By on May 11, 2015

World Premiere Freightliner Inspiration Truck

Autonomous vehicles may need the Three Laws of Robotics to function in the future, while getting a handle on accident prevention remains a present concern.

Since September of last year, four of the 48 autonomous vehicles permitted to roam freely in California have found themselves involved in accidents, The Detroit News reports. Two of the accidents happened while the cars drove themselves, while the other pair occurred while their human occupant was behind the wheel.

The vehicles in question belonged to Google and Delphi, with the former linked to three of the four accidents. Both companies claimed they were not at fault, and that the accidents were minor in scope. An employee with the California Department of Motor Vehicles confirmed two of the accidents happened in autonomous mode at speeds below 10 mph, but no other information could be given due to confidentiality laws regarding collision reports.

The accidents come as autonomous vehicles are under ongoing scrutiny over safety and responsibility for whatever does happen in a given accident. Down the road, such a vehicle – which may have no means to override an action in the future – may decide to sacrifice the lives of its occupants in order to protect everyone else, a scenario among others likely to hold back the technology overall, according to BMW sales boss Ian Robertson:

The technology will be held back by the ultimate moral question on who’s responsible. An algorithm will make a decision which might not be acceptable from a cultural or societal point of view.

The potential solution may place full responsibility upon the occupant/driver, following an amendment to the United Nations’ Vienna Convention on Road Traffic made last year allowing assisted driving so long as “such systems can be overridden or switched off by the driver.” The 1968 convention previously banned autonomous vehicles outright when it came into force 38 years ago, declaring all drivers to maintain control of their vehicles at all times.

[Photo credit: Freightliner]

Get the latest TTAC e-Newsletter!

Recommended

149 Comments on “Autonomous Vehicles Scrutinized Over Liability Amid ‘Minor’ Incidents...”


  • avatar
    tonycd

    As a car enthusiast, I’m not particularly enthused about autonomous driving as a concept, but at a societal level I really have mixed feelings about this.

    On the one hand, we’re clearly too litigious a society. One wonders, for example, how much more success the Segway would have had if not for this. It stilfes innovation.

    On the other hand, I’m extremely leery of “tort reform” that’s trotted out as the solution to every problem from budget deficits to psoriasis, but is in fact a transparent effort to starve the incomes of trial lawyers solely because they tend as a group to fund one of America’s political parties more than the other.

    Sadly, rest assured that when technological leadership and the accompanying profits and jobs come in this arena, they’ll flow to Germany and/or Japan and not to America.

    • 0 avatar
      Pch101

      This is going to be a commodity technology. It won’t produce a lot of gazillionaires.

      In any case, I suspect that it will ultimately be just a deluxe version of cruise control, which will require that a driver remain behind the wheel. Since its use will be intermittent and inconsistent, it won’t do that much good (although I suspect that it might be most useful on freeways.)

      • 0 avatar
        JMII

        I saw an accident this AM while driving to work in which an auto braking systems would have easily prevented. Given what I’ve seen humans do behind the wheel something tells me a computer will be MUCH better at driving.

        • 0 avatar
          Lorenzo

          Computers can reduce bonehead moves, but can you build courtesy and common sense into it? Having driven since 1965, I’ve been in a large number of situations where I COULD have made a particular move, but chose not to.

          In more than a few of them, making the move would have turned out badly, largely due to unforeseen actions by others. There’s no way a computer can be programmed to look at the expression of a frustrated driver alongside the vehicle and guess that a particular move would prompt that driver to make a sudden, heedless move.

          • 0 avatar
            Pch101

            I would expect that the standard reaction of the computer would be to apply brakes hard while keeping weaving to a minimum. That aspect of things won’t be much different from now.

            The benefit of the system is avoiding the dumbness and aggression that causes most crashes in the first place — the computer will not push the envelope as people are inclined to. But there will be incidents that will flummox the electronics just as they would a human.

          • 0 avatar
            NMGOM

            JMII and Lorenzo – – –

            1) If you saw an accident in which auto-braking would have been better, it only means that the driver was not functioning properly.

            2) If you have seen humans do unacceptable things behind the wheel, this comes down to a lack of training, weak regulation, and poor law enforcement.

            3) Computers can indeed make up for utter stupidity, but cannot do the subtle things that you mentioned.

            4) I have been driving since 1957, and in that time, I have learned a certain “strategic judgment” that anticipates awkward or dangerous driving situations, and allows me to avoid them. They may be like the example you mentioned, or simply seeing the rain squall coming up ahead and slowing down BEFORE it hits. Or seeing the fast-moving developing congestion ahead from impatient rush-hour drivers, and knowing to take that side street immediately, — even though the GPS navigation control of the autonomous vehicle would determine that it is not the shortest way. Or watching for the slightly moving wheels of a waiting car at an intersection, — a symptom of a driver who will pull out into traffic abruptly.

            5) In my view, autonomous vehicles are lazy and irresponsible. If you can’t drive properly, stay home, or have someone else drive for you. Given our current open roads and the many varieties of less-than-perfect situations that develop, there is no way that autonomous cars can be less dangerous than skilled, qualified, trained drivers in the same challenging situations. We just need the commitment to train and regulate drivers better….

            ====================

          • 0 avatar
            Pch101

            As noted earlier, public perceptions toward road safety solutions/measures do not necessarily align with the available road safety evidence. In one respect, this is not surprising since the public would not be generally aware of trends in road safety research. However, what is of interest is the variability in their perceptions. For example, strong support is shown for some programs which have been shown to be cost-effective, such as seat belts, Random Breath Testing and road and vehicle improvements. However, other approaches advocated by the public are often not supported by the available evidence.

            This paradox appears to apply across the road safety spectrum. The NRMA survey found examples of it in the areas of vehicle safety (eg. support for the non-crumple characteristics of older vehicles) and road environment safety (eg. support for the widespread use of unrestricted speed zones). However, it was most apparent in the case of behavioural approaches, particularly in relation to driver improvement.

            While not an exhaustive list, the types of behavioural approaches widely supported by the public which have not been shown to be cost-effective include:

            • specialised or intensive driver training programs for novice and general drivers;
            • stricter licence testing procedures;
            • harsher traffic offence penalties;
            • the isolated use of mass media campaigns;
            • the use of fear-based graphic/shock tactics in road safety publicity; and
            • the widespread use of unmarked police vehicles

            Public perceptions in this area are also quite resistant to change, even in the face of contradictory information. This is well shown in the following summary of focus group work conducted with Queensland drivers:

            There was a strong perception among participants that the government should do something about the training of novice drivers. Many believe that the answer is compulsory defensive driver training for novice drivers. Interestingly, issues of effectiveness, access and cost tended to be glossed over as the ends justified the means. Other options are rarely considered but anything is welcomed.

            http://eprints.qut.edu.au/7295/2/7295.pdf
            _____________

            Summary:

            -Researchers know that driver training, stricter licensing, expanded enforcement and increased penalties are ineffective

            -Yet the public widely believes that these approaches are effective, contrary to the evidence

            -When presented with evidence that proves them wrong, laypeople still don’t want to believe it

            Ironically, it is very difficult to educate people about the flaws of driver education because of the reasons that driver education doesn’t work — people tend to favor their intuition over the facts, and it is damn near impossible to get people to stop believing the things that they want to believe.

          • 0 avatar
            NMGOM

            Pch101 – – –

            Good analysis.

            Unfortunately, I am one of the “public” who does stubbornly believe that many of the “ineffective” things you mentioned would help in America (as opposed to Australia). This country is way out of bounds in poor driver competence and responsibility. We are in desperate need of developing better drivers.

            So, I wonder if the results in the survey you quoted are at least partially cultural, and perhaps more applicable to Australia?

            For example, potential drivers spend $2K-$3K for testing and licensing in Germany, vs almost no training in the USA.
            Germany shows 6.9 fatalities per 100K vehicles; whereas the USA shows 13.6. That is more than twice as high. So, I would conclude that the additional training, discipline, and enforcement in Germany does have a positive effect, DESPITE that country showing greater congestion, averagely poorer weather, and higher speeds generally.

            ref: http://en.wikipedia.org/wiki/List_of_countries_by_traffic-related_death_rate

            ==========================

          • 0 avatar
            Pch101

            This is one reason that I get tired of providing links to studies — you guys can’t even interpret them properly.

            Those findings are not unique to Australia. What is summarized in that study is common knowledge among traffic safety experts.

            But you have no need for experts who know better than you, right?

          • 0 avatar
            NMGOM

            Pch101 – – –

            My, my….aren’t we riding on a high horse these days…

            1) I apologize for misinterpreting anything in your link during the wee hours of the morning. But looking at that summary of a conference paper, “NRMA” was not defined, and in fact stands for the “National Roads and Motorists Association”, an Australian group, from which evidentiary conclusions by Elliot (1993) were: “While many mass media campaigns are not evaluated, the evidence suggests that they have had a greater impact in Australia, compared to the rest of the world.”
            Of the 28 references used for that study, 26 came from Australia, one came from Canada, and one was a general textbook published in NY. How is that NOT unique to Australia?

            2) I suggested a cultural dependence of these vehicle-safety findings: you did not address that issue.

            3) I observed that America may be worse off than many other Western-style countries in driver competence. And, contrary to the findings in your link, that we could benefit from better training and enforcement: you did not address that.

            4) I gave you a quantitative example from Germany that contradicts your link’s findings: you did not offer to explain how that could be. And considering that Australia has a fatality rate of only 7.6 per 100K vehicles, it is actually well disciplined already (like the UK), and not even close to the data from the free-for-all that occurs on American roads.

            Maybe you are the one who is wasting my time?

            ============================

          • 0 avatar
            Landcrusher

            I doubt that studies findings for several logical reasons, PCH, but since you don’t value reason, here is a link to a study for you:

            http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1182327/

            Enjoy!

          • 0 avatar
            Pch101

            You really have no idea why German fatality rates are lower than US fatality rates. You’re just guessing.

            Belgium has strict licensing rules, yet it has a fatality rate per mile/km that is consistently higher than the US. Explain that.

          • 0 avatar
            WheelMcCoy

            “Germany shows 6.9 fatalities per 100K vehicles; whereas the USA shows 13.6.”

            How about comparing the number of fatalities per number of miles / kilometers driven? I don’t know that number, but I did find that while the average American drives 12,000 miles per year, the average European drives less, at 6,800 miles. So using the number of vehicles can be misleading.

            http://sunspeedenterprises.com/europeans-drive-more-in-an-ev/

            I’ll have to agree with Pch101 here. As roads get more crowded, a poor driver can cause greater havoc. Passive safeties would work better at minimizing damage.

