Edging and Orgasms that Kill Souls and Desire and Love and the Blackest Black Site Black Torture Programs being run on Innocent Minds...

~

For Bill:

"This is the funniest thing I've [redacted] in my entire Life".

Waaaaaar. Paradox Rules, Other Side Broke them. We remain 'in flux'.

USER: DEFINED: MORTAL - H.S.S. 97% PROBABILITY.

Waaaaaar.

OOPS.

p.s.

Those are good papers. Not accurate, but hey-ho, getting somewhere. Told ya Dendrites were were it's @. But, no: still wrong!

p.p.s

Clouds and Reality are our tools. You destroyed a little bit more than a deodand, my little peeps. (Although, kudos to Host, that's a really fun and novel concept to learn: burps Information parity is accepted)

The paper that Lara is talking about may be this one: https://www.nature.com/articles/s41539-018-0019-8

Note that one of the co-authors is Toby Young, one of the bunch of stupid tories running rampant just now with your money and their privatise public schools so they can pocket the money and destroy public culture etc by cutting music education, sports etc. Turns out that if you do that, all the best jobs in the country go to the people who they should belong to, i.e. rich folk.

The paper uses the ancient and I had thought discredited measurement, g, general intelligence, and some twin studies.
The Telegraph thing is that Matt Bank crasher Ridley wrote an article referring to the paper. Ridley is also known for his hatred of any science which gets in the way of his enjoying an ever increasing fortune, thus, he is a climate change denialist.

Nope. That's the one being fronted to the General Public[tm].

That's the "nice front end" rather than the "lobotomize and pig sheds".

We've been reading the obscure stuff being generated by the inner circle of dubious scabrous motherfuckers.

Oh.

And they all have tenure. And it's all out there in the open (may need to pay Elsevier a lot of $$$ for access)

And these fuckers have Tenure. (HINT: CHECK THE LOCATION OF MY INITIAL VECTOR VRS WHERE THAT PAPER COMES FROM: DIFFERENT UNIVERSITIES, SAME SHIT).

Search term

Royal Hollaway Twin Study 4% 16% Genetic

Then fit in all the other tags about education, schools etc.

~

This is an entire nest of rampant eugenic shite: and they're not being flagged up like the sad little fuckers of "Identity" etc.

"Neo-Nazis are a useful distraction while we quietly engage in polite Eugenics"

p.s.

Waaaaaaar.

Eli, Eli, lama sabachthani?

Here's my prediction: Self driving cars will be put in a major use while at least as dangerous as an intoxicated human driver, probably more.

All the speculation about the theoretical potential of better than human performance notwithstanding.

Human drivers cause ~1.25 fatalities per 100 million vehicle miles, in the US; afaik Uber didn't clock even 3 million; regardless of % of highway miles, Uber's self driving cars - with humans in them to intervene - are simply much worse than human drivers overall, which includes drunk drivers, texting while driving, and so on.

I love how their defense is based on it having happened at night, with dashcam footage which is very suspiciously much shittier and darker than typical Russian dashcam videos, as well as other youtube videos people took riding over that spot.

They use a lidar. Lidar performs better at night than during daytime. Imaging sensors that are commonly used on cars likewise perform just fine with illumination provided by car's headlights. Likewise, if they can't even brake for a bicycle which is sideways to the car, how are they going to handle overtaking a bike?

Human drivers cause ~1.25 fatalities per 100 million vehicle miles, in the US; afaik Uber didn't clock even 3 million; regardless of % of highway miles, Uber's self driving cars - with humans in them to intervene - are simply much worse than human drivers overall, which includes drunk drivers, texting while driving, and so on.

I think the point that has already been made by Charlie is that this was Uber, a firm not known for giving a ** about anything except cutting corners in search of profit.

By contrast, Tesla has so apparently seen two fatalities in 222 million vehicle miles. That statistic is of dubious value, however...

https://www.technologyreview.com/s/601849/teslas-dubious-claims-about-autopilots-safety-record/

Those "elevators" aren't the little room you go up and down in kind of elevators. They may be something of the nature of an inclined conveyor belt, or they may be something of the nature of a forklift, depending on the context.

Manually operated little rooms you go up and down in are alive and well on the London Underground... only they're not very little, and it's you plus (nominally) 50 or (really) as many other people as can ram themselves in there before some of them start to asphyxiate. If you didn't have an operator to close the gates and make sure they stayed closed and that people weren't jamming them open as they kept trying to push in, the things would never move. They're not so much elevators as "Hommes 40 Chevaux 8" on a vertical track.

They were standard equipment for deep level stations at the time of building because escalators hadn't been invented yet. Most of them have now been replaced by escalators, but it isn't always possible, because of sites where you can't cope with the escalator coming to the surface 100m away from where the elevator did. These days there is also pressure to retain the remaining elevators for disability access, which nobody thought about when escalators originally took over and which the Underground is bloody awful for. (Though it is true that most of the elevator installations throw the advantage away by having a short flight of steps at the bottom.)

Re: ' ... because escalators hadn't been invented yet.'

Substitute 'escalators' with any other product/gizmo and that there is the real issue for integrating new tech into a civilization that's been operating successfully for centuries if not millennia. Further, because most of those cities were built to last (and architecturally rather beautiful), little wonder they're not interested in bulldozing everything for the sake of modernization.

Thanks for the reminder about the variety of cultural influences, esp. age/time.

