mead cohen berger shevtsova garfinkle michta grygiel blankenhorn
The Costs of Traffic Accidents
Fixing Public Health with Self-Driving Cars

With gas prices falling all over the country people are paying less at the pump and driving more. But not all the consequences are good ones. Last week, the WSJ reported:

Traffic fatalities have been trending upward since 2014, when the price of gasoline plummeted and a strengthening economy spurred more travel. The average price at the pump has dropped more than a $1.30 per gallon, or 35%, since this time two years ago, according to data from the U.S. Energy Information Administration.

States including Vermont, Oregon and New Hampshire have been particularly deadly since the upward trend started in 2014, according to preliminary estimates compiled by the NSC, an Illinois-based nonprofit that collects data from state authorities.

Nationwide, about 19,100 people, enough to fill 382 school buses, died in crashes between January and June of this year… Another 2.2 million were injured…

Traffic accidents are something we take for granted—an unpleasant but ultimately unavoidable element of our modern lives. But it’s really a far deadlier phenomenon than others that get much more media attention—gun massacres and terror attacks come to mind. Moreover, traffic accidents are the leading cause of death for people ages 15-20 years of age.

Beyond the grim cost in lives, there is a sizable economic impact. The NHTSA summarized the costs for 2010:

The economic costs of these crashes totaled $242 billion. Included in these losses are lost productivity, medical costs, legal and court costs, emergency service costs (EMS), insurance administration costs, congestion costs, property damage, and workplace losses. The $242 billion cost of motor vehicle crashes represents the equivalent of nearly $784 for each of the 308.7 million people living in the United States, and 1.6 percent of the $14.96 trillion real U.S. Gross Domestic Product for 2010. These figures include both police‐reported and unreported crashes. When quality of life valuations are considered, the total value of societal harm from motor vehicle crashes in 2010 was $836 billion.

Of that $242 billion price tag, road congestion associated with traffic accidents is a whopping $28 billion. This figure includes the costs of travel delays, wasted fuel, and environmental harm. (And this was before the latest uptick in deaths. The WSJ says that deaths in Vermont, Oregon and New Hampshire are up 82%, 70% and 61%, respectively, in the first half of 2016, in comparison with the first half of 2014.)

Advocates and analysts estimate that self-driving car technology has the potential to cut these numbers by 90 percent or more.

One thing is for sure: we can expect is a lot of pushback against self driving cars. Taxi drivers who hate Uber will loathe self-driving cars even more, and for completely understandable reasons, the Teamsters Union will fight against self-driving trucks as viciously (and as unscrupulously) as it possibly can.

There will be lots of problems with the transition to a nation of self-driving cars, but the benefits look certain to outweigh these. The public health, safety and environmental consequences of a much more efficient transportation system, and the enhancements to the standard of living of ordinary families when cars become cheaper, safer and less necessary (ride sharing in self-driving cars will mean that fewer people will have to bear the costs of owning a car for every adult in the household) mean moving up to a world of self driving cars should be a top priority for our political system.

Features Icon
show comments
  • truthsojourner

    Yes, putting prohibitively expensive automobiles, which absolutely no one will be able to afford except for those who can already buy Teslas, will certainly solve the problem. I’m afraid the era of the Jetsons is not going to get here that fast. Can we think of another, more practical solution, perhaps?

    • f1b0nacc1


      • Andrew Allison


  • Andrew Allison

    There’s a breathtakingly simple approach to reducing congestion, which I first saw in Cologne in the mid-60s and recently, and more extensively in Estonia: the so-called “green wave”. The traffic lights on arterial streets are set so that if the motorist maintains the advertised speed, the lights will all be green. It was really astonishing to heavy traffic moving steadily instead of stopping and starting.

    • Jim__L

      Done in America already, to the extent possible. It’s not easy to arrange.

      • Andrew Allison

        Where? And why not every major city in the world’s most automobile-centric country?

        • Jim__L

          In the downtown of my hometown, for one. North-south arteries during commute hours in Campbell / San Jose, too.

          You don’t notice unless conditions are exactly right, because it doesn’t work unless conditions are exactly right.

          The limits are due to a couple of factors — our road networks (road lengths, cross-street spacing, terrain features) aren’t specifically designed with that in mind, and there are just too many people trying to get from their homes to their places of work.