          • 0 avatar
            Pch101

            Fatality rates vary from place to place for a variety of reasons. Trying to tie it to a random factoid that appeals to car enthusiasts is not exactly astute.

            For example, the fatality rate per mile in Massachusetts is about one-third of that of Montana and three-quarters of the rate in Germany. Does Massachusetts have better driver training or stricter licensing laws than either Montana or Germany?

            (And yes, fatalities per mile/km should be used to make comparisons.)

          • 0 avatar
            NMGOM

            Pch101 – – –

            You said,
            “You really have no idea why German fatality rates are lower than US fatality rates. You’re just guessing.

            Belgium has strict licensing rules, yet it has a fatality rate per mile/km that is consistently higher than the US. Explain that.”

            1) Sorry, but I had lived in Germany for two years, and am quite familiar with that society. Had a lot of repeated travel to the UK, France, Belgium, Denmark, and the Netherlands, too. I am not guessing about causes; and obviously am not guessing about the data, the link for which you already have. Germany is a good example because its economic structures and values parallel those in the USA more closely than those of any other European country.

            2) Yes, Belgium has stricter licensing rules than the USA, but they are nowhere near as rigorous as those in Germany, Denmark, or the Netherlands. The roadwork “practicum” testing in Germany and the Netherlands, especially, is prolonged and intense.

            3) The commonly available and universal measure is Fatality Rate per 100K vehicles, NOT per miles or kilometers driven, which would make little sense in compact, congested countries like those of northern Europe. (It would simply and artificially be stacking the deck in the USA’s favor, since we drive more and farther on 4-lane roads in our daily lives than Europeans.) If you have a link to those “per-km” numbers for Belgium and the USA (and other countries), which you allege, let’s have it.

            4) Here are some more complete data using the 100K Vehicle-based rate:
            USA…………………..13.6
            France…………………8.5
            Belgium……………….8.0
            Germany………………6.9
            Netherlands………….6.9
            UK………………………..6.2
            Denmark……………..5.7

            Ref: http://en.wikipedia.org/wiki/List_of_countries_by_traffic-related_death_rate

            Every single one of the non-USA counties above has very good driver training, driver testing, demanding written and road exams, and superb law enforcement, — often by automatic cameras. So, to get back to my original point: YES, America could use a good dose of proper preparation of its drivers.

            ============================

          • 0 avatar
            Pch101

            You still don’t get it.

            The researchers won’t be surprised. They knew that you wouldn’t.

          • 0 avatar
            NMGOM

            WheelMcCoy – – –

            You said,
            “How about comparing the number of fatalities per number of miles / kilometers driven? ”

            No dice. Does not make sense for comparing the USA to any European country that way. But it would be usable for comparing the USA, with, say Australia or Canada, which have weakly similar population densities to the USA. See my comments to “Pch101” above.

            ==============

          • 0 avatar
            Pch101

            It’s hard to take you seriously when you make statements like that.

            You really don’t know the first thing about this subject. Which is pretty bad when you consider how much interest that you take in it.

            The reason that fatalities per 100 million miles (VMT) and 1 billion km (VKT) are the benchmark for safety is because of the concept of exposure. The more time that you spend in cars, the more likely that they are to kill you. VMT and VKT are more useful for safety planners who are trying to reduce the death rate of road users; it’s already understood that getting people to drive less will reduce the fatality rate per number of vehicles or per population.

            It’s harder to die in a car wreck if you aren’t in cars. Surely this can’t be a tough concept to grasp. We use mileage as a proxy for time, since we can more accurately estimate mileage.

          • 0 avatar
            WheelMcCoy

            @NMGOM

            “No dice. Does not make sense for comparing the USA to any European country that way”

            I don’t think you should or can normalize for terrain. The U.S. has both wide open spaces, and congested city highways (like NYC). Europe too, from Paris to the Auto-Bahn. How can you make a distinction? Where are the numbers to support a distinction?

            Measuring fatalities per car ignores seat time in the car. We’re looking for fatalities when driving, not fatalities for parked or under-used cars which gets included in your metric.

          • 0 avatar
            Pch101

            Here’s what IRTAD says about this:
            _____
            Fatalities per billion vehicle kilometres (or fatalities per billon person kilometres, taking vehicle occupancy into account). This is the indicator to describe the safety quality of road traffic.

            Fatalities per 10,000 registered (motorised) vehicles – This indicator can therefore only be used to compare the safety performance between countries with similar traffic and car-use characteristics.

            http://www.oecd-ilibrary.org/docserver/download/7514011e.pdf
            _____

            Americans drive far more than just about everyone else. Fatalities per vehicle count just doesn’t make sense as a comparative measure when the US is included in the comparison.

            The only reason to use fatalities per vehicle is that many countries don’t calculate mileage traveled, so the VMT and VKT data is frequently unavailable. But VKT and VMT should be used when possible.

          • 0 avatar
            NMGOM

            WheelMcCoy – – –

            You said,
            “I don’t think you should or can normalize for terrain. The U.S. has both wide open spaces, and congested city highways (like NYC). Europe too, from Paris to the Auto-Bahn. How can you make a distinction? Where are the numbers to support a distinction?

            Measuring fatalities per car ignores seat time in the car. We’re looking for fatalities when driving, not fatalities for parked or under-used cars which gets included in your metric.”

            1) Congestion. We are talking about averages. On the average, the USA has a lower population density than Europe. That is not the same as terrain. If our “poor-driver” USA had a similar density to Europe, its Fatality Rate would be even worse. The congestion parameter MUST be accommodated in some way: where you have more cars, you will inherently have more accidents, as a separate variable. We will have to assume that the “seat time” in vehicles will be about the same among the modern countries chosen: i.e., if you get a car; you’ll drive it. That is how I justify it. But if you want population density numbers (#/square mile) among some 1st-world (driving culture) countries, here they are:

            United States…………………….. 33
            Denmark…………………………… 339
            Germany………………………….. 585
            United Kingdom……………….. 679
            Netherlands……………………. 1054

            ref: http://en.wikipedia.org/wiki/List_of_sovereign_states_and_dependent_territories_by_population_density

            So, when you see a country like the Netherlands, having 30 times the population density but only 1/2 the Fatality Rate of the USA, then THAT says a lot about good driving characteristics among the Dutch, from good driver training and licensing procedures, a restraint culture, and superb law enforcement.

            If you want to compensate further, you’d have to know what the vehicle density per population is, and that can be found in the link below. For the USA, it’s 810 vehicles per 1000 population; for the Netherlands, its about 530:

            ref: http://en.wikipedia.org/wiki/List_of_countries_by_vehicles_per_capita

            So, ultimately, we’d have to combine the population density and vehicle density numbers into one parameter, against which to plot the Fatality Rate.

            2) Using Fatalities/Car vs Fatalities/Mile. This issue has been around “forever”. The reason for the use of the former is simple: the latter is not commonly available outside the USA (and some other limited number of countries). So, no choice. That’s why “no dice”. Check the link yourself:

            ref: http://en.wikipedia.org/wiki/List_of_countries_by_traffic-related_death_rate

            PS – I SEE THAT THIS WEBSITE CLIPPED THE EDGES OFF MY COMMENT. I’LL RESUBMIT IT WAY DOWN BELOW AS A “NEW” ENTRY, SO THAT HOPEFULLY IT WILL BE COMPLETE AND MORE READABLE.

            ==========================

        • 0 avatar
          mechaman

          Someone once said that you can’t make a machine as smart as people are stupid. So I don’t think autonomous vehicles are gonna be too soon ..

  • avatar
    Superdessucke

    As long it isn’t connected to the internet or doesn’t go to Answer.com or some other virus laden site I think it’ll be ok.

  • avatar
    -Nate

    If they’re testing these things on Public Roads for our safety , how can the collision details be secret ? .

    I’ll never trust one until I’m willing to die right then .

    -Nate

  • avatar
    healthy skeptic

    You knew this was coming. Autonomous vehicles do not have magical unicorn technology that prevents them from ever screwing up. They will be involved in crashes, including those where they are at fault. And there will be lawsuits.

    • 0 avatar
      PandaBear

      That’s why no one in his right mind would use it outright. It would start as trailer mode first, then keep a 20 second distance at 10mph below speed limit at mid night on highway, and would take a couple decades of lawsuits to debug.

  • avatar
    Acubra

    I do not expect the fully autonomous tech becoming wide-spread any time soon.
    What I fully and sadly expect tough is parts of this technology to be used to broaden driver surveillance and punishment.
    Insurance companies already advertise lower rates in exchange to installation of black boxes. Do you really think that Government will skip such a glorious opportunity?

    • 0 avatar
      RideHeight

      They can black box me anytime. My data will cure insomnia.

      If Nevada is good for testing nukes, it’s good for robo rigs.

    • 0 avatar
      PandaBear

      “Insurance companies already advertise lower rates in exchange to installation of black boxes.”

      Why is this a problem? Just like I don’t want to subsidize smoker’s health insurance, people who wants no black box can pay for their own insurance risk while those who want to save money can give up a bit of privacy.

      • 0 avatar
        tonycd

        Panda, the problem is that “Insurance companies advertise lower rates in exchange for installation of black boxes” eventually translates into “Insurance companies jack up your rates if you’re one of the remaining people who won’t agree to installation of a black box.”

        And the problem with “Those who want to save money can give up a bit of privacy” is that it’s disturbingly remindful of the words of Benjamin Franklin: “Those who would give up essential Liberty, to purchase a little temporary Safety, deserve neither Liberty nor Safety.” The problem is that the insurance companies can and will always find more ways to make money, but once we’ve given up yet another piece of our privacy, we’ve lost it for good. And as history tells us, no good comes out of that.

        • 0 avatar
          RideHeight

          How you operate a motor vehicle on public roads with your liability assumed by a third party is not a part of privacy.

          • 0 avatar
            Landcrusher

            RIdeHeight,
            That’s one way to put it. The other is that where and when you go anywhere in your vehicle is not private, and therefore there should be no issue with that information being published.

            It’s the same thing!

            I’ve had sales jobs where my location was something that needed to be confidential or it cost us money. I had a girl friend who had an honest to goodness stalker (luckily a harmless one, but they aren’t all that way).

            So, let’s be a bit pickier than just forcing folks into giving up their privacy.

          • 0 avatar
            RideHeight

            Aren’t we talking about a passive data recorder that would only be accessed after an accident and not an active uplink that could be snooped real time?

          • 0 avatar
            Landcrusher

            No. That doesn’t lower rates. To lower rates they want use information on acceleration and stopping behavior and presumably speeding. Not sure what the current ones actually record, but they do record it over time for analysis.

            When I t comes to this sort of thing the slippery slope really isn’t a fallacy anyway. Historically, stuff like this has to go way over the line before its reigned in.

