back to article Dumb autonomous cars can save more lives than brilliant ones

Autonomous cars only need to be good enough to reduce the number of road deaths to be worth permitting: eliminating fatal accidents can wait until later. That's the result of an analysis from military-industrial complex darling RAND Corporation. That idea isn't entirely new, since it was put forward by America's National …

  1. Notas Badoff

    Better than the next guy

    It's been just a week since a driver ran a stop sign and boom. A second earlier and our car would have done serious damage to his passenger. A second later and his car would have killed me, likely outright.

    There are any number of checks that would have stopped this. Like there was a stop sign. The street he was trying to cross (the witness said "blow thru") is so major just reference to a map would've said nuh-unh. The speed he was going on that side street was at *least* double posted or reasonable. The angle that street had to the intersection prevented seeing a whole direction of traffic. And at that time of day the odds that any 4-second period wouldn't have a car crossing is pretty low.

    Also, computers don't have mid-life crises causing 50-year-olds to drive like 15-year-olds.

    Count me as positive that automatic driving has got to be better than my fellow drivers. Even now. If I could ask his passenger I think that'd be another 'yea' vote.

    1. NickyD

      Re: Better than the next guy

      Upvote for 50 is mid-life :)

    2. jdoe.700101

      Re: Better than the next guy

      Autonomous cars need to be better than the next guy ALWAYS. If they decide to return control to the human, or stop, when they get confused, they're a liability. The problem with driving is that it is full of edge cases (road works, snow, flooding, dropped loads, protesters, cows, etc...) which humans are remarkably good at dealing with. Attention and reaction time however is where computers win, let them deal with that, and leave the driving to the human, until they can deal with the edge cases.

      1. Francis Boyle Silver badge

        Re: Better than the next guy

        You're conflating two very different cases here. If a car is designed to safely stop and refuse to proceed autonomously because it detects a situation it can't cope with, then at most it's an inconvenience. If you are talking about cases where the car finds itself in a situation where it is unable to extricate itself safely, it only needs to be better than the average driver. The premise of your final sentence is spot on. It's just that the conclusion doesn't follow.

        1. Anonymous Coward
          Anonymous Coward

          Re: Better than the next guy

          Ignoring the dubious credentials of Rand, I think this does highlight a theme, a halfway house that is being ignored in the wider debate about AVs, and that is simply limiting the vehicle to obey the laws and road conventions, but not taking responsibility for direction, speed or braking beyond that (I'd include collision mitigation/avoidance braking). I don't think that's what Rand are getting at, but I think this is a better way forward, whether as a permanent change, or as a stepping stone to full automation. I suspect that making all vehicles stick to the rules would get that 10% reduction in casualties.

          In the case of the OP, it would mean that the other vehicle couldn't have exceeded a posted speed limit, it would have had to have stopped if the signs or traffic signals indicated that. Ultimately it might not have stopped the car pulling out in front of oncoming traffic, but that sort of fully automated driving is (I would guess) an order of magnitude or two greater complexity, than simply "rules of the road" automation.

          Now, as an enthusiastic driver who doesn't always obey the rules, I'm in no hurry for this on my own account, but looking more widely, this would be an improvement where driving standards are low (like the US). And maybe I over-estimate my own abilities and would be an ideal candidate to have a car that could only obey the rules.

        2. jdoe.700101

          Re: Better than the next guy

          You're conflating two very different cases here. If a car is designed to safely stop and refuse to proceed autonomously because it detects a situation it can't cope with, then at most it's an inconvenience.

          Assuming that the car can safely stop, that other cars nearby can also handle the car stopping, and that the car has manual controls, then yes it could be argued that its an inconvenience. However, if an autonomous car does the bulk of the driving, and leaves the unusual driving to the human, you could argue that they are making driving more dangerous. Because in such situations, the human will have less driving experience than previously.

          1. Baldrickk

            Re: Better than the next guy

            It's less about the experience - I know new drivers who are safer than peope who have been behind the wheel for years.

            It's the amount of attention that they are giving to the driving of the car, given that they are no longer the ones driving, until suddenly, they are.

            I did of course get what you meant though,

          2. Francis Boyle Silver badge

            Re: Better than the next guy

            "However, if an autonomous car does the bulk of the driving, and leaves the unusual driving to the human, you could argue that they are making driving more dangerous. Because in such situations, the human will have less driving experience than previously."