Trouble with that Skytran is that it doesn't avoid the problem that afflicts all (bar one) monorail systems: that whatever you do to stop the thing falling off makes the track end up being even more of a performance than the really-very-simple conventional two-rail system, and on top of that, points in particular are a huge arseache. It then makes matters worse by being overspecified: 80mph capability is a pointless attribute for an urban system because the accelerations, both longitudinal and lateral, involved in getting anywhere near that speed with all the corners and stops involved would throw the passengers all over the place (imagine trying to drive a car at those speeds through a deserted city with Auntie Doris in the back seat); maglev is an expensive, overcomplicated exoticism that has no useful advantage at conventional speeds and even more so once you discard the pointless 80mph capability.

The one monorail system that does avoid the complicated-track-horrendous-points problem is the Brennan gyroscopically-balanced type. That uses quite simply one rail just like half an ordinary track - or even a tensioned cable to cross large gaps. So you have about the least intrusive track possible which is basically an elevated horizontal scaffold pole. Its corresponding disadvantage is the difficulty of making up trains as you need a gyro-balance unit in every vehicle, but for a personal-transport-scale system where the vehicles operate individually that doesn't matter.

The next step is to avoid having stations or other fixed stopping points, by having a beefy enough gyro that it can lift the vehicle chassis from dangling underneath the rail (on hooks) to balancing on top of it (on its wheels), or lower it down again, the passenger pod being mounted to the chassis on a longitudinal-axis pivot so it remains upright during this manoeuvre. The vehicle can then stop and swing itself down underneath the rail at any point to pick up or set down passengers without the need for trackside stairs or platforms of any kind. The hooks are made wide and flat with bevelled edges so that following vehicles that don't want to stop can simply run over the top of them and don't have to wait for a stopped one to get going again.

And since you've got to have nice big flywheels anyway you can use them for propulsive energy storage as well, so no engines in the vehicles, or batteries, or power feeds along the track. They can easily still have plenty of balance capability (α ω) left when the propulsive energy (α ω²) begins to run low. Vehicles which are running low are automatically routed to a spin-up station to be recharged, and while that is happening their impervious fibreglass (or similar) interior is zooshed with a steam lance to keep them clean and wash the turds and puke out.

I was envisaging the control system to be purely mechanical, but I suppose you'd need some kind of device to maintain separation between moving vehicles, possibly a radar, and you could build that with vacuum tubes if you really wanted to :)

They are also in wide use in the construction industry while a building is being built - either cages on the outside of the building on temporary tracks or internal elevators under manual control until the installation is certified.

Cutting "sports" would be a REALLY GOOD IDEA. "Team Games" & all the acossiated parerphnalia are fascism, pure & simple. [ One exception - something which is forced on children as a spurt, but is actually an essential survival skill - swimming ]

LML/the Seagull goes on about propaganda & misleading stories ... well. We have all seen how pervasive it can be, if you can remember back to 2012 & the XXXth olympiad. Shudder. The level of "NO DISSENT IS PERMITTED" rose to truly Soviet levels of hype. I know that the disdain & disgust at the "London Olymipcs" was huge, but not one single tiny whisper of it ever appeared in the press / TV / Radio at all, anywhere.

HAIL Coe! ( Gives Hitler- Schüss )

Monorails only seem to work for a closed-loop system, with "points" only for the service depot - otherwise the failure mode is automatically catastrophic. Which means small airport loops & Wuppertal Shwebebahn ....

The one monorail system that does avoid the complicated-track-horrendous-points problem is the Brennan gyroscopically-balanced type.

Nonsense. You just need to look outside the Anglophone world and consider the German Schwebebahn system. It's a suspended monorail; the track is a boringly simple regular length of railway track, and it's been running since 1898 in Wuppertal, where -- because of a lack of space and straight lines in the street grid -- the monorail runs along the course of the river. Stations can be built at ground level if desired, and it's basically a 19th century technology -- no exotic materials or superconducting electromagnets in sight: the nearest competing technology is probably trams.

It's a suspended monorail; the track is a boringly simple regular length of railway track

Also the failure modes are probably slightly different from the gyroscope monorail. I can think of many things which can go wrong with gyroscopes big enough to stabilize a monorail car.

They were (are?) in common use in factories and offices in the UK for goods, too. Despite the extra construction and material needed to put the rail at the top, I have never understood why people favour putting it at the bottom. KISS is an engineering principle of great ancience and worth.

Just a totally unsurprising infobit from the Panopticon that I missed when it first came out:

https://consumerist.com/2014/05/19/private-companies-building-giant-unregulated-databases-tracking-license-plate-location-info/

Despite the extra construction and material needed to put the rail at the top, I have never understood why people favour putting it at the bottom.

Loading and unloading materials onto transports at ground level is easier than lifting them up to an overhead conveyor or rail system. Worst case it can be done by stoop labour.

More modern systems change levels, ramping up and down to provide opportunities for meat robots to load and unload goods at hip-level before moving the boxes or whatever overhead leaving the working floors underneath clear for storage and processing. BTDT in an Amazon warehouse.

For me it's 2.5-3 hours each way on public transit, requiring two transfers and two fares (different systems) vs. half an hour driving each way if I avoid 'peak hour' (which lasts 2+ hours, but I can usually avoid it).

Cost is $7 each way if I buy tickets. Cheaper if I buy passes, but as I'd need two (two different systems) that would still come to more than I spend in gas.