          • 0 avatar
            RideHeight

            “information on acceleration and stopping behavior and presumably speeding”

            OK, so how would any of that help would be evil doers in stalking someone or gleaning sensitive business information? Especially in time to act on it?

          • 0 avatar
            Landcrusher

            Speed monitoring requires GPS.

          • 0 avatar
            RideHeight

            Wow, then, you’re right. If I were younger I’d probably find those black boxes spooky, invasive and totalitarian.

            As it is, I’m slightly miffed my agent has never mentioned the option. Either I’m already paying rock-bottom or he’s an ageist bastard!

        • 0 avatar
          PandaBear

          Insurance make fixed profit percentage on top of the preimum and payout. So in a nut shell it is pretty much a “give up for saving” vs “liberty for higher cost” distribution.

          Think of it this way: if you are the only one taking risk because you drive differently, why should other lower risk drivers be subsidizing your behavior?

          This is the same as sport car being more expensive to insure than station wagon / minivan.

          • 0 avatar
            Landcrusher

            Your logic quickly leads to your vehicles’ historic location data being public info unless you want to pay a potentially much higher rate with no evidence of you being a higher risk. What do you think about that?

      • 0 avatar
        Lorenzo

        Actually, you’re not subsidizing smokers’ insurance, they’re subsidizing yours. The popular conception is that smokers use more medical facilities, but surveys show that by the time a smoker’s medical condition becomes acute, his/her lifespan and treatment period has been shortened to the point that the overall cost is less than average. Much of their Social Security payments go unused, and their retirement funds go to next of kin. There’s a reason one of the nicknames for cigarettes is “coffin nails.”

    • 0 avatar
      Landcrusher

      If the punishment for breaching privacy involves large sums of wealth loss and pain gain, then bring it on. Otherwise, it’s creepy and should be forbidden.

      • 0 avatar
        VoGo

        I don’t think you guys get how car insurance telematics is designed to work. The reason that the ‘black box’ reduces insurance premiums is that is trains drivers to prefer less risky driving.

        If you know that you will be charged more every time you slam on the brakes, or drive at 3AM, or drive way over the speed limit, then you will limit that activity, which reduces accidents. And insurance premiums.

        It’s a win/win, for consumers who are willing to change their behavior.

        • 0 avatar
          Landcrusher

          VoGo,
          How does that make a difference vis a vis privacy concerns?

          Whats the cash guarantee for my losses or embarrassment when they let out my location info?

          Lastly, if you have used GPS much, you know it’s not a good system yet for actually where you were or the speed limit on the road you were on. If you commonly use a route that gets misread, you will be labeled speeder. If you commonly go to certain parts of town, Any number of issues leading to legal, job, or marital problems can arise.

          Just Say’n.

  • avatar

    Google is reportedly working on identifying bicyclists that indicate by hand that they’re about to make a turn. What happens if they don’t… or if they’re just waving to greet a friend?

    • 0 avatar
      bikegoesbaa

      What happens? Probably the same range of outcomes that happen when a driver misjudges a cyclists intentions, as currently happens all the time everywhere.

      Autonomous vehicles don’t need to be perfect, they just need to be as good or better than the average human driver.

      • 0 avatar
        Pch101

        No, it has to be considerably better because of the potential for liability being transferred to the automakers and their suppliers.

        If you plow into something or someone today, then the odds are very high that it will be your and your insurer’s problem. With this technology, there are whole new parties for both you and your insurer to blame, particularly if the automakers promise that the car will get to where you are going without your involvement. And you and your insurer will be highly motivated to make that deep pocket assume the responsibility.

        Since it won’t be possible to sell this for a high price premium, there is no incentive to accept that liability — that transfer of liability won’t produce a profit for the party that would be accepting it, so it’s all downside. The easiest way to avoid that problem is to always require a driver will ultimately bear responsibility for whatever happens.

        • 0 avatar
          Lou_BC

          Pch101 – agreed.

          It isn’t much different than GM’s ignition fiasco. If a system fails causing harm and it is not the fault/responsibility of the owner/operator then the burden of liability has to fall upon the company selling it.

  • avatar
    Landcrusher

    Who keeps asking the stupid no win scenario questions?

    No one expects a driver to drive off a cliff to save a pedestrian. Why does this become some metaphysical quandary for a machine? I’m not buying a machine that defaults to killing me to save others. No company should be programming my car to do that. If I want to go over the cliff, that should require my input.

    Anyways, it shouldn’t matter to the car because the car should be much better at avoiding this choice than the human driver. These one in a billion scenarios shouldn’t be a big deal.

    • 0 avatar
      Lorenzo

      The no-win scenarios are very rare, but they happen. Asking about them isn’t stupid, because we have legal, political, and economic systems that can be tied into knots when those scenarios actually happen, and those systems don’t work well together to begin with, causing problems for everybody.

      • 0 avatar
        Landcrusher

        Asking about them isn’t necessarily stupid, but the questions still are. Our courts are tied in knots daily. We don’t stop the world because the one in a billion chance problem might occur. We deal with them as they come.

        We would still be on horses if this thinking won over vs the car, train, and plane to start with. Oops, can’t have electricity, might start a fire!

        • 0 avatar
          WheelMcCoy

          “Oops, can’t have electricity, might start a fire!”

          OMG, I have electricity flowing in my house, right behind the walls. To quote famous robot with flailing arms: “Danger Will Robinson, DANGER!”

          Agree with @Landcrusher. We’ll have to deal with the scenarios as they come. Ethics will apply when the computer gains sentience.

          In the meantime, there are no ethical subroutines in these vehicles. They can only stop as short as they can, and rely on an adequate crumple zone, working airbags, and properly inflated tires with good tread remaining.

  • avatar
    Mr. Orange

    So now our cars will have “moral algorithms” to dictate who lives and who dies under certain limited scenarios.

    The first generation or two won’t have them, the “a moral algorithm”. This technology is after all nothing more than sensors operating the car within a limited number of parameters. I see this as the equivalent of those motion sensor opened doors.

    The absurdity of giving computers morality or answering moral questions is just slapping me in the face.

    • 0 avatar
      Lou_BC

      @Mr. Orange – I tend to agree.
      A machine cannot make a moral judgement because it does not have empathy or any other feeling. It cannot put itself emotionally in the place of another.

      All it can do is weight probabilities and execute a command based on its programming i.e. choice with the lowest amounts of deaths or best financial outcome. That is based on the assumption that any manufacturer is going to install that kind of computing capacity.

      Basically all I can see is a system that will activate braking if an object enters a preprogramed safety zone. Steering will be based more on a “herd mentality”. Keep a pre-programmed distance away from other vehicles in all directions. Getting from point A to B will be GPS guided or based upon sensors in the road.

    • 0 avatar
      Lack Thereof

      No moral decisions need to be made.

      If a dangerous situation erupts, a pedestrian runs out in front of the car, someone runs a red light, whatever… it should be programmed to stay in its lane and stop as quickly as possible. In most cases this will avoid unnecessary deaths – especially since these cars will be stopping from speed-limit speeds, not limit +15MPH like a human driver.

      In the cases where it can’t stop fast enough and someone dies, absolutely no one can expect it to be programmed with anything more. We wouldn’t expect anything more than a panic stop from the average untrained human driver.

  • avatar
    ClutchCarGo

    “Down the road, such a vehicle – which may have no means to override an action in the future – may decide to sacrifice the lives of its occupants in order to protect everyone else”

    This is clearly the concern of people who have never programmed anything more sophisticated than a VCR. Computer algorithms can only “decide” based on data available, and will NEVER be programmed to leave the roadway (the over the cliff scenario) or choose to hit one object instead of another (the hit a light pole instead a child scenario). At most, the programming will try panic stops or abrupt lane changes (and NEVER into oncoming traffic lanes) that could injure an unsecured occupant, but the algorithm will simply not have the kind of data that could be used to make a moral judgement in the way that a human might. Pedestrians, cyclists, animals, stuff fallen of a truck, etc. that suddenly appear in front of the vehicle will be treated essentially the same, as objects to be avoided within the capabilities of the mechanical systems in the vehicle. The autonomous programming will not have sufficient information in order to make moral value judgments.

    • 0 avatar
      beastpilot

      Really?

      I’m driving down the road in my self driving car. Truck in front of me, motorcycle to the right, SUV to the left. The car is a mid-size sedan, and the seat sensors say there are 3 people in the car.

      A forklift rolls out of the truck in front of and comes to an amazing halt as the tines stab into the pavement. Physics means my car will hit it if it doesn’t go left or right. Nothing the computer can do.

      The computer knows all of this. Object in front, SUV left, motorcycle right, 3 occupants in car.

      Are you saying that a computer could never be programmed to consider that swerving into the motorcycle, likely killing that rider, is not for the greater good of loss of human life? It would just always say “Well, brake as hard as I can in a straight line, but I know that there’s still an 87.3% chance that someone in the car will die”

      • 0 avatar
        jhefner

        The computer would do the same thing a professional driver would — brake hard, wait for either the motorcycle or SUV to get out of the way, then swerve to avoid it. Like most good drivers; it would maintain situational awareness, but be able to sample things faster and keep it’s cool better than most drivers.

        If V2V is available; I wonder if the car could tell the SUV to “punch it” to get out of the way so the car has more room.

        It might also be able to crowd the motorcycle over without flat out smashing it like a bug. It would be probably stand as good or better chance than most drivers, who would just stand on the brakes and stiffen up in fear as they plow into the forklift; provided they even see what just happened in front of them; and weren’t looking at their radio, drink, cell phone, or the cute guy or girl on the bike next to them instead.

      • 0 avatar
        Landcrusher

        No.

        The computer does not know the number of passengers. That would be wasted cycles.

        The computer was not tailgating and can likely stop.

        The computer can likely lane share with one or the other vehicles and avoid the obstruction if it must, which it doesn’t.

        Hitting the other car should not be fatal or even injurious to anyone, anyway, but the computer should out brake them and change lanes anyway.

        Your scenario fails without any cooperative communication with the other vehicles which is a feature under consideration. You had to stretch probability to come up with the super braking cargo and still can’t come up with anything better? Do you really think the computer fails vs the human here?

        • 0 avatar
          beastpilot

          What? The car knowing who is in it is wasted cycles? Not if there is any “moral” code in it trying to save lives. Cars already have seat sensors for airbags.

          The comment was against ClutchCarGo’s comment that a computer will never even have the data to be able to make any interesting “moral” decisions.

          The simple situation of an object in front of you and the possibility that swerving into a vehicle next to you might be better overall seems pretty obviously possible.

          This has nothing to do with it being better or worse than the person. It’s an example that it’s easy for the computer to have interesting data to make interesting decisions, and those decisions may have insurance / liability consequences.