            You're ignoring the fact that most of the cases you cited have little to do with driving skills per se. (They simply involve driving slowly and using common sense. Or telling the car what to do.)

            Driving in snow is the exception. But then, while it may be something autonomous vehicles have difficulty at present, unless I'm missing something critical, it's the sort of thing that computers can be very good at given the training/programming (follow the road, keep control of the vehicle and don't hit anything).

      2. teknopaul

        Re: Better than the next guy

        Did you read the article? The point was that they dont need to ALWAYS be better, they just need better safety over all.

        1. teknopaul

          Re: Better than the next guy

          Thats probably quite easy to achieve. Drive at the speed limit and don't jump reds.

          If the driver has to manually override to jump a red there would be no getting away with it in court if they hurt someone.

  2. DonL

    Estimates

    The problem is that "10 per cent better than humans" is an (probably optimistic) estimate and it may very well end up being worse than humans. If they think it's 90 percent better then it's probably still optimistic but at least going to save lives.

    I'm all for adding safety features (i.e. automatic breaking) to cars, but fully automatic driving still seems like a stretch at this point.

    1. jmch Silver badge

      Well, duh!

      "allowing autonomous vehicles on the road when they're merely 10 per cent better than humans, versus waiting until “their safety performance is 75 or 90 per cent better than that of average human drivers”."

      and " “under none of the conditions we explored does waiting for significant safety gains result in fewer fatalities.”"

      So both the scenarios tested involved cars who drove better than humans and in no case would more fatalities be expected. Of course not! Proper test would require also a solution where autonomous vehicles are 10% less capable than humans!

      And in any case how do you even gauge what is the performance of the "average human driver"??

      1. Dr. Mouse

        Re: Well, duh!

        "Proper test would require also a solution where autonomous vehicles are 10% less capable than humans!"

        But they will not be allowed to be released if they are worse than humans. They will only be out there when they are demonstrably better drivers than humans (which, I suspect, they already are). So you are modelling an unrealistic scenario in that case.

      2. Robert Moore
        Angel

        Re: Well, duh!

        With apologies to the late great George Carlin.

        Think of how bad the average driver is, and realise half of them are worse than that.

    2. Chemical Bob
      Trollface

      Re: automatic breaking

      The auto industry has been trying to get this working for years as a way to boost sales.

      1. Robert Moore

        Re: automatic breaking

        When I was a kid, (Several million years ago) my father bought an AMC Hornet Station wagon, bright yellow with fake wood grain panels.

        This car had a very robust and reliable implementation of automatic breaking. As I recall the car spent more time in the shop than on the road.

        I always felt the yellow colour was quite fitting for a complete lemon.

  3. Anonymous Coward
    Anonymous Coward

    As I've said many times

    It needs a 90% reduction in injuries and deaths over the toll for human drivers before it will be generally accepted. 1) Because most people believe they are a better than average driver. 2) More importantly, those numbers are skewed by drunk drivers and other people not paying attention.

    Sure, if I was trying to get home after having a dozen beers and a few shots I'd use a car that's only 10% better than human drivers rather than drive myself (though truth be told, because I'd be worried about ending up in jail) But for my daily drive where I'm sober and paying attention, FUCK NO - I know for a fact I'll be better off than the statistical average which includes all the drunks and millennials who have withdrawals if they go five minutes without sending a text.

    A car that's only 10% better will only be suitable for hire on New Year's Eve, St. Patrick's Day, and during a 10 minute window after closing time for the local establishments.

    1. Anonymous Coward
      Anonymous Coward

      Re: As I've said many times

      And that's exactly why the machine can do better. Because you CANNOT know for a fact. For the basic reason that you're a meatsack that, like it or not, reacts to emotions, even on a subconscious (unknowing) level. And unlike you, a machine can't be placed under duress for any number of reasons. If you're on your way to work, you may be tired or frustrated. On your way home? Stressed from the day or in a hurry. It can be patient in traffic jams, less reactive to sunrises and sunsets, and so on.

    2. Trilkhai

      Re: As I've said many times

      "More importantly, those numbers are skewed by drunk drivers and other people not paying attention."

      My thought exactly. There's a good reason why insurance companies charge different amounts to drivers based on our age and driving record — there's a big difference between "all drivers on average" and "middle-aged drivers with 12+ years experience and a spotless record."