          Again, telecommuting for all computer-tied jobs is the solution here. Fiddling around with an imposed order, like one they dream is possible with self-driving cars, will not solve the problem, no matter how much people with pathologically orderly minds want to believe in it.

          • Andrew Allison

            In a speed-linked traffic light system, conditions are always right. As my recent experience in Tallinn confirms, it doesn’t matter whether it’s one vehicle or dozens.

          • Tom Chambers

            Ah, this brings back memories….in a city where I once lived, the traffic lights on the North Parkway were perfectly timed for 50 mph. Just drive a steady 50 mph and for 5 miles, every light would be green when you got to it. There was just one problem… the speed limit was 35.

          • Jim__L

            I think the civil engineers who designed the traffic light timing were winking at you. 😉

          • Jim__L

            Scale dozens to hundreds, and Tallinn would have problems.

  • Dale Fayda

    The BEST thing for technocratic nincompoops like WRM and the rest of the commentariat, along with those in our ruling class is to leave us the f@#$k alone! They have all shown themselves to be 100% wrong on every single issue of importance. The “wizards of smart” can’t “fix” anything; they can and do only make it worse. That goes double for this idiotic “self-driving car” idea.

    • FriendlyGoat

      Don’t have a coronary over this, but I agree with you that the hype over self-driving cars is ridiculous.

      • Anthony
        • FriendlyGoat

          Thanks. Vance has a point (several, actually) but this stood out:

          “When people read Breitbart every single day and convince themselves that Barack Obama is a foreign terrorist, that is not a problem of government. That is a problem of community failure and we have to recognize that.”
          It’s not just about Obama. When climate change is called a hoax, income taxes are called “theft”, everybody with any sense is “liberal elite”, etc.——we have serious problems with how people are relating to each other.

          • Anthony

            You’re welcome. Also, I agree and on another site I submitted an observation regarding our inability to frankly try “Reconciliation and Truth” rather than deception and deflection. Here’s something related:

          • FriendlyGoat

            Thanks. I wish I had more patience and sympathy for black writers such as Issac Bailey who would justify George W. Bush, Mark Sanford, Lindsey Graham, Nikki Haley and Tim Scott——but I don’t. Without those kinds of Republicans having been elected regularly in the past 30 years, there would have been no rise of Donald Trump.
            It’s not “okay” (in my mind) for a person to think high-end tax cuts, opposition to public schools, opposition to PPACA, voter ID, gerry-mandering, ridicule of climate change, harassment of LGBTs, no sense on the 2nd Amendment, blather about “small government”, and the rest of the GOP agenda is fine and dandy—-BUT—-that Donald Trump is eliciting too much racism.
            This analysis strikes me as sorta clueless. (Curiously, he noted that other black voters seem to “get it” with Tim Scott—-but not the author.)

          • Anthony

            I don’t know whether the writer being black, white , yellow , etc interferes with content of essay ((and FG, Americans come in all stripes – black Americans have never been monolithic despite propaganda of some). I think his analysis, not being clueless, reflects an attempt by a Southern American – who by experience in South as a black man – to place bigotry in context familiar to his experiential senses. You have exhibited (in our past exchanges), a distaste for a Carson, Scott, Thomas, Bailey, et al. but like you people may their choices. You must prevail on them their mistake

          • FriendlyGoat

            Well, Mr. Bailey is lamenting the latent racial problems of the South and how they are being dredged up and given new life by this presidential candidate in this election. My feeling is that if we had been busier defeating “conservatism” in the South, in the North, on Fox News, on talk radio, on Breitbart, and in the churches that we would be having a lot, lot, lot less problems of the type Mr. Bailey now finds with Mr. Trump. Americans did not just wake up in 2016 and decide to revive racism for the heck of it—–or because Trump told them to.

            A form of multi-purpose disdain approaching hatred has been cooked and cooked and cooked in this country since the likes of Limbaugh and Hannity went on the air over 20 years ago. It has been vocally “against” a lot of things for a LONG time, and that list is now expanded to immigrants, Muslims and the urban neighborhood. But the people who respond to Donald Trump on these matters are responding with the experience of an entire generation’s timespan of “conditioning” on how to be a serial disapprover.

            For a writer like Mr. Bailey to tacitly approve of what most of Republicanism now stands for, such as voter ID and no Medicaid expansion, but be critical of it if spoken with racial overtones by a guy like Trump strikes me as conflicted. The entire GOP platform has not been “minority-friendly” for a long time, and I have to wonder why a black writer who can identify that the right side of the political sphere now sounds racist didn’t seem to know that.