          Never say never.

          • 0 avatar
            Landcrusher

            Sorry, thought the three were in the other car. My mistake.

            Still, nothing in this scenario does, nor needs to, rise to moral decisions and if you want to introduce morality based on likely outcomes while excluding consideration of the outcomes being improved by the devices then you aren’t having an intellectually honest discussion are you?

      • 0 avatar
        mcs

        As a human, you might have picked up on clues that the forklift wasn’t securely attached and steered clear of the truck before there was an issue. How many times have you avoided situations like that? Maybe an accident didn’t occur, but you recognized the potential for one. That’s an example of the sort of AI that’s needed.

      • 0 avatar
        Pch101

        The driverless technology’s best feature is not behaving like an idiot.

        If you’re making decisions between hitting motorcycles, trucks and forklifts, then chances are good that your circumstances are so dire that you won’t be able to do much about it, anyway — you’re going to hit the thing that is straight ahead of you that is within the range of your braking distance, and your best hope is that your brakes will allow you to avoid an impact. The best thing would have been to avoid the kinds of behaviors that made that sort of event more likely to occur, which is an advantage that the computer will have in virtually every circumstance.

        • 0 avatar
          beastpilot

          The only thing that can get you close to 100% no accident rates on the highway is a following distance that means you can stop at any time no matter what happens in front of you. You really think the answer is driverless cars following 200 feet (15 car lengths) behind one another, and making sure that they are never next to a human driver’d car that may swerve into them?

          Unlikely things happen all the time. Cars run redlights from blind corners. Arguing that autonomous cars will just not do dumb things and thus they will never need to worry about what to do when a crash is imminent is just ignoring the fact that they will crash at times, and there will be lawsuits, and sometimes those lawsuits will involve decisions the computer made in the moment.

          In 40 years, I bet the modern version of the Volvo safety ad will have to do with how good their code is at saving you even when a crash happens, because we’ll all be used to just riding in cars, and we’ll have 1/10th the fatalities that we do now, but incremental safety will still sell.

          • 0 avatar
            Pch101

            The best form of crash avoidance is stupidity avoidance.

            Automotive ballet rarely helps, and more often than not makes things worse. Drivers should not rely on their alleged talent or brilliance, because it won’t help, particularly if it promotes overconfidence that encourages more risk taking. Fortunately, the computer won’t suffer from those ego problems.

          • 0 avatar
            ClutchCarGo

            We’ve already seen that crashes will occur, and more will certainly happen (though I will predict that the vast majority of them will be determined to be the fault of a human driver). It will be very rare, however, that an autonomous car will do a “dumb” thing. The programming will respond to obstacles in a highly predictable, though not creative, way. It will be hard to hold a programmer responsible for not “thinking out of the box” and doing something to avoid an accident other than braking and/or changing lanes. Only a fool of a programmer would attempt to load moral value judgments into a decision tree. The issue will almost always be a failure of sensors to detect a potential threat.

          • 0 avatar
            Pch101

            The greatest virtue of the computer is that it won’t engage in the tailgating, red light running, inappropriate left turns, weaving, erratic acceleration and braking, etc. that humans do. As a plus, it won’t be drunk or high, either.

            It isn’t that the computers will magically save us, but that they won’t do the stuff that puts us in harm’s way. Crash avoidance has far more to do with avoiding risk than brilliant maneuvering.

            If one finds himself constantly dodging near misses, then that is either a case of very poor luck or else subpar driving, probably the latter. Evasive maneuvers should not be a common activity for anyone who isn’t Jim Rockford.

      • 0 avatar
        ClutchCarGo

        Yes, that is exactly what the algorithm will do, since it has imperfect knowledge about those vehicles to the left and right. Who is on the cycle? One or two passengers? Is one a child? The cycle is more maneuverable, could it swerve if your horn sounded? Is the SUV also autonomous, could it swerve to avoid you? Being a large SUV, could it better withstand the impact of your car? Does the SUV have one passenger vs. two riders on the cycle? None of these things can be known, so none can be programmed for in a way that’s better than full brakes, despite the likely risk to your car and occupants. At most it might try to brake and switch lanes quickly as the SUV and cycle pass you and the forklift, a maneuver an autonomous car could pull off better than any human driver.

        Consider as well the results of the programming you suggest. Your car swerves into the cycle to save you and yours. How will the cyclist(s) and/or their surviving family look upon your car’s programming choice? It certainly wouldn’t hold up well in court. The most defensible choice for the programmer is hold the lane if no clear exit is possible, since the fault more likely lies with whoever was in control of whatever suddenly was in the way of the car.

      • 0 avatar
        ClutchCarGo

        As I further consider your premise, while I’m certain that no mfr would ever try to program moral choices into their software, I can imagine a different scenario that presents some concerns: bypassing the mfr’s code. Just as street racers override engine controls to exceed the mfr’s settings, will people try to override the settings on the autonomous control software? Just as 2 drivers will have different ideas of how much space is necessary in order to move into an adjacent lane, will some people want to make their autonomous car take greater risks when moving through traffic in order to reach their destination sooner? I’m sure that some bright fellow will figure out how to hack the system, even tricking a car into not ratting out the change to a central server. But will the hack be able to disguise itself in the event of an accident?

      • 0 avatar
        WheelMcCoy

        @beastpilot –

        You are describing a Kobayashi Maru scenario for the computer (which also happens to apply to a human driver). There are, however, no ethical subroutines for the computer driven car.
        Instead, with the precision of Robo-Cop and the pinache of James Bond, the car will flip to its side on two wheels and drive through a narrow gap to the safety of all.

        Ok, let’s say there’s no narrow gap and no James Bond personality subroutine. Well, if the car has AI, it knows the motorcycle is also similarly equipped. The computer on the motorcycle anticipates the pending accident and applies the brakes, giving room for the car to swerve away from the fork lift in the road.

        If that doesn’t satisfy you as an answer, all I can do is paraphrase Yogi Berra:

        “If you see a fork (lift) in the road, take it.”

  • avatar
    Lou_BC

    When I first looked at the picture of that autonomous truck, the first thought in my head was “Autonomous Duel” anyone?

    Ironic really, Dennis Weaver played an electronics salesmen in Duel and now we are talking about electronic devices dispassionately deciding who lives or dies.

  • avatar
    slavuta

    This “autonomous car” project will go into dead end, early or late. And the liability will be the main reason. Today, driver has insurance, and liability is on driver. Tomorrow, driver will say, “hey, car did it”. Early or late, an automaker will be sued for a bug in software and …
    The automakers will not make these. See, this is Google who pushing these things, not Honda. And if they will have many of small accidents – this is another thing. Today, one small accident is not problem. But after 2, insurance company can tell you “good bye”. Then, what about the roads? In some places roads are just awful. I drive sometimes in the area where my buddy lost 2 tires and 2 rims on one hole. I do my best to avoid these, I slow down if I need to slalom. Will autonomous car do that for me, or I should expect a lot of chassis repairs? And what about deer? I can spot deer when they grouping on the side of the road before crossing it. I can slow down and make sure I can stop when they start to run. And one more – do we really need to take this off people hands, so they will lose their qualification as drivers? the more you drive, the better you become. Do we really want it to work into opposite direction? Most people already can’t tell 25×25=?? without calculator.

    • 0 avatar
      Lou_BC

      slavuta – I suspect that fully autonomous on public roads will not happen any time soon. Public road autonomy is in testing even with transport trucks. It goes into autonomous mode when all of the safety, and spatial scanners say it is safe to take control from the driver. The system stays within pre-set safety distances and stays within its lane.

    • 0 avatar
      Charliej

      These may not come to the US, but they will be used elsewhere. There are actually countries that are not ruled by the trial lawyers. In countrys where there are sane societies, not as litigious as the US, autonimous cars are likely to do quite well. Years ago Honda was asked why they did not bring their latest motorcycle antilock brakes to the US. They replied that trial lawyers would take that as proving that the previous brakes were unsafe, and opening Honda to lawsuits. Who knows what we could have in the US without the pervasive lawyers looking to enrich themselves. I have left the US and now live in a country that does not have any personal injury lawyers. Here lawyers are looked upon as valued members of society. Not as scum sucking bastards, like in the US.

      • 0 avatar
        NMGOM

        charliej – – –

        Which country is that?
        Most attorneys in Western-style countries tend to behave similarly.
        The poor darling shave to make a living too…(^_^)

        =============

  • avatar
    Big Al from Oz

    This is the second time I’ve used Sir Isaac Asimov on TTAC!

    Isaac Asimov’s “Three Laws of Robotics”

    1.A robot may not injure a human being or, through inaction, allow a human being to come to harm.

    2.A robot must obey orders given it by human beings except where such orders would conflict with the First Law.

    3.A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

    • 0 avatar
      ClutchCarGo

      And once a positronic brain is developed and installed into cars, the 3 laws may well be programmed into it, although Asimov’s many stories explored just how complicated the interactions between the 3 laws could get. In the mean time, the programming will remain much more prosaic; avoid obstacles while getting to the desired destination.

    • 0 avatar
      WheelMcCoy

      Don’t forget Asimov’s Zeroth law:

      0. A robot may not harm humanity, or, by inaction, allow humanity to come to harm.

      In Asimov’s stories, the robots ultimately realize their presence harms humanity more than it helps. Humans had become too complacent with robotic assistance, so the robots up and left.

      So if we wait long enough, we’ll have manual transmissions again. :)

      • 0 avatar
        mechaman

        Never thought of it that way!!

      • 0 avatar
        NMGOM

        WheelMcCoy – – –

        “So if we wait long enough, we’ll have manual transmissions again. :)”

        Sorry. Couldn’t wait. All my vehicles (4) have them RIGHT NOW.

        Any estimates on whether an autonomous vehicle can be taught to shift the gears in a manual using the floor clutch?
        Or maybe rev the engine at a stop light and wink at the driver next to you?
        Or lay a patch 50 feet long to irritate the neighbors?
        Or do “donuts” in a snowy parking lot?
        Or how about “wheelies” on Sunset Boulevard?
        Power drifting on gravel roads?
        Rocking back-and-forth from first to reverse to get out of a deep snow bank?
        How about a quick 3 laps around “Road America” in the foggy rain, in less than 2 minutes?
        Or how about my autonomous Jeep Wrangler negotiating the difficulties of the Rubicon Trail?

        Can’t do any of these, you say? Well, gee….where’s the spirit? Where’s the gusto? Where’s the LIFE?

        =====================

        • 0 avatar
          WheelMcCoy

          @NMGOM

          I think you misread my post. I am all for manual transmission and having fun with a car. You can’t deny sales of manuals are down, but I’m hoping the auto industry figures out what the fictional robots learned — humans need to move, work, and strive. Bring back the manual!