      1. Dr Dan Holdsworth
        Facepalm

        Re: As I've said many times

        The researchers here are not implying that semi-autonomous cars (which a lot of us are already driving, myself included) are 100% safe all the time, but instead that adding in protection against human drivers doing something bloody stupid does tend to reduce accidents and save lives.

        If you take a broad view of this, then ABS braking is a primitive form of semi-autonomous car system; it removes the task of optimising the brakes from the driver. Emergency Stop assist systems do the same; drivers often don't jam the brakes on hard enough in an emergency. Frontal radar reduces the number of accidents where through inattention the driver hits a car in front.

        Motorway Lane assist systems are another form of semi-autonomous car control system; they reduce the number of "dozy driver wandering about motorway" sort of accidents. This is the sort of thing the researchers were arguing for; not fully autonomous vehicles but systems which reduce the number of just plain stupid things a motorist can do through inattention or misapprehension.

        1. Charles 9

          Re: As I've said many times

          "Motorway Lane assist systems are another form of semi-autonomous car control system; they reduce the number of "dozy driver wandering about motorway" sort of accidents. This is the sort of thing the researchers were arguing for; not fully autonomous vehicles but systems which reduce the number of just plain stupid things a motorist can do through inattention or misapprehension."

          But some would argue that this may just result in stupider drivers that in turn create edge cases that the semi-autonomous systems aren't capable of coping. Plus there's always Trolley Problems.

  4. John Smith 19 Gold badge

    Now is that "10 per cent better than humans" or "10 per cent better than American humans?"

    Because American drivers do appear to be epicly s**t.

    "It's been just a week since a driver ran a stop sign and boom.

    ---

    Like there was a stop sign. The street he was trying to cross (the witness said "blow thru") is so major just reference to a map would've said nuh-unh. The speed he was going on that side street was at *least* double posted or reasonable. The angle that street had to the intersection prevented seeing a whole direction of traffic. "

    I guarantee that if that Ahole remembers it happening they think it shows what a superb driver they are.

    IMHO the trouble with American road deaths is not the number. It's the number of incompetent motherf**kers who kill others but don't die themselves, and then get behind the wheel and do it some more.

    BTW the US has roughly 5x the population of the UK and 10x the road body count, despite having no roundabouts and significantly lower average population density.

    1. Charles 9

      Re: Now is that "10 per cent better than humans" or "10 per cent better than American humans?"

      "BTW the US has roughly 5x the population of the UK and 10x the road body count, despite having no roundabouts and significantly lower average population density."

      But it has a higher NUMBER of concentrated areas, and that's where the accidents pile up. There are also more total road mileage in the US, particularly in freeways/motorways where speeds are higher and flow is expected to be smoother. You think London is bad? Try Los Angeles.

      1. Pen-y-gors

        Re: Now is that "10 per cent better than humans" or "10 per cent better than American humans?"

        @Charles 9

        "But it has a higher NUMBER of concentrated areas, and that's where the accidents pile up"

        But in concentrated/built-up areas vehicles are usually travelling fairly slowly, so there will be accidents but should be relatively few fatalities (except for those evil unAmerican lefties who dare to walk)

        1. Charles 9

          Re: Now is that "10 per cent better than humans" or "10 per cent better than American humans?"

          "But in concentrated/built-up areas vehicles are usually travelling fairly slowly, so there will be accidents but should be relatively few fatalities (except for those evil unAmerican lefties who dare to walk)"

          You've never been to New York City or Los Angeles, have you. Built-up AND high-speed travel aren't necessarily mutually exclusive.

        2. Anonymous Coward
          Anonymous Coward

          Re: Now is that "10 per cent better than humans" or "10 per cent better than American humans?"

          "But in concentrated/built-up areas vehicles are usually travelling fairly slowly"

          There's this little thing called a freeway or expressway that typically moves at 90 to 140 kph, depending on different factors... with as many as ten lanes of traffic going each way. These are found in, and most heavily used in, built up areas.

          For some people the daily commute is on the order of 200 km each way, though most are no more than 120 km, and many are less than 80 km each way.

      2. Martin Gregorie

        Re: Now is that "10 per cent better than humans" or "10 per cent better than American humans?"

        You think London is bad? Try Los Angeles.

        A lot of those traffic jams is directly due to Good Ole GM. Back in the '30s LA was building a Metro system - until GM bought it up and demolished it.