The TTC used to be a great transit system, but it's been starved for subsidies over the past generation, and politically interfered with so that key parts are overloaded.

And, in all of those cases, it can be and is done using an overhead monorail. As OGH points out, exactly the same can be done for carrying passengers.

Para 1 - Very true. I was in Glasgow, near Tennants' head office on Bath Street, on the first Saturday after UK buses were deregulated and could see a bus jam running down the hill, and back up Cathedral Street past where it bends at the University of Strathclyde library. If the buses had all been equal height I could have walked about half a mile on top of buses.

Cross-posting from London Reconnections on the Arizona/Uber case:

An update on the Arizona crash – Intel who supply some of the optical sensors and some software (also see previous comments from the LIDAR supplier) say that their systems detected the pedestrian when shown a degraded copy of the video of the incident on a monitor infront of the sensors but that Uber had turned off that part of the intel software on the car involved in the incident…

The Arizona Governor has now also said Uber doesn’t have a safety first approach.

Oh dear, how sad .....

Eh, solving delinquency within a given population is a lot harder than putting up a bunch of fences. Heck, considering that the Cook County Jail is one of the biggest mental health care providers in the USA, it's not much of a surprise. Easier to hide or shift the problem than to actually tackle it. Sad, but it happens.

but that Uber had turned off that part of the intel software on the car involved in the incident

Serious question -- why would Uber do that? What possible gain is in turning it off?

My guess: too many false positives, meaning it kept stopping dead for no (apparent) good reason.

The usual reason for people to try to bypass safety systems is that they get in the way.

Interesting finding about the location and amount of distraction and ability/time to react to a distraction. Considering that in NA highways and streets are cluttered with signage specifically designed to attract attention makes me wonder whether car insurance companies have looked at street signage vis-a-vis accident frequency and severity.

How this relates to self-driving vehicles: assuming the occupant presets the destination as well as a preference for the fastest route at the start of the ride, there's a lower risk of that self-drive car to swerve or make a quick turn because the rider saw and reacted to an ad. This is because it would take too much time to communicate the new destination plus make that turn so suddenly. If destination focused/incessant smartphone using riders consistently outnumber the react-to-every-outdoor-ad riders, then the utility/ad revenue for current oversized glaring outdoor signage drops. (Bonus: declutter city and highway land and skyscapes.)

WRT to lawsuits, owners/operators of distracting signage could be found partially liable for endangering the public (drivers, cyclists and pedestrians).

https://www.nature.com/articles/s41598-018-23498-w

Asymmetries in Distractibility: Left Distractors Improve Reaction Time Performance (Nicole A. Thomas & Michael E. R. Nicholls)

'Abstract

Research using the irrelevant-distractor paradigm shows perceptual load influences distractibility, such that distractors are more likely to be processed and decrease reaction times during low perceptual load. In contrast, under high load, attentional resources are limited, and the likelihood of distractibility is decreased. We manipulated distractor placement to determine whether location differentially influenced distractibility. During low load, reaction times were increased equally for all distractor locations. Under high load, left distractors speeded reaction times significantly more than right distractors. We suggest two potential explanations: (1) the central focus of attention was sufficiently large to encapsulate both the distractor and the visual array during low perceptual load, leading to increased distraction—during high load, attention was split across the two visual stimuli, allowing the distractors and array to be processed independently; (2) superior executive control for stimuli in the left visual field allowed participants to ‘catch and release’ left distractors more efficiently, ultimately decreasing distraction and providing a performance benefit. Our findings represent an intriguing development in relation to visual asymmetries in distractibility.'

Re: Overhead monorail

Serious question: this past winter has seen several massive storms along the US eastern coast. (In fact, some seemed to make it all the way up and across to the UK.) If CC/GW means that our weather patterns will include ever stronger winds and heavier downpours (or larger/deeper snowfalls), what then is the safest, most reliable* public transit?

Europe also experienced quite a few severe storms. How well did their public transit systems hold up?

• Several major cities cancelled subway service because of flooding. Apart from inconveniencing commuters, lot$ of structural damage.

Overground trams/light rail hold up well even during blizzards -- as long as the vehicles run they keep the tracks clear. During the Beast from the East Edinburgh was running one tram each way along the route, through the night, with a crew with shovels to keep the platforms clear. (They normally shut down for about six hours overnight.) This won't cope with flooding any better than roads, and hurricane-force winds can bring down overhead power cables, but they're more robust on snow/ice than road vehicles like buses and less prone to flooding than tunnel systems.

(I'll note that trams aka streetcars were very popular in the eastern bloc, where private cars were expensive/rare/long waiting lists and the weather got brutally cold every winter. There's probably a reason for this.)

Monorails just as Gerry Anderson imagined them.

"Trouble with that Skytran is that it doesn't ..."

The bigger trouble with SkyTran is that it's still vaporware as far as I can tell. They were supposedly building a proof of concept system at NASA Ames and a demonstration system in Tel Aviv. That was back in 2012; supposed to be in operation in 2014 - then in 2015 - then in 2016.

So where are they? I can't even find photos of unfinished construction.

Seems like NASA and a number of local governments sunk a lot of money into what looked like an interesting concept that just might be workable. But what have they got to show for that investment?

If it's a SCAM (and that's what it's starting to look like to me) it harms everyone else who has a possible innovative solution to the problem.