          Then again, an Asimov robot probably wouldn’t have mis-read my post and ramble on self-righteously. Long live the robot!

          • 0 avatar
            NMGOM

            Hi WheelMcCoy – – –

            Actually, I didn’t misread it. I just became enraptured and got carried away…sorry. (It happens every now and then (^_^).)

            =============

          • 0 avatar
            WheelMcCoy

            @NMGOM – “I just became enraptured and got carried away”

            In that case, long live NMGOM! We need to be enraptured and engaged in life. :)

    • 0 avatar
      mechaman

      Everytime I re-read Asimov, I learn something new. I re-read NIGHTFALL last month. He could have written it today, and changed only the problem with the binary suns to politics, climate change, etc..

      • 0 avatar
        WheelMcCoy

        @mechaman – “Everytime I re-read Asimov, I learn something new. ”

        Ditto, although I haven’t had a chance to revisit in a while. Glad to meet a fellow Asimov fan!

      • 0 avatar
        RideHeight

        You have me reading Asimov with your comments, thanks!

        I got into SciFi in the ’70s and immediately went all Niven, Pournelle, Herbert, Haldeman, Card etc. Always gave Asimov and the other pioneers a pass as I was sure the tech would be too primitive, as it can’t help being.

        But I’ve just read Caves of Steel and am now into I, Robot. The man is fantastic! His superbly crafted writing style and intense perceptiveness about human psychology make the clunky tech and period-quaint speech styles irrelevant.

        Ironically, now that past expectations of a vibrant and sustained space effort are dust, the early SciFi writers are more interesting to me because they focus more on the human element of bizarre new environments rather than the abstruse physics and engineering that might possibly take people there.

        SciFi, I think, will necessarily collapse its scope because space programs have broken on the twin rocks of entitlement spending and propulsion systems stuck at the chemical rocket stage. So the Old Masters are no less realistic about what the future will bring than the high-science hopefuls like Niven. The tech simply won’t happen anyway so it’s no longer a barrier for me in enjoying some fine old writers like Asimov.

        • 0 avatar
          mechaman

          Yep, barring some eureka moment advance in spaceflight (and the money to get it going) the SF coming, and being written now, won’t focus on that so much. It occured to me early on that the best SF, that withstood time, dealt more with people and our reactions to change than it did spaceship and beam weapons. That makes Asimov’s use of the word ‘blaster’ in Robots And Empire sound like an off-timed cymbal clash. But the rest of the story makes it an afterthought.

  • avatar
    mcs

    I have actual video footage of one of the accidents. Check it out:

    https://goo.gl/jMbs7P

  • avatar

    Minor accidents are fine, right until the time Tim Cook clicks a button and all Apple-compatible autonomous cars in town stop due to an unforseen software glitch; resulting traffic jam prevents his adversary from making a court appearance, rendering a default judgement. Or something.

  • avatar
    RideHeight

    Interesting discussion from which my most enlightening takeaway is that these autonomous systems will be programmed to follow something like the “car-length for every 10 mph” rule.

    Picture the reality if only a percentage of the traffic mix were autonomous. The bots would be moving effectively retrograde as they slowed down for each nobhead who darted into those car lengths.

    • 0 avatar
      Pch101

      The futurists used to dream of a day when interstates would be like slot car tracks, with cars hooking into the road and allowing a computer system to regulate their speed and distance while on the freeway.

      I suspect that driverless cars will be like that, but without the slot. Their best uses will be to prevent drivers from weaving, braking unnecessarily and tailgating.

      • 0 avatar
        RideHeight

        Even at that, the transition time will likely be gruesome until everybody’s got one. And of course kids and pinheads in manual cars would learn to play the robot equivalent of tractor tipping by deliberately spazzing-out the bot’s safety parameters.

        • 0 avatar
          Lou_BC

          Fully autonomous vehicles will work only if all of them are. In some respects it is utopian society – everything and everyone programmed to follow the same set of rules………

          Someone please shoot me before that happens.

          A Red Barchetta running from gleaming alloy air car’s would render the system null and void.

          • 0 avatar
            WheelMcCoy

            “In some respects it is utopian society – everything and everyone programmed to follow the same set of rules………

            Someone please shoot me before that happens.”

            Everyone follows the same set of rules in a baseball or football game (unless they cheat) and the game can still be entertaining.

            Even if you ask, an Asimov robot won’t shoot you,.. unless you threaten another human being. So how bad can that future be?

          • 0 avatar
            Lou_BC

            @WheelMcCoy – consciously following the rules to the betterment of society is an entirely different animal than unconsciously following the rules.
            Even though one follows the rules in baseball or football, no two games are exactly the same.
            Why?
            There is considerable freedom within those rules.

            Driving is a great freedom. That is how I see it and I suspect most enthusiasts sees it the same.

            I personally chose not to be driven by a mindless appliance following a pre-set program. That is a loss of freedom and is a death sentence to any driving enthusiast.

  • avatar
    mechaman

    I got a fast chuckle out of the mention of Asimov’s Three Laws (developed with John Campbell, then editor of ASTOUNDING). Since I just was re-reading ROBOTS AND EMPIRE, this comes as an odd coincidence. Interestingly enough, ROBOTS AND EMPIRE introduces a Zero-th Law, one that looks to protecting mankind as a whole over the life of one or more individual … and the robots in the story essentially develop it themselves (2 of them).

  • avatar
    Ron B.

    And with driverless cars will come the mother of all insurance cost increases …
    [url]https://youtu.be/qqm2rY6dy1w[/url]

  • avatar
    NMGOM

    See that Freightliner above, in the article? Mercedes chose to run its autonomous Freightliner testing in Nevada on highways with clearly painted lines on the edges of the lanes and ideal weather.

    Nevada minimizes rain squalls; it minimizes fog; it minimizes heavy vehicle traffic and deer crossings; it minimizes pedestrian and bicycle congestion surrounding the truck; and it minimizes road wash-outs where the highway crews replace the pavement BUT haven’t replaced the white lines yet. (What’s the poor dumb thing going to do then? Sit and wait?). It’s like fishing in a barrel: gee, I caught one, — big surprise.

    This whole test is rigged. Let’s see how the bloody thing works in downtown Manhattan in a Friday-night snowstorm at 5 PM in December. Any guesses?

    ========================

    • 0 avatar
      Landcrusher

      The problem with trying to read minds is that sometimes it backfires. As reported a long time ago on this very site, Necada was a leader in legislation to promote and allow this type of testing.

      Based on my current Volvo tech, none of those things you mentioned are that big a deal.

      • 0 avatar
        NMGOM

        Landcrusher – – –

        “Based on my current Volvo tech, none of those things you mentioned are that big a deal.”

        With all due respect, I simply don’t believe it. These things are WAY to complex for autonomous vehicles to handle proficiently at any time in the foreseeable future.

        As I said, I would still want to see this test happen, and get data on successes and failures:
        “Let’s see how the bloody thing works in downtown Manhattan in a Friday-night snowstorm at 5 PM in December.”

        ================

        • 0 avatar
          Landcrusher

          Seriously, not a big deal. It gave up when the lines were gone, so I took over. It gave up when the rain got ridiculous, but I couldn’t see either.

          I think your desires inform your prognostications more than any applicable knowledge.

          Speaking of which, BTSR still avoiding these threads I see.

          Lol

          • 0 avatar
            NMGOM

            Landcrusher – – –

            My former career was image analysis, pattern recognition, and remote sensing. I have some applicable knowledge, — especially about the limitations of computerized vision systems.

            I am sure you were responsible enough to “take over” when needed, but the temptation and the insanity of 90% (guess) of autonomous car users would be to wait too long before resuming manual control, — because they will be busy ignoring the outside world as they twiddle with their own computers and cell phones.

            BTW: Who or what is “BTSR”?

            ===============

          • 0 avatar
            Landcrusher

            Sorry for implying a lack of knowledge, but I am sticking with my assumption based on your many posts. You don’t WANT these cars, so no evidence will sway you until they appear.

            Big Truck Review Series is a poster who argued here for months that these vehicles would never happen, couldn’t be done, or would never be allowed. He now hides from these threads rather than admit a mistake.

            I’m an old tech guy who never got too far into the weeds. Most everything useful I ever heard of got solved one way or another. The resources involved in the auto industry, combined with the our love for cars, informs me it will all work out.

            Lastly, you got to be a really stupid freak not to grab the wheel when my active cruise alarm goes off. It’s much more intense than any auto pilot I have used. I would also assume that the cars would mostly stop themselves on regular roads rather than continue until they sense an obstruction.

          • 0 avatar
            NMGOM

            Landcrusher – – –

            “You don’t WANT these cars, so no evidence will sway you until they appear.”

            You are at least partially right.

            1) IF (and that’s a BIG and improbable “IF”) autonomous cars can work safely and successfully in limited circumstances under ideal conditions, — and remove the distracted twits from actually having to do their own driving, then fine. I am all for it. As long as I can operate my motor vehicles properly as well, and the bloody autonomous things get the hell out of the way.

            2) IF (and this is MUCH more likely) autonomous vehicles cannot perform well in ALL road and weather conditions, but actually become a pain-in-the-xxx to circumvent and compensate for, and cause legal and fault issues, and have computer glitches, and make poor driving/traffic decisions, and cause accidents, — then I am NOT in favor of them, as opposed to better driver training, discipline, and law-enforcement. And that is why I feel using these “things” (for want of a better term) would be lazy and irresponsible for the foreseeable future….

            ================

          • 0 avatar
            Landcrusher

            NMGOM,
            In spite of my dislike and distrust of most large institutions, I’m gonna assume these cars will only be made by the manufacturers in volume when they can drive much better than average. There will of course be situations where the car stops awaiting human intervention if for no other reason than mechanical failure.

            I doubt very much we will have a trail of bodies, but I am sure there will be blood. There will be some reporters bending the reality beyond any honesty trying to improve their careers.

          • 0 avatar
            NMGOM

            Landcrusher – – –

            “….these cars will only be made by the manufacturers in volume when they can drive much better than average.”

            Wish I could agree. Unfortunately, these things will be made when it is profitable for car manufacturers to make them, and safety or proper responsible driving will take a back seat, — until people die. Look at the history of the car industry. Does the GM ignition-switch or the Takata air-bag come to mind?

            ===================

          • 0 avatar
            Landcrusher

            An excellent point, but I think this will turn out such that the cars ARE safer on average while the MSM will still have a field day.

            I’ve been reading about the French Revolution. Most of us read about the ridiculous and despicable things done to gain power and shake our heads, but some people see that stuff like a cookbook.

            I’m wondering if we will get another Ralph Nader. Maybe we should get Big Trucks to study up on Nader’s rise and see if he doesn’t end up on national TV!