        However, that doesn't affect the fact that the average USAian is a godawful driver by comparison with almost anywhere else in the world.

        1. Charles 9

          Re: Now is that "10 per cent better than humans" or "10 per cent better than American humans?"

          They tried again in the 90's...until they accidentally created a sinkhole in downtown LA. These days, sprawl is the main reason it's not considered viable since LA built OUT while New York (out of necessity) built UP.

    2. Trilkhai

      Re: Now is that "10 per cent better than humans" or "10 per cent better than American humans?"

      Doesn't the US also have a much greater dependence on cars in general than the UK? I'm just thinking: if Americans have to spend a lot more time behind the wheel on average than the British do, then it could be that the fatality rate per hours driven is equivalent, even though the per-capita rate is higher.

      I can't say for others' experiences elsewhere, but the handful of simple 1-lane "traffic circle" roundabouts we have in my town are both faster & safer than the traditional intersections are. (If we had the multi-lane type that's common across the pond, I might feel differently.)

      1. Anonymous Coward
        Anonymous Coward

        Re: Now is that "10 per cent better than humans" or "10 per cent better than American humans?"

        if Americans have to spend a lot more time behind the wheel on average than the British do, then it could be that the fatality rate per hours driven is equivalent, even though the per-capita rate is higher.

        Per mile driven, the US still has far higher fatalities. However you cut the figures, the US has amongst the worst road safety records of developed nations (check out OECD IRTAD data). Poor driving standards are probably the main contributor, but I'm with you that the US aversion to roundabouts, and excessive use of flat junctions on high speed roads are another major contributor, and these poor road layouts are very unforgiving of poor driving standards. We should also bear in mind the federal nature of the US which introduces unnecessary inter-state variations that certainly won't help matters, albeit a dick in Minnesota is probably still a dick in California.

        What's particularly bad about the US casualties, is that they aren't making much progress in improving road safety - casualties per 100k km data shows the situation getting marginally worse between 2010-2015, whereas other poor performers (Czech Republic, NZ, Belgium, Korea) all achieved 10-20% improvements. The US is also particularly poor for road fatalities affecting the elderly, with death rates per capita 5x the best performing countries, although as proportion of fatalities the US sees fewer pedestrians killed than European nations, perhaps as you'd expect from a nation that drive everywhere.

        1. John Smith 19 Gold badge
          Coat

          casualties per 100k km data..getting marginally worse.particularly poo..elderly,death rates

          per capita 5x the best

          So even normalized to a standardised section of road (100K Km?) they still come out badly.

          As for death rates among codgers being 5x that of other G7 nations well...

          Think of it as "evolution in action."

          Gotta cut the welfare bill somehow.

      2. Anonymous Coward
        Anonymous Coward

        Re: Now is that "10 per cent better than humans" or "10 per cent better than American humans?"

        "Doesn't the US also have a much greater dependence on cars in general than the UK?"

        Yes, which is not surprising given that the UK is tiny (smaller than the Great Lakes, or Michigan, or any of ten other states), and most of the urban areas were laid out and built after people stopped traveling on horses or steam trains.

      3. Anonymous Coward
        Anonymous Coward

        Re: Now is that "10 per cent better than humans" or "10 per cent better than American humans?"

        "Doesn't the US also have a much greater dependence on cars in general than the UK?"

        Yes, which is not surprising given that the UK is tiny (smaller than the Great Lakes, or Michigan, or any of ten other states), and most of the urban areas were laid out and built after people stopped traveling on horses or steam trains.

        Also, the UK doesn't tend to extreme weather... for example heavy icing and blizzards are likely much less common in the UK, particularly in cities where there will be urban 'heat islands'. Climate data shows that average lows in the UK are above zero in all months.

    3. Ugotta B. Kiddingme

      @John Smith 19

      "...despite having no roundabouts..."

      Actually, these are beginning to appear - and with predictably amusing results. There are a few recently constructed in my area which have confused the populace and, temporarily at least, actually made traffic worse.