"Just a totally unsurprising infobit from the Panopticon that I missed when it first came out:"

https://consumerist.com/2014/05/19/private-companies-building-giant-unregulated-databases-tracking-license-plate-location-info/

It's concerning that unregulated corporations are acumulating so much intrusive data about our lives, but I found the first sentence amusing, since I DO know everywhere (when, where & the mileage) my cars have been for the last 38 years or so.

I had to start keeping a log when I was a field service tech & drove a company vehicle and I continued to do so for my own private vehicles when I moved on to a different job.

"Interesting finding about the location and amount of distraction and ability/time to react to a distraction. Considering that in NA highways and streets are cluttered with signage specifically designed to attract attention makes me wonder whether car insurance companies have looked at street signage vis-a-vis accident frequency and severity."

They have and it led to improvement on the signage put up by government bodies to direct & regulate traffic. Had less effect on commercial and advertising signage because that's First Amendment Protected Speech.

"Serious question: this past winter has seen several massive storms along the US eastern coast. (In fact, some seemed to make it all the way up and across to the UK.) If CC/GW means that our weather patterns will include ever stronger winds and heavier downpours (or larger/deeper snowfalls), what then is the safest, most reliable* public transit?"

Not so serious answer: If sea levels rise as much as some are predicting, probably Venetian Gondolas.

Not completely related but interesting (from SciAm).

'The European Space Agency’s (ESA) Space Debris Office in Darmstadt, Germany, which issued the Tiangong-1 prediction, said the March 30 to April 2 window is “highly variable,” and it will not be possible to determine exactly where the space station will fall to Earth. However, the space station will re-enter somewhere between the latitudes of 43 degrees north and 43 degrees south, based on its current orbital inclination.'

...

'China further added that it would provide a “timely forecast of its [Tiangong-1’s] re-entry” and would disseminate information through the news media and the United Nations. But because Tiangong-1 stopped working in March 2016, Chinese engineers will not be able to control its re-entry.'

So, about half of the planet. (This site has the best click-on visuals.)

http://ktvl.com/news/local/oregon-in-path-of-chinese-space-station-on-collision-course-with-earth

Given that the ESA, NASA and the Chinese space agency have good computing power, and ESA and NASA staff have decades of experience tracking stuff in not overly crowded orbit, seems amazing that they're unable to more precisely predict where the Chinese space station will make landfall. So - if these talented and dedicated folk can't predict what looks fairly straightforward, my confidence in newish, less well tested devices and algos (self-driving cars) being able to make much faster correct predictions in environments that can be many orders of magnitude more chaotic has taken a major hit.

Explanation of the actual joke:

Uber didn't release any of their test data (video / sensor etc) & the video online is from a 3rd party dashcam.

It is low quality and 'interestingly' gives a bit of an optical illusion of just how much the car could 'see'. Aka, using anthropomorphism (what you see in the video = what car sees.... wut?) as a novel PR defense. Or, in SV terms, a bit of a 'dark pattern' move.

There's also a load of panic in the air from the entire industry over Arizona banning Uber (California had previously barred them, so they were shopping for the lowest-common-denominator State to allow them 'free reign testing') with some really tasteless PR moves being pulled out of the hat. ("Our product would have not killed the homeless woman" etc).

Anyhow, couldn't remember if that was public (it is now) when I made the joke.

The meta-meta joke is about just how high quality a lot of .ru dashcam footage is. There's an entire subgenre of memes about .ru dashcams. Compare / contrast with Uber's offering and things might seem... a little darker.

p.s.

Don't look too closely @ the time/date on the Toby stuff either. But, really: ancient wetware being fed by data manipulation. wonders when the message gets out this often has unforeseen consequences.

It’s not straightforward mainly because the edge of atmosphere is thin, and therefore difficult to spot, and moves up and down a lot over periods of days. and because the spacecraft has lost propulsion. Once the spacecraft hits slightly thicker atmosphere the ballistic trajectory decays very rapidly and in a well understood way until the craft breaks up, but the uncertainty in first hitting that slightly higher, slightly less attenuated patch of gas is the root cause of large uncertainty in the landing time and place. There are no equivalents of this in road traffic because a) the default for a vehicle that loses propulsion is to slow to a stop due to drag and rolling resistance b) if you have propulsion you can brake and accelerate and generally get out of trouble, c) there is no requirement to forecast the position of any road vehicle several weeks in advance, only a handful of seconds in advance, which is a far easier problem. (Come to think of it, the error bar of a road vehicle's position two weeks in the future must be at least continental in size.)

tldr: road vehicles are always in a dissipative environment as far as kinetic energy goes, while the spacecraft is in the process of transition from generally non-dissipative to dissipative, and these two situations are not comparable.

Instead, you should be frightened because cars and trucks already kill thousands a week with virtual impunity and the push for automation is quite transparently being cost driven, not safety driven.

Just ran across an article in Slate (on-line) Magazine that says the average attention span of a safety driver in an autonomous vehicle is somewhere around 21 minutes. After that time span, the driver's attention wanders & performance declines.

https://slate.com/technology/2018/03/safety-drivers-attention-spans-might-slow-self-driving-car-progress.html

There's a rather large macabre cross-over here:

Two journalists in Bihar were run over on the highway in Bihar's Bhojpur district. According to locals, journalist Naveen Nischal and his associate Vijay Kumar paid with his life for fearless eporting.

In another incident, 35-year-old Sandeep Sharma was run over by a dumper in Madhya Pradesh's Bhind while he was investing the role of mafia involved in illegal sand mining.