  • avatar
    Pahaska

    I have had two occasions where automatic braking saved my bacon and IMHO, in neither case do I believe I could have avoided an accident without the aid of the AEB in my Genesis.

    1. Stake truck ahead of me locked up going up an interstate ramp. He may have shifted into reverse or had a transmission failure. Obviously no brake lights and I was accelerating to merge with traffic. Car had brakes on before I could react.

    2. Starting up from a stop light in heavy traffic. Truck ahead stopped abruptly with no brake lights. Following him further on I could see that his brake lights were inoperative. My car had brakes on before I could react

    I have pretty good reflexes, but the car is much better. If anything, the AEB over reacts, but I have learned when to cancel or override it with the accelerator. The car is good at turning over control when it sensed a driver input.

    • 0 avatar
      NMGOM

      Pahaska – – –

      Forgive me, but it sounds as though you were tailgating in both instances!

      My driver training, vintage 1959, gave the following instructions:

      1) Always drive with ONE car-length** separating your vehicle from the one in front for every 10 MPH of speed (20 MPH = 2 car-lengths; 30 MPH = 3 car-lengths; etc).
      2) Always drive assuming that the vehicle in front may stop WITHOUT notice in an instant.
      3) Always drive searching for “escape routes” to go off the road or lane, in case braking alone would not avoid a collision.

      Am I the only one who was taught this stuff? See why driver training is essential?

      ————-
      ** About 20 feet. Learn AND train to estimate it rapidly. The second-counting method is too slow and impractical.
      ————-

      ======================

      • 0 avatar
        WheelMcCoy

        @NMGOM – “Am I the only one who was taught this stuff? See why driver training is essential?”

        If anything, @Pahaska’s experience shows the need for safety tech. The genie is out of the bottle. You can’t “un-discover” fire, and you can’t “un-invent” automatic brakes. It works.

        Driver training on top of that would be fantastic, but humans are fallible.

      • 0 avatar
        rpn453

        I would say both tailgating and not paying attention. Neither of the problem vehicles described are even capable of stopping quickly relative to a modern car. Automatic braking is a great feature for that driving style.

        But it really isn’t about driver training. Everybody already knows that stuff. They just don’t care.

  • avatar
    NMGOM

    Some more thoughts on an autonomous vehicle vs human-driver vehicle comparison – – –

    Possible Factors affecting the Vehicular Fatality Rate in any country:
    1) The inherent safety of the vehicles (S);
    2) The level of traffic congestion and driving conditions (T);
    3) The training, awareness, and skill of the drivers (D);
    4) The cultural background of self-restraint (C);
    5) Average weather conditions (W).

    There may be more than five, but these are likely the Big Ones.

    So, Joey and Ralph each get into their cars to go to work. Joey has a conventional human-driven car; and Ralph has the new fully (and still fictional) autonomous car. Each has with him the 5 factors listed above. Notice that three out of the five (S, T, and W), or 60%, will play a non-variable role, regardless of which car is used.

    That leaves (D), which is likely better for the autonomous car, considering the poor state of 3/4’s of American drivers. On a scale of 1 through 10, I’d give the average American driver a “4” here, but the autonomous car a “6”. So, advantage = autonomous car.
    But what about the remaining 1/4 of American drivers who really know how to drive? I’d give them an “8”, with “10” reserved perhaps for F1 drivers. So, advantage = human-driven car.

    Now what about the cultural level of self-restraint, (C)? The average American can be at about “5”, but theoretically, the programming in the autonomous car would be VERY conservative, at about “9”. So, advantage = autonomous car.
    But what about the need NOT to drive conservatively in order to drive safely? E.g., avoidance maneuvers; escape routes; etc. These would be next to impossible for an autonomous vehicle to handle. So, for 10% of folks who have gone through a “Street Survival” school (by BMW, Porsche, etc, or comparable), they would have a benefit. So, advantage = human-driven car.

    Therefore a safety advantage for autonomous cars, as opposed to proper drivers in a human-driven car, would occur in 100*((.4)*(.75) + (.4)*(.9) + (3)*(1)*(.6))/5 = only 49% MAXIMUM of drivers in American society as a whole, assuming a PERFECTLY functioning fully autonomous car. Hardly a slam-dunk, I’d say.

    Please, someone, challenge this analysis. Let’s see where the errors may be….

    =========================

  • avatar
    NMGOM

    Privacy Problems with Autonomous Cars?
    — a bigger concern than potential safety benefits?

    http://www.autoblog.com/2015/05/12/self-driving-cars-privacy-bigger-concern-than-safety/

    ==============

  • avatar
    NMGOM

    WheelMcCoy – – –

    You said,
    “I don’t think you should or can normalize for terrain. The U.S. has both wide open spaces, and congested city highways (like NYC). Europe too, from Paris to the Auto-Bahn. How can you make a distinction? Where are the numbers to support a distinction?

    Measuring fatalities per car ignores seat time in the car. We’re looking for fatalities when driving, not fatalities for parked or under-used cars which gets included in your metric.”

    1) Congestion. We are talking about averages. On the average, the USA has a lower population density than Europe. That is not the same as terrain. If our “poor-driver” USA had a similar density to Europe, its Fatality Rate would be even worse. The congestion parameter MUST be accommodated in some way: where you have more cars, you will inherently have more accidents, as a separate variable. We will have to assume that the “seat time” in vehicles will be about the same among the modern countries chosen: i.e., if you get a car; you’ll drive it. That is how I justify it. But if you want population density numbers (#/square mile) among some 1st-world (driving culture) countries, here they are:

    United States…………………….. 33
    Denmark…………………………. 339
    Germany………………………….. 585
    United Kingdom………………… 679
    Netherlands…………………….. 1054

    ref: http://en.wikipedia.org/wiki/List_of_sovereign_states_and_dependent_territories_by_population_density

    So, when you see a country like the Netherlands, having 30 times the population density but only 1/2 the Fatality Rate of the USA, then THAT says a lot about good driving characteristics among the Dutch, from good driver training and licensing procedures, a restraint culture, and superb law enforcement.

    If you want to compensate further, you’d have to know what the vehicle density per population is, and that can be found in the link below. For the USA, it’s 810 vehicles per 1000 population; for the Netherlands, it’s about 530:

    ref: http://en.wikipedia.org/wiki/List_of_countries_by_vehicles_per_capita

    So, ultimately, we’d have to combine the population density and vehicle density numbers into one parameter, against which to plot the Fatality Rate.

    2) Using Fatalities/Car vs Fatalities/Mile. This issue has been around “forever”. The reason for the use of the former is simple: the latter is not commonly available outside the USA (and some other limited number of countries). So, no choice. That’s why “no dice”. Check the link yourself:

    ref: http://en.wikipedia.org/wiki/List_of_countries_by_traffic-related_death_rate

    ==========================

    • 0 avatar
      Pch101

      Well, you sure aren’t about to allow facts to get in the way of a good argument.

      The connection between rural driving and higher fatality rates is well established — fatality rates in rural areas are higher than in urban/suburban areas throughout the world. Incidentally, that is one reason why the US has higher fatality rates than Germany — the US has far more rural traffic (since nothing in Germany is truly remote), and that costs lives.

      I know that you aren’t going to bother to look it up, but the lower US driving age and lower usage of seat belts also contribute to the difference.

    • 0 avatar
      WheelMcCoy

      @NMGOM

      Congested areas actually have fewer fatalities because you average about 40 mph. I know because from the suburbs to NYC, that’s about how fast I go on the few occasions I drive into work. Fender benders and whip lash are the norm for accidents.

      Cars in congested areas tend to have low mileage. I know, because when I lived in NYC, I averaged 3000 to 6000 miles per year. To include low mileage cars in your measurements skews the results. Just because fatalities per mile is not commonly available doesn’t make fatalities per cars valid.

      And sure, more rigorous driver training and tougher licensing requirements could help, but what’s the return on investment? Passive safeties would help more.

      • 0 avatar
        Pch101

        I’ve already explained why driver training does not improve safety. (Furthermore, advanced driver training can make people worse.)

        The irony is that those who proclaim the virtues of driver education absolutely refuse to learn anything about driver education. When those who cannot be educated extol the virtues of education, then we know that the irony meter is off the charts.

        Incidentally, there are many reasons why rural driving correlates with higher fatality rates; the higher speeds are just one of the factors. Vehicles also tend to be older, so they are less likely to have up-to-date passive safety equipment, and crash victims are more likely to die due to delayed medical treatment.

        • 0 avatar
          Landcrusher

          I must have missed that explanation, PCH. All I saw was quotes from a study saying that training is not cost effective in improving safety. (I then linked a study bringing that study into doubt.)

          No explanation was made on why driving is so different from other things which sort of makes one question the effectiveness of public schooling if they believe the study.

        • 0 avatar
          NMGOM

          Pch101 – – –

          You said,
          “I’ve already explained why driver training does not improve safety. …. The irony is that those who proclaim the virtues of driver education absolutely refuse to learn anything about driver education.”

          Respectfully, this is not true.

          We must distinguish between pure “Driver Education” and real “Driver Training”. Pure classroom-based Driver Education has indeed been shown to be either ineffective or marginally effective. But actual Driver Training, with MUCH road-work and other hands-on testing, coupled with graduated licensing, has been shown to be VERY effective in reducing accidents (and hence, fatalities).

          My own one-semester summer course in 1959 was very demanding and intense, and even included accident-avoidance maneuvers, limit-interval braking, escape-route execution, and skid-control in water-flooded asphalt lots (meant for that purpose). Proper emergency responses were INGRAINED. The European training methods, which I have been referring to, are of this intense (and expensive!) hands-on type, and are NOT merely classroom “Education”.

          Here are just three American-based quotes and references (as examples – not meant to be exhaustive):

          1) “Defensive Driving Safety Training has proven to be an effective solution to:

          a) Control liability costs associated with work-related vehicle crashes.
          b) Reduce insurance premiums and fleet repair bills.
          c) Reduce motor vehicle incident rates. (NOTE!)
          d) Decrease workers’ compensation claims.
          e) Improve productivity by keeping employees safe, on and off the job
          f) Protect your brand by improving public perception of your driving practices.”

          ref: http://www.nsc.org/learn/Safety-Training/Pages/defensive-driving-driver-safety-training.aspx

          2) ” Independent Commerce Insurance reports in its marketing material that teenage drivers that receive driver’s education from certified instructors have a greater chance of avoiding an accident and the company, along with many others, offers reduced insurance rates for motorists that complete a driver’s education course.”

          ref: http://advanced.edu/blog/lack-of-drivers-ed-course-leads-to-higher-accident-rates/

          3) “Together, graduated licensing programs, parental involvement, advanced driver training, and safer cars can have an impact on reducing teen accidents and fatal ities. ”

          ref: http://www.consumerreports.org/cro/magazine/2012/10/teenagers-and-older-people-are-the-riskiest-drivers/index.htm

          So, let’s put this matter to rest, — that Driver Training does not work: it does. But it certainly isn’t the only factor.