      I, on the other hand, was fortunate enough to know what they are and how to handle them. The medium sized city where I grew up in the 1960's had them at either end of the major north-south highway through town. The south "traffic circle" was the larger of the two and really more of a rounded square nearly a mile in circumference. Few ever had trouble with that because the size meant you had plenty of time to ponder which lane you needed for whatever action was required at the next inlet/outlet. The NORTH circle, no more than 75m in diameter, was closer to a classic British roundabout. This, due to both the smallish size and the fact that it was double-laned, led to no end of amusing confusion for non-local drivers. It was not at all abnormal to see a non-local stuck on the inside lane go round three or four times trying to find a gap to get into the outer lane and exit the circle OR miss his intended exit and have to go round another time or two in order to get it right. The north traffic circle and a particularly narrow bridge (inevitably always facing an oncoming lorry) were more or less required parts of the final exams of local driving instructors who got lots of sadistic pleasure watching white-knuckled teenagers trying to navigate these obstacles. Those of us who learned how to drive there are better for it.

      1. Anonymous Coward
        Anonymous Coward

        Re: @John Smith 19

        This, due to both the smallish size and the fact that it was double-laned, led to no end of amusing confusion for non-local drivers.

        We've got two or three double lane roundabouts (ie two way traffic). But then we put a mini roundabout at each junction. Search on Wikipedia for "magic roundabout hemel", and imagine US drivers arriving at that. Funnily enough, it's always worked well when I've been on it.

        1. John Smith 19 Gold badge
          Unhappy

          "and imagine US drivers arriving at that. "

          Some years ago the USG had a web site page that revealed the #1 killer of USians abroad.

          Not in fact terrorists. Traffic accidents. Especially the roundabout.

          Can't recall if it was the Transport or Commerce departments that had the page.

        2. Kevin Johnston

          Re: @John Smith 19

          Hey

          We Moonrakers in Swindon were there first...and ours has 5 mini-roundabouts

  5. Kevin Johnston

    Must read the sub-head more carefully

    I was sure they were suggesting that the self-drive cars should be tanks. This was reinforced when I saw they were a military-industrialist mob.

    It would keep the people in the self-drive tanks safer...at the expense of everyone else though

  6. cookieMonster Silver badge

    replace driver airbag

    With 4' metal spike..

    Watch the accident rate drop dramatically.

    1. Pen-y-gors

      Re: replace driver airbag

      a 4' metal spike? How do you sit in the driver's seat?

      Cue small 18" high trilithon being lowered into sight...

      1. 's water music

        Re: replace driver airbag

        a 4' metal spike? How do you sit in the driver's seat?

        Might need a run up. I doubt you would be subsequently involved in a RTC though

    2. Charles 9

      Re: replace driver airbag

      "Watch the accident rate drop dramatically."

      Oh? Explain the collateral damage caused by ghost drivers and DUIs, both of which are categorically incapable of risk assessment (the former has a death with, the latter is too sloshed to give a damn).

  7. 0laf

    I think this is where Sci/Eng meets the the Daily Mail.

    It's perfectly logical and rational to introduce a technology that will reduce death figures overall even if that technology may cause some deaths. i.e. Introducing autonomous vehicles may save 1000 lives whilst costing 10.

    990 more people is walking around is better, yes? Shame about those 10 but 990 is bigger than 10 and the needs of the many...yada....

    Except when YOUR autonomous vehicle crashes into a bus stop killing 5 kids whilst avoiding the garbage dropped in the middle of the road that looks vaguely like a person to the car you'll see Daily Mail headlines of "Killer technology murders family, where will it end, the machines will murder us all".

    Then the politicians will panic, pass knee jerk legislation and basically the autonomous vehicle 'thing; will get hobbled for 20yr.

    Got to think of the bigger picture and the number don't always work outside the lab in the irrational meatware that will make decisions.

    1. Anonymous Coward
      Anonymous Coward

      "Got to think of the bigger picture and the number don't always work outside the lab in the irrational meatware that will make decisions."

      Yes, this exactly the issue. Bandying numbers and percentages around is all very well in a "think tank" setting (pardon the elevation of the RAND corp) but let's take an IT example. When a new IT project goes live, it is always used in ways the designers didn't fully envisage. In the case of autonomous vehicles, this will also be true. Even given that, it must be possible to envisage some new form of danger from autonomous vehicles, and the question is whether society will accept such a new form of danger. We tolerate car deaths because of a greater good, we believe, and the human agency associated with driving. I am not so sure we should tolerate death-by-automaton as easily.

    2. Anonymous Coward
      Anonymous Coward

      Spot on. Let us not forget that the Daily Mail recently tried to get bouncy castles banned.