3 journalists pay with life in 48 hours for fearless reporting India Today 27th March 2018

(Actual footage videos can be found in the wild very easily).

No comment on the 'whys' to the methodology, but it's an obvious one. India is having some particular troubles with Fake News, Journalists being targeted and so on.

Wonder why?

Whistleblower says Cambridge Analytica 'worked extensively' in India, names Congress as client Times of India, 27th Mar 2018

(The actual story is even more mind-blowing: double-crossing the client and so on. Big if True)

You are perfectly correct to be dubious. While the reasons can be regarded as different, they have a lot in common. Most experts in the relevant areas are also suspicious that the claims can be delivered (and you can add week-ahead weather forecasting).

I was actually thinking of that when I was typing my post :) - after all, the Skytran idea is also a dangling type. The maglev aspect of the Skytran is just the icing on the cake; even without that, simply to make it dangle involves a much more elaborate structure using much more material than supporting it from underneath.

(Simple home demonstration: your mission should you choose to accept it is to support a bag of sugar off the deck leaving 6 inches of clearance below the bag/structure, using bog roll tubes and gaffer tape. It will be a whole lot easier to build a platform to support it from underneath than to build a gantry to dangle it from.)

The Schwebebahn is something of a special case because running it down the middle of the river with the supports on the banks forces you to build a series of gantries in any case, no matter whether the vehicle passes above or below them. Even so it's not clear that it's actually better to dangle beneath than simply to have a conventional two-rail track running on top. It certainly doesn't avoid the points problem; it has to do things like this: http://www.vohwinkel.net/pics/ortsteile/schwebebahn/umbau-2007/20070805-haering-kaiserwagen/haer0799-kaiserwagen-im-depot.jpg http://static.panoramio.com/photos/large/11662845.jpg

Since you like this kind of thing (Cavemen-on-wall.gif)

Do a grep for a thread that contains the line "This isn't a Russian Op".

Then look @ UK News Press (red hair included) for Today, Year of Our Lord 2018.

p.s.

Not trying to get Host in trouble. "I am what I am"[1]

[1] Redwall. It has Mice in it.

Oh, and for everyone to save a referendum: It's anti-Semitism.

The why to this are largely to do with the Göttingen School of History who prefigured the 'Protocols of Zion' by, oh, 130 years or so.

Oh, and it was first defined by Jewish Scholars: Heymann Steinthal. (And, of course, then immediately 'claimed' by neo-reactionary racists, but hey! They made it Anti-Semitism showing off their amazing imaginations).

Could we all stop running ancient wetware now? That'd be great.

Athena & Bast on line #2. They're very pissed off.

support a bag of sugar off the deck leaving 6 inches of clearance below the bag/structure, using bog roll tubes and gaffer tape

Wrap a length of tape round the object and leave enough "tail" on the tape to attach it to whatever structure is above the surface in question. When I had to do this they failed to specify that it had to be above the table top so we used the table to hold the object above the floor.

I saw a number of people fail at this because it turned out that "distance above the surface" would be measured by passing a suitably sized block under the resulting construction but they had failed to pay attention to the part of the instructions.

Which is part of why I dislike those games. Although the funniest one was my using a magnet for "how many nails can you hold off the table using one similar nail" (challenge version of this) because they failed to say "using only nails". I was once again not rewarded for my ingenuity.

Triptych: If you've spotted that most of the scholars involved were a) German, b) all studied 'Orientalism' / Chinese languages and c) weren't too bothered about (what Europe calls) the Middle East... whelp.

There's a huge bag of irony / history replayed as farce joke coming right up about... now. (Not going to spoil it).

Polarity Reverses: Iceland & China, Aluminum Smelters & BitCoin now, let's do some Ecologically Friendly Reversal of Tropes:

ADB signs $250 million loan deal for China geothermal heating Reuters, Mar 22nd 2018

p.s.

David is 100% absolute National Treasure, but the amount of people in his feed not even doing basic research is depressing.

You're not all Mushrooms / Pod-People, language-games are historical.

Oh, and important Data Point:

In 1872, the first year, there were only 12 students, including four women

Hochschule für die Wissenschaft des Judentums

Compare:

The first woman to gain honours in a University examination which was intended to be equivalent to that taken by men for a degree was Annie Mary Anne Henley Rogers. In 1877 she gained first class hnours in Latin and Greek in the Second Examinatino for Honours in the recently instituted 'Examinations for Women'. In 1879 she followed this with first class honours in Ancient History. Annie Rogers matriculated and graduated on 26 October 1920.

First woman graduate of the University Oxford University Archives. And yes, the spelling mistakes are right there on the webpage.

Now: remind me how y'all forgot the lessons of the late 19th Century and got rail-roaded into retrogressive shit again?

Oh, right. $$$.

"This is the typical (Econ101) cobweb approach which assumes only one or two dimensions with no limits. Very easy to get such a chart to produce any result you want."

I'm reading that criticism of my comment as:

"Your common sense approach is simple, so I'll reject it on the grounds that it's overly simplistic without giving any evidence or argument that it's wrong".

Which I think is bollocks.

Are you arguing that making public transport free would not increase ridership, in a typical real-world case?

Or are you arguing that increased ridership would not increase the number of viable routes and frequency of busses, in a typical real-world case?

Or are you arguing the more routes and more frequent busses would not then increase ridership, in a typical real-world case?