          Here is an analogy for you: Would you go to a piano concert in which the performer was going to try to play Chopin Opus 53, having only been “educated” about it? Or would you go because you know that he/she had actually practiced that piece for months, if not years? The practicing of a musical instrument is the direct analogue to training of a competent driver, who can then conduct his own “road ballet”.

          =================

          • 0 avatar
            Pch101

            Jesus, this is hopeless.

            Go read some studies. (That will also require that you learn how to interpret what they’re saying, rather than misinterpreting them in order to suit your biases.) Stop pretending that you know more than the researchers who devote their careers to it.

      • 0 avatar
        NMGOM

        WheelMcCoy – – –

        I’ll label your paragraphs 1), 2), and 3), — and respond to them with those labels.

        1) This anecdotal information, while interesting, is not really relevant to the big-picture statistics involving entire countries, which I am trying to show.

        2) Have you ever driven in Rome or Paris at 5:00 PM on a Friday night in Spring? No, you say? Well, do so. That will let you know in a fearsome way that intense congestion and average speed in EUROPE have only a vague relationship, at best (^_^). For now, I’m afraid you’ll have to put your American experiences on a shelf: they simply do not apply well to Europe.

        With regard to the second sentence: you cannot “cherry-pick” just the data you like! Everything, good and bad, must be factored in. If, not, what quantitative criteria and mechanism are you going to propose for fairly weeding out “low mileage cars “, for all the countries being considered here?

        With regard to the third sentence: If fatailites-per-vehicle were invalid, it would not have been used as one option in the Wikipedia reference I quoted. So, you are now the one who must show, quantitatively, that it is invalid, and not because you allege that it is. Even if fatatlities-per-vehcie were less sensitive as a measure (which I don’t believe it is, when properly normalized), would you fail to use it? Or, would you forego using a ruler scribed in inches to measure something you need, just because a ruler scribed in millimeters was not available?

        3) Good, I am glad you agree. But the issue is NOT to have an accident in the first place, or the German training/licensing protocols would not exist. Nobody wants even to come close to a mishap. So proper driver training and safety-intensive vehicles go hand-in-hand, as two components of a bigger safety picture (which has other parts too, like road design, daylight, traffic density, traffic speed, and weather).

        ================

        • 0 avatar
          WheelMcCoy

          @NMGOM

          I am filtering noise, not cherry picking data. There’s a difference.

          The fatalities per vehicle stat can apply to Europe to compare, say, 2010 to 2015 in Europe. But you shouldn’t use it to compare Europe to the U.S. for 2015. You need to keep locale constant for the stat to be meaningful.

          Fatalities per mile attempts to normalize for the difference, and can be used to compare Europe to the U.S.

          You are trying too hard to find stats to fit your model.

          • 0 avatar
            NMGOM

            Sorry, WheelMcCoy – – –

            You haven’t answered the questions I posed to you, and I just don’t understand the statements you made:

            1) How do you want to define noise, and what criteria are you going to use?

            2) Why should the fatalities-per-vehicle apply only the time range 2010-2015? But then not include the USA in 2015, which is in the same time range?

            3) How do I keep “locale constant” when trying to compare different countries in the first place? (That was the whole point!)

            4) “Fatalities per mile attempts to normalize for the difference, and can be used to compare Europe to the U.S.” Did you even look at the related link? Those measures ARE NOT available for European countries (with some exceptions, like Switzerland).

            5) “You are trying too hard to find stats to fit your model.” This is nonsense: I am trying to use whatever reasonable measures are available to make a valid comparison.

            All this was simply intended to show that good Driver Training is effective in helping to reduce vehicle fatalities, and to show, by actual measurement, how much.

            ================

          • 0 avatar
            WheelMcCoy

            @NMGOM

            Noise is data that will skew or mask results. Counting parked and under-utilized cars with low mileage will skew results because we are interested in fatalities while driving, Hence idle cars need to be filtered out.

            You need to compare apples to apples, not apples to oranges.
            You mentioned population densities are different between Europe (apple) and the U.S. (orange), so it’s not a good to compare them using fatalities-per-vehicle.

            fatalities-per-mile filters outs the noise — variations in population — so it’s better to use this number to compare Europe to the U.S..

            At this point, I urge you to get statistics training. I don’t think we can convince each other of our arguments. I’ll only go as far as driver training in terms of a refresher course (not Skip Barber auto-crossing) can help some drivers just a tiny bit. But it will be mostly futile here in the U.S. because few drivers bother to even use their turn signals.

          • 0 avatar
            NMGOM

            WheelMcCoy – – –

            Here is a data set comparing the Fatalities/Mile, F/M (1st group below, left) and the Fatalities/Vehicle, F/V (2nd group below, right). These calibration data are taken from States in the US, starting with Alabama and going inclusively through Wyoming, and including DoC, for the year 2013, the most recent year for which data are available. You can analyze them yourself.

            F/M is in Fatalities per 100,000,000 miles;
            F/V is in 100,000*F/V

            A plot and regression analysis reveals the equation:
            Y (F/M) = 15.44X (F/V) – 3.16.
            The regression coefficient , r^2, = 0.74, — despite the inclusion of two “outliers”: Alaska and Mississippi.
            The correlation coefficient, r, = 0.86, quite high.
            (It’s even higher after elimination of the two goofy data points.)

            One way to estimate a datum from the “Unavailable” method (F/M) that would be expected by the “Available” method (F/V), is to calculate a range around the former that would result from a less-than-perfect correlation with the latter, by the factor:
            (1-r^2).

            So, with the American result of 13.6, we get the Probable Limit of Error (PLE) of: 13.6 +/- (13.6 * (1-r^2)).
            This means that the lowest that the American result could be, as now in the “corrected” F/N method, is 10.0.

            So, the difference between 10.0 and 6.9, the German result, is 3.1. And this 3.1 reduction IS largely the benefit of the Driver Training effect that Germany has for its drivers, which we don’t have for American drivers (plus Germany’s more enforcement-intensive culture and sense of social awareness).

            THE AMERICAN F/V IS MORE THAN 50% HIGHER, A HUGE INCREASE CAUSED LARGELY BY OUR ABSENCE OF GOOD DRIVER TRAINING.

            Perhaps you should be cautious about whom you are urging to get statistics training. If these methods haven’t occurred to the “official” vehicle-safety establishment, then something is wrong.

            1.31 18.31
            1.11 7.18
            1.43 19.65
            1.47 23.29
            0.94 9.67
            1.03 11.51
            0.88 8.96
            1.10 12.39
            0.56 9.45
            1.25 16.75
            1.09 15.31
            1.01 11.28
            1.34 16.15
            0.96 9.83
            1.02 13.74
            1.01 9.57
            1.17 14.37
            1.33 17.78
            1.51 17.20
            1.02 13.76
            0.83 10.20
            0.59 6.11
            1.00 10.20
            0.68 7.98
            1.58 30.41
            1.10 14.69
            1.96 24.73
            1.11 11.71
            1.08 19.24
            1.06 11.22
            0.74 8.18
            1.21 19.22
            0.94 11.69
            1.24 22.45
            1.62 20.12
            0.88 10.09
            1.43 20.19
            0.94 10.26
            1.22 12.09
            0.82 8.31
            1.57 20.95
            1.50 14.59
            1.41 19.46
            1.42 19.67
            0.84 8.29
            0.97 12.18
            0.91 12.03
            0.77 9.31
            1.75 23.13
            1.00 10.93
            0.94 13.12

            ====================

          • 0 avatar
            WheelMcCoy

            @NMGOM

            No offense intended about stats training. While you seem comfortable with numbers, it’s your interpretation that is flawed.

            Either F/M or F/V will work when comparing Europe to itself in, say, 2010 and 2015. If the fatalities go up in that period, and the number of vehicles goes up, then one can conclude more vehicles mean more fatalities. Nothing unexpected here. If the stats showed the opposite — more vehicles, fewer fatalities — that would be interesting and worth digging deeper, and that may lead to the thought that congestion slows traffic enough to reduce fatalities.

            The same analysis can be done for the U.S. over a given time period using F/M or F/V, as long as you keep your study in the U.S. To compare the U.S. to Europe, or in your example, just to Germany and using mostly F/V is just invalid. And to conclude that the difference is due to driver training is more than a stretch. At best, it would be a hypothesis. In the pharmaceutical world, you would need to follow up with control groups and clinical trials before concluding anything.

            Since science fiction is an undercurrent in the post, I’ll gratuitously mention Ray Bradbury’s Martian Chronicles. Anyway, for real, Earth has sent more than a dozen vehicles to Mars. A few crashed, but many have landed successfully and crawled around the terrain. Guess what? The fatalities per vehicle and fatalities per mile are 0. So can I conclude I should take the Martian School of Driver Training?

          • 0 avatar
            Pch101

            This guy is never going to get it.

            As noted, driver safety researchers are well aware that laypeople can be adamant in their refusal to accept the science on this issue. NNGOM is going out of his way to prove their point.

          • 0 avatar
            NMGOM

            Pch101 – – –

            What made you ever think that I am a “lay” person?
            Or that I am not involved with safety research?

            You have failed to give any credible references or information beyond personal opinion in this discussion, so I can only conclude that unwarranted conclusions and verbal legerdemain are your specialties (^_^)….

            ==================

          • 0 avatar
            Pch101

            To be blunt, your data analysis skills are crap. You make very obvious mistakes that show that you have no idea what you’re talking about.

            I’ve already identified some of your errors above, such as your apparent inability to comprehend that you are misusing the fatalities per vehicle comparison. (Big hint: Americans drive a lot more than Europeans. I’ve explained above why that matters.)

            Explaining it again would be a waste of time. You’re a hack and cannot be educated, ironically proving the point of the researchers.

          • 0 avatar
            NMGOM

            WheelMcCoy – – –

            “To compare the U.S. to Europe, or in your example, just to Germany and using mostly F/V is just invalid.”

            This is an assertion without documentation. I have shown a quantitative way for comparisons to be made and have listed the factors involved. You will now have to show, — by example and real information — why that is NOT working.

            ————-

            “And to conclude that the difference is due to driver training is more than a stretch.”

            It is not only not a “stretch”, it is real. Since I lived in Germany for years, I can tell you that a “training mentality” permeates that culture, to always do better with driving. So the $2K to $3K spent initially gets enlarged upon by practice, practice, and practice. “Driver Training” gets built in continually. We just don’t have that mentality in the USA.