  8. Nick Kew

    Backlash

    Trouble is, if autonomous cars are only, say, 99% safer than human drivers, what happens when a case is reported where one is responsible for killing someone?

    There are lots of precedents for damaging backlashes. In the field of transport, just look at the completely different standards applied to rail deaths compared to road deaths. If the cost-per-death of the Hatfield disaster were applied to motorists, a year's insurance premium would cost more than a new car, and it seems statistically likely that the rail disruption that followed it caused more deaths (by driving people onto the roads) than the crash itself.

  9. steelpillow Silver badge
    Meh

    likelihood vs. magnitude

    How do you quantify security? What is the probability of some malware bot, at some point in the next thirty years and for whatever reason, infecting some leading model and causing a global mass pileup at speed? You can't do that to human drivers. How low does the risk have to be, before we judge autonomy as "better than"? How do you manage it down to that level?

    1. Zippy's Sausage Factory

      Re: likelihood vs. magnitude

      What is the probability of some malware bot, at some point in the next thirty years and for whatever reason, infecting some leading model and causing a global mass pileup at speed?

      Given recent events, I'd say we're looking at somewhere between 99.999% and 100%.

    2. Anonymous Coward
      Anonymous Coward

      Re: likelihood vs. magnitude

      "You can't do that to human drivers."

      Sure you can. Fog, sun glare, black ice, and you can have 100-car pileups easily. A gigantic flare set off over the Springfield Interchange (outside Washington, DC: infamously complicated and crowded at almost all times) at rush hour could cause similar carnage, and the technology already exists.

  10. imanidiot Silver badge

    But WHEN are they better

    I might well be that autonomous cars are 500% better (what does that even mean?) in normal driving conditions, accounting for most of the mileage, but absolute shit in unusual situations where it's much more likely to kill it's occupant than a human driver is. SO overall it's like 100% better right?? THe question becomes, would you accept a car that runs perfectly fine most of the time but which is likely to kill you next time a heavy snowstorm hits?

    1. Francis Boyle Silver badge

      Re: But WHEN are they better

      If it's 500% better then it's not going to be "likely to kill you" in any determinable situation. Yes, it will be more dangerous in some situations but, by definition, those situations will be rare, or the danger won't be great. So no different from the current situation where some conditions are clearly more dangerous for human drivers.

      Yes, people are terrible at evaluating risks and tend to overestimate rare risks with serious outcomes, but most of us are fine with taking greater risks in some circumstances (or no one would ever speed let alone drive while tired drunk etc).

      *Will a reduction in the number of fatalities per hours driven by a factor of 5 do?

  11. MikeCarew

    But who will make the decision?

    When the insurance companies decide, through actuary science, that it is cheaper for the car rather than the person to drive, what effect will that have?

    1. Anonymous Coward
      Anonymous Coward

      Re: But who will make the decision?

      "When the insurance companies decide, through actuary science, that it is cheaper for the car rather than the person to drive, what effect will that have?"

      The roads become safer.

  12. Davidcrockett

    Surely the best combination (currently) is a human driver with a robot back seat driver that takes control if you're going to bin the car. Humans are better at edge cases whilst robots have better all round vision and don't get tired, seems like a good combo.

    1. 0laf

      Sounds a lot like what we already have with traction control, ABS and anti-skid.

      Problem is before these systems kick in the human has got the car in a very bad place already.

      And what if the human has chosen to throw their car off a cliff to avoid hitting that crowd of toddlers crossing the road. Might be a bad idea then for the car to take over and keep itself on the road.

      1. Dr Dan Holdsworth
        Facepalm

        Yes, ABS, traction control, frontal radar, lane-keeping and emergency brake assist all step in when the human has gotten into a bad spot and guess what, these systems all work and all save lives.

        Semi-autonomous cars will not be designed by Daily Fail idiots, but by a combination of automotive engineers, actuaries and software designers, all of whom are ruled by intelligence, not by raw emotion. The changes will be slow and mostly not noticeable to the average driver, much as the huge leaps forward in car safety have not been noticed by the average driver, but they will be there.

        What will happen is that the robotic systems will creep in and take over the bits of driving that humans find hard, and leave the easy stuff to us. Quite frankly we should be amazed that a bunch of apes that cannot run anywhere near 20 MPH can safely control vehicles travelling at ten times that speed.

  13. c1ue

    Idiotic.

    If saving lives is the metric that matters - then ban motorcycles.