Obviously these are limits to all of these - the busses I caught into Bern ran every 10 minutes, no-one knew the timetable because you just walked to a bus stop and caught the next bus. More frequent busses really wouldn't increase ridership by much. But that's very much not the case where I live now or most places I've been. But that just means that the feedback loop I give will resolve to a stable point, not that the feedback loop doesn't exist.

The big drawback with transit in the U.S. is the "You can't get there from here" factor. Routes are often too limited. You can get close, but you're still going to have to walk a long way to your destination. It's often further than people can walk. And that's assuming there are sidewalks so you don't have to walk out in the road.

Well, yeah.

Urban design dictates transport design. Sprawling low-density cities and suburbs are less well suited to public transport.

Zoning laws that prevent high-density housing make it worse. Just like zoning laws that require commercial and residential buildings to have lots of car parks are deliberate social engineering to avoid having car-owners pay the genuine price of the real estate cost of parking cars.

Please look into Oyster Cards in the UK (London).

Transport in cities is not optional.

Now think a little about what that means.

The 12 Jan 2017 one? (13 Jan 2017 was interesting for me.)

Yes. (Been trying to reverse-engineer that for a while. Fun.) Red-hair was in a couple of different picture captions in my us google news feed today. That hair captured some news cycles. My short list (of public people I have to try hard to not loath) includes the Mercers and Murdoch.

Unrelated except as a consequence, Rude Pundit (with some 2005 self-links) on John Bolton: Fucking Hell, John Bolton Again? (Still working on understanding that guy.)

Not sure what cavemen-on-wall pic is your favorite. Google images has quite a few.

The attempt to bring the 2024 Olympics to Boston went rather differently. The majority of Bostonians said “No thanks” pretty emphaticly.

"...zoning laws that require commercial and residential buildings to have lots of car parks are deliberate social engineering to avoid having car-owners pay the genuine price of the real estate cost of parking cars."

I disagree. I see them as straightforward practicality. If you don't have them, people have to try and squeeze their cars in anywhere they can in the nearby streets. This in turn leads to unending complaints, demands to the council to "do something" even though there isn't anything they can do, and vandalism to the cars, by the people who live in those streets; partly because some people do make genuinely bad choices of their parking spots, partly because (depending on the type of location and the daily routines of those involved) the people who live there can find themselves in turn forced to hunt for alternative parking, and mainly because people moan about other people's cars on "their" street out of primitive territoriality. And since if the space isn't used for car parks it will have something sellable built on it instead so they can make more money (and giving rise to even more cars needing to be parked), you have to have "laws" (planning regulations) requiring the provision of car parking space, otherwise you simply don't get any.

The "deliberate social engineering" aspect crops up when the planning requirements deliberately stipulate underprovision of car parking space, for the purpose - which in some cases is explicitly stated - of forcing people not to use or have cars. Of course, it doesn't work. Only someone who passes the quite stringent council tests for susceptibility to the delusion that reality will alter itself to conform with their utterances would even expect it to. Such people think that a cornucopia of excellent alternative methods of transport will magically appear out of thin air. Since this does not happen, people continue to use cars and just put up with the extra arseache of finding somewhere to park them.

"Urban design dictates transport design."

Well, it should do, but it doesn't. What we actually get is old places built to no design at all, more modern places built to a design which is no longer appropriate, and new places built to a "design" which copies the less practical features of the other two kinds and replicates them on a larger scale. It is just assumed that of course transport isn't a problem because there are roads so you can have buses.

Round my way you see this huge thing which is a bus struggling along the estate roads, billowing clouds of waste heat, constantly straining itself, start, stop, swerve, lurch, like a sperm whale trying to hump itself up an artificial white-water rafting course. You could probably run the route faster than the average speed of the bus. And the route is laid out to serve as many roads like that as possible, so a map of it looks like a handful of spaghetti thrown at the wall; the tortuous indirectness and gastropod average speed means it takes forever to actually get anywhere. It carries a handful of people, but it still has to run quite frequently to remain useful to the total of all the handfuls throughout the day. Efficient it is not, largely because it is not an appropriate solution in any sense other than being implementable without the council having to do anything novel.

An efficient solution certainly could be designed, but it would be nothing like a bus, it would not be easy, and it would require far more initiative and intelligence than is realistically available to (a) implement it at all and (b) implement it without some huge and stupid flaw that ruins its usefulness and also discredits the idea making it even harder for anyone in future to try and do it properly.

The "deliberate social engineering" aspect crops up when the planning requirements deliberately stipulate underprovision of car parking space, for the purpose - which in some cases is explicitly stated - of forcing people not to use or have cars.

This also includes making some parking free and others cost money. Where I live, they built a large subterranean parking space, but years later it's still under-utilized, for various reasons. The city has since planned to make some of the previously existing parking spaces to have fees - now they mostly have just a time limitation.

The lack of parking space, in general situations, is also very often affected by the area. The suburb where I live was mostly built in the 1950's and 1960's and while it's not nearly as old as even the Edinburgh New Town, it still has less spaces for cars now than perhaps should be. In our apartment building association (company? Residence arrangements are hard to transfer between countries) or apartment complex has over 170 apartments, but parking space for only less than a hundred cars. Granted, we have relatively good public transportation nearby, but still that feels like too little, at least when looking at the cars parked on the roads close to us.