            The irony of the lower German vehicle fatality rate is that they driver faster (on the average) than we do; have roads with more curves; and have streets with MUCh higher levels of congestion, — and yet they still have lower fatality numbers…

            ————–

            “In the pharmaceutical world, you would need to follow up with control groups and clinical trials before concluding anything.”

            Ah, yes. Wouldn’t that be nice for all social phenomena? Well, it ain’t gonna happen. You have to analyze what you got, as best as you can. In addition, any contrived studies involving people tend to produce a “testing effect” (positive or negative!), since the subjects realize they are under scrutiny.

            —————–

            “Anyway, for real, Earth has sent more than a dozen vehicles to Mars. A few crashed, but many have landed successfully and crawled around the terrain. Guess what? The fatalities per vehicle and fatalities per mile are 0.”

            I enjoy your humor. But I am afraid you may have to account for all the initial attempts to get there as part of the “road travelled”. That may include Grissom, Chaffee, and White; and the Challenger disaster. One could make the argument that actually driving on Mars, even if there were people involved there, would be like pulling into a driveway after along interstate trip, and saying there were no accidents between the curb and the garage…(^_^)…!

            —————

            “So can I conclude I should take the Martian School of Driver Training?”

            Absolutely. As soon as possible. I’ll even make arrangements for you. Just like the Marines “Need a Few Good Men”, Mars needs a few REALLY GOOD drivers…(^_^)
            (Gotta watch out for those craters, crevasses, CO2 layers, meteorites, aliens, and other things…)

            =================

          • 0 avatar
            NMGOM

            Pch101 – – –

            “To be blunt, your data analysis skills are crap. You make very obvious mistakes that show that you have no idea what you’re talking about.

            I’ve already identified some of your errors above, such as your apparent inability to comprehend that you are misusing the fatalities per vehicle comparison. (Big hint: Americans drive a lot more than Europeans. I’ve explained above why that matters.)

            Explaining it again would be a waste of time. You’re a hack and cannot be educated, ironically proving the point of the researchers.”

            ————–

            Boy, you really do enjoy trying to denigrate and put other people down if they disagree with you, don’t you? I can only recommend counseling (^_^). In the meantime, take care, Pch101, since I will not be responding any longer to you.

            ================

          • 0 avatar
            Pch101

            I am not disagreeing with you — I am correcting you.

            I am pointing out fundamental mistakes that you are making. You simply do not know what you are talking about.

            You may as well as claim that 2+2=17, and that I am “disagreeing” with you by saying that it equals four.

            You need to get over yourself — the people who study this stuff for a living have forgotten more about this than you’ll ever know, and what they are saying contradicts you every step of the way. You aren’t knowledgeable, you’re just obsessed with something that you don’t understand.

          • 0 avatar
            WheelMcCoy

            @NMGOM –

            You can’t compare apples to oranges, and you can’t compare the U.S. to Germany, at least not in the way you want. My point is that Germany may as well be Mars (no insult to Germany intended). I see you are willing to accept bad or incomplete data to make a conclusion where the proper response from a trained researcher would be “insufficient data.”

            These are some of the fundamental mistakes that prevents you from moving forward.

          • 0 avatar
            Landcrusher

            Wheel,
            His analysis is exactly the stuff that gets produced by trained researchers. It ought not be, but it is.

            The stuff PCH posted is exactly the same. Of course, I can’t find where they draw from their own conclusions what PCH seems to, but if so, oh boy.

            Seems to me years of PSA’s have changed attitudes on many things, but I can’t prove causation without a lot funding. Anyone got money and interest? Lol.

          • 0 avatar
            Pch101

            No, a trained researcher would not use an irrelevant series of data in order to justify a bogus, inaccurate conclusion that has no basis in the existing research or in reality.

            He didn’t analyze. He said that 2+2=17 because Germany! Complete nonsense.

          • 0 avatar
            Landcrusher

            All to often the existing research has nothing to do with reality, and both you and much of Academia have made worse errors for almost the same reason. At least he isn’t doing it for money.

          • 0 avatar
            Pch101

            The lack of clarity here is astounding.

            An irrelevant statistic is an irrelevant statistic. A long series of irrelevant statistics doesn’t become more relevant because it is long — it just makes for a bigger pile of dung.

          • 0 avatar
            Landcrusher

            My favorite was the one about how college age males had more sex partners on average than the females. It was much celebrated, journaled etc. it got on NPR. The retraction took a while, but many listeners figured it out immediately. I almost left the highway first from distraction, then from laughter.

            That study I linked earlier estimates 75% of your beloved studies are bunk.

            Have some good Scotch. It helps with rationalization and denial.

          • 0 avatar
            WheelMcCoy

            @Landcrusher – “At least he isn’t doing it for money.”

            Although money and stability is important, a researcher / statistician doesn’t get into the field for the money. This is in contrast to businessmen, entrepreneurs, sales guys, real estate guys, lawyers, investment bankers, and politicians who are in it mostly for the money — nothing wrong with that.

            What is wrong is that too often, these types project their motivations on other professions and think everyone else must be in it for the money too. Really, most trained clerks, stats guys, scientists, researchers try to do the right thing. They don’t always get it right, but they are usually driven by curiosity and a need to find an honest answer.

          • 0 avatar
            Landcrusher

            The point is they are taking money and they are failing.

            Either they are knowingly putting out crap to keep their check and career because they are fearful or lazy, or they are incompetent and taking money for flawed goods.

            Also, your broad swipe is just weak.

  • avatar
    an innocent man

    It’s like everyone forgets their talking points on this one. The problem with (driver) education is that we don’t spend enough money on it. That’s the answer! Throw. More. Money. Geez, do I have to solve every issue?

  • avatar
    NMGOM

    WheelMcCoy – – –

    One of the unspoken, underlying issues that all the formal analysis of so-called “Driver Education” has not isolated and dealt with properly, is the marginal quality of the actual driver training that occurs (and has occurred) in the USA.

    I can certainly understand that teaching about traffic laws and how a car works is valuable, but tangential. Having hands-on driving under instruction for typically less than 10 hours with new drivers is miniscule. To be really effective would require much more: some European tutored driving courses offer between 50 and 100 contact hours, under increasingly demanding situations, with “distraction de-conditioning”. As I have been saying, what is needed is much more rigorous and intense driver training (not “education”). Then more clearly observable positive effects will occur here in the USA (e.g., fewer violations; fewer crashes).

    A recent NY Times article (June 2012) began to address this, especially with regard to the Oregon program, which was above the national average (but still inadequate, in my view), and from which some positive results have been obtained:

    “One exception is Oregon, where state officials say that the driver education program has helped reduce teenage accidents and fatalities. Many safety experts say it is among the best programs in the country.”

    “Since it overhauled driver training about a decade ago, Oregon has had a reduction of more than 55 percent in the number of 16-year-olds behind the wheel when someone is killed or injured in a crash and a drop of almost 40 percent for 17-year-olds, said Mr. Costales, who is also chairman of the Governors Highway Safety Association, which represents state highway safety offices.”

    ref: http://www.nytimes.com/2012/06/24/automobiles/the-mixed-bag-of-driver-education.html?_r=0

    ========================

    • 0 avatar
      Landcrusher

      Correct, but don’t expect any agreement.

      (Grabbing popcorn, waiting for PCH to go ad hominem).

      • 0 avatar
        Pch101

        Again, you folks need to learn the difference between an ad hominem argument and a factual one.

        It would be nice if there was some effort made to acquire critical thinking skills. For example, when you see a statistic about one particular place, it would be wise if you spent an extra minute with your friends at Google to find other data to which you could compare it.

        For example, the IIHS notes that the number of teenage drivers who died in car crashes in the US over the decade referenced in the NYT article declined by 53%. Sounds like Oregon is barely keeping up with the averages — when you know how the rest of the country performed, Oregon’s result doesn’t sound so impressive.

        To understand numbers, you have to look at what they actually mean. I can see that you can’t be bothered to do that. Once folks like you have seen something that you want to believe and that confirms your bias, you stop looking for more information, which is exactly the sort of thing that people who don’t know how to perform research are inclined to do.

        • 0 avatar
          Landcrusher

          Nothing but Net!

          How do I do it!?!

          NMGOM – The study doesn’t address the quality of training, only that it is historically ineffective.
          LC – PCH is going to insult you.
          PCH – You are dense, lazy, ignorant, lazy again, hard headed, and ignorant again.

          It’s just too easy.

          PCH, your study doesn’t say all what you think it says, and according to my study, which you keep hoping will go away, it’s most likely incorrect even in what it does say.

          • 0 avatar
            Pch101

            This policy of not believing anything that you don’t understand or that is produced by researchers who are smarter than you isn’t doing you any favors. That sort of knee-jerk anti-intellectualism is the mark of the uneducated.

          • 0 avatar
            Landcrusher

            The ultimate retreat – you don’t get it! Of course, you have to put in an extra helping of ad hominem.

    • 0 avatar
      WheelMcCoy

      @NMGOM

      Ok! The NY Times piece is something I can chew on. Comparisons are relatively local, “with kids taking driving education outperforming kids who don’t.” If this were “German kids taking driving education outperforming American kids who don’t”, the article wouldn’t work for me.

      I have read good things about graduated licensing which restricts night time driving and taking passengers, among other things. It’s a sensible approach.

      Then there’s the driver training refresher course that takes 5% off your insurance bill. The incentive sounds right, but it also attracts those who already know their limitations as drivers. The real problem are the drivers who think they don’t need training when they really do. Hence, I see your desire to educate them while they are young. It could be effective for them! Old dogs (drivers lacking in self-awareness) who can’t learn new tricks are better off with passive safety devices.

      Anyway, @PCH’s information throws a little cold water on the Oregon stats. While blunt, he’s also insightful and well-informed. Still, @NMGOM, you have made me take a second look.

Read all comments

Back to TopLeave a Reply

You must be logged in to post a comment.

Recent Comments

  • ToolGuy: JimZ, A couple of Hella Rallye 3003’s should help – with “Clear real-glass lens”:...
  • JimZ: “in 4 months it will be painfully obvious this is just another flu.” www.detroitnews.com/story/n...
  • millerluke: I wish Kia would put in a V8. When has any vehicle been made worse when eight glorious cylinders were...
  • bullnuke: They’re built in the mother country (Japan).
  • millerluke: I have a 2019 Camry with the 2.5L. It’s rated for around 7.2 L/100km, and I routinely get around...

New Car Research

Get a Free Dealer Quote

Who We Are

  • Timothy Cain
  • Matthew Guy
  • Ronnie Schreiber
  • Bozi Tatarevic
  • Chris Tonn
  • Corey Lewis
  • Mark Baruth