          • Anthony

            On your first and second paragraphs, I agree 100% and stand by you. On third paragraph, I can recognize your identification of “conflicted” in public square. And I’m certain Bailey knows of what you cite (and has know a long time) but obviously has made his own compromises of which manifests in your “conflicted” (this for sure is speculation only as prior to article I was unaware of Issac Bailey).

            Overall, FG, I find nothing to disagree with in your commentary and will take liberty (if you will) in restating observation I shared on another site (which underlays much contention):

            On this continent, we yet grapple with the original sin: Slavery based on skin color (melanin), the psychological constraint of “White Supremacy”, implicit racial social hierarchy, racial iconography, etc. Here’s a proposal: for the last 80 years of the 21st century let us Americans (coming of age generations) end the psychological benefit of the implied racial wage in our country; let’s not only live up to our ideals but also exemplify to the Global World that America means what its Creed states. This proposal’s (ending the racial bribe) execution won’t come easy but the generations born from 1980s onward must take advocacy and agency to end odious and destructive covering on the body politic.

            To that end Bailey’s rendition is correct: “…that racism and bigotry remain primary drivers of the American populace in ways understood and not, that they are confined to a few imagined trailer parks in Kentucky and Georgia and not the sole property of one major political party, even if that party is infected by it more than the other.” My answer to Mr. Bailey’s lament remains All of Us or None as Michelle Alexander has said.

            Indeed, Bryan Stevenson (At EJI -Equal Justice Institute) provides a beginning towards our “Truth and Reconciliation” journey for the remainder of the 21st century. That is, address the myths of racial differences that still plague us (Slave History, Terrorism History and its Trauma, Jim Crow – marginalization, racial subordination, segregation, etc) as well as what W.E.B. Du Bois called the “public and psychological wage paid consciously or unconsciously to our many brothers and sisters who in America identify as white – an implicit social status in today’s America causing much discussed angst and anxiety perhaps,

          • FriendlyGoat

            You have made a worthy proposal. I hope we can work toward it and make it come true. Aside from the Civil War legacy, we have some new racial problems which are not making this any easier.
            1) We have noted that the GOP this cycle has decided via Trump to allow overt racism to be one of the main sales tools to sell the rest of conservatism. We had better not allow that to win or we will be taking several steps backward. (In other words, if a crap strategy somehow works, we can only look forward to receiving more of it.)
            2) We now seem to be mad at Asians—-not only China for its various potential threats, but at all of the TPP countries because of the possibility of a TPP agreement (this one or some other one negotiated in the future).
            3) We seem to be mad at Latin Americans for being hired here by our employers.
            4) We seem to be really mad at Middle-Easterners because of ISIL and terrorism.
            That’s a lot of black and brown people for us (white America) to be mad at all at once. Too many.

          • Anthony

            Insightful and germane observation – the quality of our democracy and the moral character of American society is potentially at risk. A historian once said “even a long democratic history does not necessarily immunize a country from becoming a democracy without democrats.” I’m of of the view that it’s more than about Clinton or Trump and frankly my circumstances change none whomever wins and probably betters materially under Trump; but country as you state cannot afford Trump or tenor he represents.

            Social and political phenomena are complex, FG, and have many causes (in reference to your 4 points above). Instances in which Western Democracies have embraced intolerance, less openness, and weakening democratic political institutions are historical fact. We have to contend against the rising intolerance and incivility (among some) and the eroding generosity and openness that impact the attitudes and behaviors in transitioning times – 30 years of general economic stagnation has not helped. Leadership, leadership, leadership – in conjunction with a society’s moral ethic (moral norms favoring whole societal development).

          • FriendlyGoat

            “eroding generosity and openness” is a well-turned phrase.

          • Anthony

            Thanks, FG.

          • Anthony
          • FriendlyGoat

            Thanks. Vance raises a lot of thoughtful points. One that I have wondered about concerns the destruction of black wealth which occurred as a result of the financial crisis of 2007-08. I wish I could sit down with Ben Bernanke and ask why we could not have saved the banks and the mortgage bonds by spending that printing-press money to pay down the mortgages of individuals instead of just buying up the “toxic” securities to make the institutions more solvent.
            We weren’t supposed to have sold over-priced houses with too-high variable interest rates to so many black people, causing them to lose everything—–and supply the vultures with another way to get richer by re-buying those properties on the cheap for rentals.