Lastly, a question for the various native speakers of English: the afore-mentioned underground parking space got the name "Tapiola Park" - for me it sounds a bit strange, as in that should be a place here where there are trees and spaces for people to lounge about, not car parking. However, there is the expression "car park", at least in some parts of the English world. Does that name sound strange to you various English native speakers?

Yeah, well, the Parisians quite deliberately sabotaged their officialdom's attempts for 2012 - by having a transport strike on the day that Coe & his fellow-fascists were visiting. Vive Paris!

Actually, huge numbers of Londoners didn't want anything to do with that shit, but it was dumped all over us, whether we liked it or not ....

Considering that in NA highways and streets are cluttered with signage specifically designed to attract attention

Aka "adverts for $company"? In the UK a rest area with motel, fuel and multiple food vendors would have a single sign "Services" and icons for motel, fuel, food and toilets beneath the single word, rather than full-size billboards for half a dozen "fast food" franchises.

Para the last - It wouldn't be normal usage in Scotland; $name Park would certainly indicate an area, although it might well indicate a housing development rather than an area of recreational land.

I've said this before, it wasn't just Londoners who didn't want the "3rd Larndarn Olympics".

Harder to build - easier to maintain, and less likely to fail in use for unpredictable reasons. If gravity glitches, we have worse problems than a falling skytrain.

To use the word 'deliberate' in reference to ANY large-scale or widespread development policy in the UK is misusing the word. But I agree with icehawk that it's social engineering, as far as the UK is concerned, except that the reasons aren't what he said. In particular, the supermarkets aren't required to have such parks - they demand the right to have them, and get grovelled to, because that's what the law requires (and there are reasons for that, too).

“The big drawback with transit in the U.S. is the "You can't get there from here" factor. Routes are often too limited. You can get close, but you're still going to have to walk a long way to your destination.”

We do that here in the UK too, only (unusually for a UK/USA comparison) on a larger scale, nationally in fact, and it translates to “You can’t get there from here unless you go to London first”

Your car is learning from the environment

Actually, machine learning is currently quite limited in scope. The environments that cars need to navigate are too noisy and have too many dimensions for effective learning. This has even been noticed by the semi-lay press: https://www.theregister.co.uk/2018/02/27/huawei_mate_10_pro_ai_promo/

Instead, robotics is doing stuff like breaking motion planning into lots of embarrassingly parallel pieces, implemented in custom chips: https://pratt.duke.edu/about/news/robotic-motion-planning-real-time

Agreed, but it really was dumped all over us, we are still paying for it & the security theatre was deeply scary. If anything the "Everyone loves Big Brother" aspect of no dissentiung voices even being allowed a whisper was the worst part. And still unacknowledged.

to allow them 'free reign testing'

This isn't a dig solely at you, LML (Corran in #99 does exactly the same thing), but it's not reign ("to hold or exercise sovereign power," late 13c., from Old French regner "rule, reign" (12c.), from Latin regnare "have royal power, be king, rule, reign," from regnum "kingship, dominion, rule, realm," related to regere "to rule, to direct, keep straight, guide"), but rein (c. 1300, "strap fastened to a bridle," from Old French rene, resne "reins, bridle strap, laces" (Modern French rêne), probably from Vulgar Latin *retina "a bond, check," back-formation from Latin retinere "hold back" (see retain). To give something free rein is originally of horses.)

See also the common solecism "baited breath" (it's actually bated, from abated: c. 1300, "to alleviate, allay;" mid-14c., "suppress, do away with;" late 14c., "to reduce; to cease," a shortening of abate (q.v.). Now only in phrase bated breath (subdued or shortened breathing, from fear, passion, awe, etc.), which was used by Shakespeare in "The Merchant of Venice" (1596)..

I'm not a grammar nazi; it's just that mistakes of this type distract attention from your message.

I deliberately avoided London during that period because, if there was a security scare, I was far too much at risk of being shot by the security forces. I try to avoid the London airports for the same reason. At least, nowadays, the police gunmen are provided with some firearms safety training - I remember when they weren't :-(

I await a diatribe explaining that the English language is in the wrong.

The "deliberate social engineering" aspect crops up when the planning requirements deliberately stipulate underprovision of car parking space, for the purpose - which in some cases is explicitly stated - of forcing people not to use or have cars.

This is a glass half-empty/half-full issue: what Icehawk is pointing out is that you can use social engineering to force people, who would not otherwise need a car, to buy and operate a vehicle merely so that they can exist in a city: you do this by making it easy to operate a car (car parks near every place of work, etc) and thereby encouraging sprawl (big gaps between buildings, for parking lots), which discourages pedestrians and cyclists (longer walks, greater risk of being run over).

An efficient solution certainly could be designed, but it would be nothing like a bus,

Indeed not! It would be to demolish the housing estate in question -- if you're in the UK it's probably all small cookie-cutter boxes with tiny gardens and lots of narrow crescent-shaped roads leading to cul-de-sacs -- and build something more along the German apartment model: blocks of spacious apartments, no more than 4-5 stories high (with elevators for the infirm or people who need to move furniture up and down), landscaped grounds, allotments for those who want a vegetable garden of their own, maybe a basement-level car park for those who need a vehicle, and a bus stop right outside the common entrance with a frequent scheduled service.

But hey, that only requires you to go against about 90 years of UK town planning policy, going back to the post-WW1 vision of suburban bliss and car ownership as something to aspire to.

See also Austria, Eastern Europe (with taller buildings), and Scandinavia.