    1. Anonymous Coward
      Anonymous Coward

      "If saving lives is the metric that matters - then ban motorcycles."

      It is true that a recent study found motorcycles had five times the rate of accidental deaths and six times the rate of serious accidents, but we allow natural selection little enough scope in our society as it is. It could be a mistake to reduce it further.

  14. Zippy's Sausage Factory
    Facepalm

    It doesn't matter

    My theory is that the first time a crash involving an autonomous cars features on the evening TV news, the Daily Fail, Excess, Scar and all the other red tops will run "BAN THESE KILLER CARS" headlines. Circulation will spike on that day, so they'll think "aha, another cause".

    Within two years, the government will cave, and ban driverless cars and computerised driving aids from British roads.

    1. Charles 9

      Re: It doesn't matter

      Then why hasn't it happened already, given we already have had "close enough" instances?

      1. Zippy's Sausage Factory

        Re: It doesn't matter

        Then why hasn't it happened already, given we already have had "close enough" instances?

        Probably because the papers haven't decided it will sell enough extra papers to offset the loss of advertising revenue from annoyed car companies.

        1. Charles 9

          Re: It doesn't matter

          You would think newspapers would probably have Plan B's in place to replace car ads should things head that direction.

  15. N2

    Yesterday

    Some one waved at me, indicating to slow down, I dont drive too fast on the country roads around us as you never know whats around the next corner. So I slowed down anyway & around the next corner was a loose horse getting a bit agitated. Passing it may have caused a problem even though the road was clear, but waiting a couple of minutes until it calmed down did the trick.

    I'm not sure how many robot cars can recognise that.

    1. WolfFan Silver badge

      Re: Yesterday

      That would depend on the robot car. KITT would recognise the problem, and would help fix it. KARR would run over the horse, back up, and run it over again.

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

      I'm showing my age again. I have to stop doing that.

      1. Charles 9

        Re: Yesterday

        Don't. I actually got the reference, seeing as how I grew up watching the original Knight Rider and recently obtained the series for old time's sake.

      2. Zippy's Sausage Factory
        Coffee/keyboard

        Re: Yesterday

        KARR would run over the horse, back up, and run it over again.

        That genuinely made me laugh. Have an upvote.

  16. GordyBUK

    So we'll just ignore those cases where the driver saved the day?

    All this hype about self driving cars reducing the number of deaths on the roads ignores the data about the number of situations where a human driver has taken evasive action, and as a result, prevented an accident from happening, and people from being injured or killed. This is the sort of thing that defensive driving course teach you about.

    For some reason, this data isn't routinely collected, but if we want to make an informed decision we need it. In aviation, it is suggested that the number of incidents (near misses) is about 10 times higher than the number of accidents. I suspect that the number is higher on the roads, given the number of cars, traffic densities etc. I admit that some of the avaition incidents are due to things like interruptions etc. and so can be at least partially attributed to "human error", but many are not.

    1. Charles 9

      Re: So we'll just ignore those cases where the driver saved the day?

      "For some reason, this data isn't routinely collected, but if we want to make an informed decision we need it. In aviation, it is suggested that the number of incidents (near misses) is about 10 times higher than the number of accidents."

      WHERE is this suggested, because firsthand experience tells me it's the opposite?

    2. Daniel 18

      Re: So we'll just ignore those cases where the driver saved the day?

      "I admit that some of the avaition incidents are due to things like interruptions etc. and so can be at least partially attributed to "human error", but many are not."

      One of my less common pastimes is reading air accident reports from around the world. Aviation accidents that are not partially or wholly caused by human error are very, very rare. Most are caused by multiple human errors.

      These are not always pilot errors. Sometimes the error lies in procedures, or maintenance, or aircraft design, but usually you will find one or more people at the root of critical failures that resulted in the accident.

      Properly designed, built, maintained and operated machines that have strong safety standards associated with them are very unlikely to suffer a catastrophic failure.

  17. llaryllama

    The fact that humans are still allowed to control machines with hundreds of horsepower and weighing several tons just to buy milk from the store is baffling to me.

    1. Anonymous Coward
      Anonymous Coward

      "The fact that humans are still allowed to control machines with hundreds of horsepower and weighing several tons just to buy milk from the store is baffling to me."

      Anyone doing that should downsize to an SUV or light truck. Not only will it cost less to run, the parking will be much easier.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like