          • Anthony

            I got your general point but it was not just housing policies nor Fed choices that impacted many lives beginning Fall 2007. Yet, J.D. Vance offers a refreshing insight to issues from a very seldom considered American perspective. I referenced his piece to you because of it consideration of both Donald Trump and Republican politics.

            Still, I agree mortgage (principal) consideration versus financier relieve could have had reverse priority – but ,FG, that’s where politics come in (we get the government we elect).

          • FriendlyGoat

            Ah, yes, Mr. Trump and Republican politics. After he loses, if he does, we can ask the thoughtful Republicans how come their sheep have lost interest in any of the traditional GOP shepherds. Or, put another way, if liberals do not believe GOP leadership—-AND—–the Trump-following idiots do not believe GOP leadership, who do they think they’re playing to? Just the Chamber of Commerce gang?

          • Anthony

            This entire election process disturbs me and I have not yet deduced why. Cable shows (Fox, CNN, MSNBC, et al ) are not informing as much as trying to grab an audience. And I’m of thought that country needs informing more now than ever. The internet comment boards (generally) are way stations for the liked minded and serve little purpose beyond “letting off steam” – which in itself has some societal benefit I presume. Nevertheless taking the long view and wanting succeeding generations both social and political progress and stability, this vexing nature of our politics leave me pondering. It’s more than Trumpism, Chamber of Commerce, etc. Something, FG, is absent (pardon my digressing).

          • FriendlyGoat

            I’m in a light mood today, so I will offer some answers why “this entire election process disturbs” you.
            1) You (we) aren’t crazy about Hillary Clinton.
            2) You (we) aren’t crazy about Donald Trump.
            3) You (we) have no idea why our country is stuck in this loop of polarization where roughly 48% of the people do not agree with roughly 52% of the people on anything.
            4) You (we) are over-exposed to polls and pundit spin.
            5) You (we) are weary of a gridlocked Congress.
            6) You (we) are frustrated that nobody with any power or influence even cares that the growing wealth divide also means we suffer a growing horse-sense divide, hence a willingness from spinners to try to sell almost any theory on any subject—-no matter how ridiculous.

          • Anthony

            Thanks, FG. Your lists are always instructive (and especially now). And last paragraph exposes much in a few words, again thanks.

        • Jim__L

          Most institutions (media, government) are wholly in the hands of a Politically Correct elite that really doesn’t have the interests of anyone but themselves at heart, and they are systematically suppressing churches because they won’t toe the Statist line.

          He’s absolutely right that elites look down on traditional Americans and our values.

          He’s absolutely right about the Gold Rush optimism of Silicon Valley. There are amazing projects going on here that can change the world, and are thrilling to be a part of. That’s why I’m absolutely convinced that using technology to connect people all over the country to these projects, is the way to solve a lot of America’s problems.

          By the way, you’ll see more of the black working class at the defense contractors still in Silicon Valley, than you will at any of the tech companies.

          Trump is no more dangerous than “What, me worry about exposure of classified information?” Hillary. Less so, because Trump’s election would lead to the strengthening of checks and balances, while Hillary and the Democrats would continue to dismantle them.

          I’m not convinced Trump’s defeat is in any way inevitable. As one of my Silicon Valley techie friends pointed out recently, immigrants here are trying to convince him in broken English why he should vote for Trump. That does not bode well for the narrative that Trump’s support is in any way lily-white… which means his support is far, far greater than any of them could imagine.

          PS – Why in the world does anyone respect Ezra Klein?

      • f1b0nacc1

        While I agree with both of you about self-driving cars (the path-finding and obstacle-avoidance issues are way too complex right now for a reliable, robust solution), trucks might be a different issue. Long-haul trucking between fixed locations would work, and from what I can see most of the auto-truck solutions are targeted at that model. There are a ton of bugs to hammer out still, but it is a fixable problem, and as long as we stay with relatively simple environments (interstates with simple off-ramps leading into large truck stops), I don’t see anything that would prevent some implementation within the next 5-10 years.

        • FriendlyGoat

          Even though TAI took the opportunity to slam the Teamsters for their supposedly vicious and unscrupulous opposition to this, I suspect there will be plenty of opposition from drivers of cars on the interstate and from the insurers of trucks. If you can’t insure that self-driving truck for high liability limits, you can’t have it.