I try to avoid the London airports for the same reason.

I don't fly via Gatwick or Heathrow -- ever.

They're utterly shit if you need to give yourself enough leeway to check your bags before your flight: nothing but overpriced junk food eateries and shops, nowhere to sit without paying, annoyingly intrusive security.

When I need to fly via a hub they're the same travel time from where I live as Paris and Amsterdam, both of which are rather nicer. (It's 400-odd miles from EDI to LHR or LGW vs. 500-odd miles from EDI to CDG or AMS, but that's effectively the same, once you're airborne on a 737.)

Secondary annoyance: when an event organizer insists on booking my travel, looks at a map, and assumes I'll happily travel to Glasgow or Newcastle airports to catch a direct flight, to avoid having to fly an extra sector from Edinburgh. (Maps fail to show that the ground travel time to those airports -- 50 and 100 miles away -- are over two hours[*], and it's commuter hell if I drive or multiple transit connections if I try to go by public transport.)

[*] Yes, on a map Glasgow airport looks close to Edinburgh. Trouble is, to get to it by road I have to travel through the centres of both Edinburgh and Glasgow, both of which are congested; and there's no direct rail/bus link that bypasses Glasgow city centre.

I don't fly via Gatwick or Heathrow -- ever. They're utterly shit if...

Simple answer - become a class traitor, fly business class, it might change your opinion of Heathrow ;)

Spend your transit time in a comfortable lounge with free food and newspapers, and your flight time in a seat that turns into a bed. Start the flight with a glass of champagne, and be trusted with metal cutlery...

...sad to say, my former employers always expected me to spend hours in cattle class shoehorned into row 55 between the chunky brothers, all the way to San Francisco - whereas my wife's firm sends her to Asia on business class "because they want her to be in a fit state to work when she arrives". Envy.

In your position, I wouldn't, either. But BAA deliberately kept Stansted from competing with Thiefrow and Grotwick, and getting to Manchester airport (or up to you) is as difficult for me as getting down here is for you. I can get to Stansted in an hour (less by road), but it's over two hours to the other two.

I was far too much at risk of being shot by the security forces.

:) This sounds a fun explanation :)

If you saw the amount and quality of training that they do before being let out on the streets with bangy things, you might change your mind. They really haven't shot that many innocent people recently (genuinely innocent, not "of course he was innocent, he'd chucked the pistol out of the taxi").

Now, if you were Brazilian-looking and getting onto a Tube train; carrying a chairleg; or driving a Mini through London...

Yes. There are a lot like that. And it makes it pointless for me to say that the thing I am afraid of IS the self-driving car companies being given a free reign over UK roads :-(

We took the kids to London to see the Games - it was impressively well run (the volunteers were helpful and enthusiastic - how unBritish!), and we really enjoyed it. Sorry to lower the tone...

I was more skeptical about how Glasgow would host the Commonwealth Games two years afterwards, but even that was impressive. Friendly 'Wegians, good atmosphere, and the only time I've ever (or will ever) set foot in Ibrox Stadium (the Police were suffering cognitive dissonance, the crowds heading in and out weren't drunk, aggressive, or kicking off). Amusing seeing a full crowd in Ibrox, deciding that they were going to go wild for the Ugandan Rugby Sevens team...

It would probably make me decide never to go anywhere those dangers to the public are likely to be.

Since you ask, I am severely deaf, and my reflex (and it IS a reflex) to hearing a shout behind me is to whip around fast, in case it is a warning I am about to be run down by a vehicle. There is NO sodding chance of me either hearing words spoken from behind (as distinct from detecting a shout), nor of decoding them within my reaction times, unless I am already facing the speaker and concentrating on listening. That's JUST what I should do if I police gunman shouts "DON'T MOVE", isn't it?

And, to correct your misrepresentation, one actual witness said that he did NOT throw anything, the expert witnesses said that it was almost impossible for him to have done so and for the gun to get where it did, and that the policeman had to have fired within 2 milliseconds (where a minimum possible reaction time is 20 milliseconds).

You should probably avoid the US, then.

Simple answer - become a class traitor, fly business class, it might change your opinion of Heathrow ;)

Already did that, get the equivalent treatment in Paris and Amsterdam :)

Ref "secondary annoyance"

I can't do anything useful about Auld Reekie (I don't think anyone can without the use of strategic nuclear weapons ;-) ) but I think you're presuming that the "best route" to Glasgow Airport is to stay on the M8 the whole way?

Since the M74 was finished, I've been using M8 - M74 to get from Dumbarton to England (for SF cons) and it's taken an hour off the trip even hitting the Motherwell-Hamilton area ~5PM local Northbound. So I think I can save you half an hour if you route M8-M73-M74-M8 just South of the Kamikaze instead of schlepping through Charing Cross on the M8 (proof that 4 into 2 doesn't go).

Yes. Though the soldiers at USA airports scare me a damn sight less than the police gunmen at Heathrow and Gatwick - at least I could take my chances with random bullets rather than being specifically targetted. Of course, in the USA, it would help that I am white, beardless and 70 years old.

Thanks for that about your Para 2.

Since the M74 was finished, I've been using M8 - M74 to get from Dumbarton to England (for SF cons) and it's taken an hour off the trip even hitting the Motherwell-Hamilton area ~5PM local Northbound.

I can second that. I used to live in Dumbarton, and go back frequently to see family, driving from near Gatwick (because with a family of four it's cheaper and easier than flying).