          • Jim__L

            I agree about the liability limits. However, drivers on interstates probably won’t be able to tell much difference between automated and human-drive trucks — they’re both a pain to deal with, unless the highways are spacious enough.

          • f1b0nacc1

            While I am usually willing to blame unions for just about all bad things, in this particular case the Teamsters really don’t have the leverage to do very much. Fortunately their days are numbered, and they won’t go mourned by most….

            Regarding the liability limits, that is an issue right now, but I suspect we are likely to see that disappear before the technology is ready. Drivers and their attorneys might be an issue at first, but given that these would be on federal interstates (a very driver-friendly environment), I suspect that they would find a very unfriendly environment for litigation. We would likely see an interim period with autotrucks carrying an ‘observer’ (just in case) for a few years while the bugs get shaken out of the technology, and this would tend to undercut the options that plaintiff attorneys could bring to bear.

            I will offer the following wager…10 years from today there will be at least some autotrucks in use on the highways (likely federal, possibly others) in a commercial capacity. Any takers?

  • Y.K.

    Self-driving cars will reduce ordinary car accidents. One thing they will not do is reduce congestion. Rather the opposite. The slightly better driving will be more than matched by the enormous induced demand, e.g. by all of the people who couldn’t get a driving licence before but would be legally able to use a self-driving car.

    There are some intriguing ideas regarding completely AI-directed traffic. These unfortunately require removing *all* human drivers from the road. This is much harder than it seems. Here, one doesn’t need to only replace all ordinary cars and trucks, but also every special use car like police vans, ambulances, tow trucks, perhaps even military vehicles… Budgets alone will prevent this from happening for the next 2 or 3 decades. Until then, AI cars will make congestion much worse.

  • from the Sinai

    A few comments on autonomous vehicles:

    1) Human-driven vehicles (HVs) are too unpredictable to share the road with autonomous vehicles (AVs). It follows
    that you can’t start with a small network of AV roads and grow the network, because every trip that can’t be completed on a single network will require transshipping the all the cargo, human and otherwise. (This is also the Achilles Heel of maglev trains.)

    2) Is there going to be just one make and model of AV running just one version of software? If not, how do you ensure that every manufacturer’s AV software is 100% compatible with every other manufacturer’s? Remember, every little glitch means someone will probably die. How do you make sure that the software updates are all compatible with each other and with every previous release that may still be running on an AV somewhere? How do you make sure all the updates are correctly installed? Practical answer: it can’t be done. As they say in New Jersey, fugghedabowdit!

    3) AVs are a terrorist’s dream. Get a dressmaker’s dummy, dress it up like your grandmother, fill it with C4, call an autonomous taxi, send granny to your target, and boom!

    • Y.K.

      1) Judging by current tests, this is actually manageable. The first AVs will drive *very* conservatively, and their faster reaction time should allow them to survive humans.

      2) This will be a huge problem with some V2V or V2I schemes if the industry doesn’t coordinate. But this won’t prevent AVs themselves. So long as the AVs assume all other cars are dangerous and driven by maniacs (I mean humans) they’ll be fine. If our AV can survive humans, it can survive another AV.

      3) You’re not thinking big enough. Imagine a terrorist manages to hack into a model and gain full remote control of other people’s cars… Security is and will be a huge problem for AVs.

      • Jim__L

        1) Judging from current tests, AV’s can’t actually cope with roads that are not specifically prepped for their use.

        Telling people in rural areas (or suburbs) where the roads are not AV-compatible that they have to give up their cars so city folk can ride everywhere drunk or stoned, is not going to go over well.

        If AV’s can’t coexist with HV’s, AV’s are a non-starter. Period.

        2) Anyone who generalizes humans as “maniacs” does not have the perspective to participate in this conversation.

        3) So how is it again that replacing HV’s with AV’s is going to reduce the number of deaths and injuries? Oh wait — it’s not.

        These are pipe dreams. The money and technical expertise expended on AVs, not to mention the time and effort expended on both sides of the fight about them, would be far better invested elsewhere.

        • Y.K.

          1) So far, AVs are actually doing well in testing. Arguably better than HVs already, but the scenarios aren’t completely comparable (yet).

          2) This was from the AV’s ‘perspective’. AV would never pass speed limit, switch lanes haphazardly or be drunk. Also, AVs are unable to participate in conversations.

          3) That is a strong criticism. If the HVs didn’t also start adding hackable computerized systems with similar defects…. At the moment most people ignore the issue which alas suggests it won’t be an impediment until the first disaster.

          • Jim__L

            Doing “well” depends on whether or not you’re talking to a drooling booster or someone with a more honest assessment. I’m getting my information straight from the Silicon Valley techies who test self-driving cars, by the way.

        • Andrew Allison

          You overlook the fact that 90% of accidents involve human error ( There’s a huge dividend to getting human drivers out of the loop. It’s not just accidents.It’s been shown (sorry, I’ve forgeten the source) that congestion is a result of human frailty.

          • Y.K.

            I recall a Japanese model of highway congestion suggesting a main cause are clusters of slow drivers choking traffic. But inter-city traffic will definitely not be improved (or only barely so), and I strongly suspect the other AV congestion improvement would be swamped by induced demand. At least for the first models.

            There are some ideas for AI-driven traffic management and for a public transport model, but the first requires removing all HVs from the road, and the latter won’t happen (incompatible with how modern society uses cars and also with the commercial interests of most entities involved).

          • Andrew Allison

            The extensive real-world testing by Google and others, pursuit of AVs by major manufacturers, and Uber’s plans refute the argument that AVs can’t co-exist with HVs

          • Y.K.

            Never said AVs can’t co-exist with HVs, in fact argued the opposite above.

            I’m referring to AI-directed traffic which is different thing: a further refinement on AV if you will. AV deals with a single car (or truck or whatever). AI traffic management (I don’t recall if that’s the technical term) uses cooperation between cars to improve traffic. Imagine for example the following scenario:

            A large group of AV cars in what would look like a traffic jam. The cars negotiate among themselves and find out most cars are driving straight. They set up traffic so that the entire procession moves at a single fast speed. The cars collaborate so that every car is moving at the same speed, and if any car needs to exit (they know the destinations in advance), the formation can allow it to.

            This entire scheme requires negotiation. If any HV were around, there’d be noone to negotiate with (from AV perspective). Also humans aren’t that reliable with regards to keeping a single speed or signalling before turning… If the HV is unwilling, it can easily break the entire formation – leading to the clustering around the slowest HVs, and negating the benefits of cooperation.

          • Andrew Allison

            My mistake. I agree that AI-directed traffic can’t co-exist with HV.

          • Y.K.

            Here’s an amusing look at traffic, and how AI-cars could help alleviate it:

            Unsurprisingly, the video finds that the best solutions rely on AI coordination and ‘removing’ all humans, but AI cars are still an improvement by themselves.

          • Andrew Allison

            Yup, the solutions are relatively simple. The morons-in-charge simply can’t deal with new thinking. Case in point: a lifetime-or-two ago I was on the Santa Clara County Transportation Committee, and pointed out that extremely easy fix to freeway congestion was to make room for incoming traffic by shutting down the right hand lane a few hundred feet before the on-ramp. As you might expect, I was laughed out of court. A decade or so later, the “traffic engineers” figured it out.

          • Jim__L

            Pretending there is no such thing as a computer error — or that those errors would somehow be less dangerous than human errors, when you’re traveling at highway speeds — is a dangerous fallacy.

            And congestion is a result of too many cars on the road. Some mornings there are simply no accidents on the road, but roads slow down anyway. It gets crowded.

            Only telecommuting is a real solution. That, or distributing jobs more efficiently through our suburban areas instead of stuffing them all into cities.

          • Andrew Allison

            No one is suggesting that autonomous vehicles are fool-proof, just that they are an order of magnitude more fool-proof than human drivers. Simply put, AVs will be involved in far fewer accidents than HVs per mile driven. Furthermore, congestion is not a result of too many cars on the road, but of the inability of human drivers to deal with heavy traffic.

          • Jim__L

            You are asserting “computers are better than humans at a complex activity that requires dealing with the unexpected” that without any evidence.

            To suggest that congestion is not a matter of too many cars on the road is absurd.

          • Andrew Allison

            Google’s autonomous vehicles have caused one accident in almost two million miles, the average for humans is more than 10 times greater.

            As for conjestion, see: et al.

          • Jim__L

            And the testers at Stanford and Google I talk to, tell me that they are behind the wheel for every “oh s**t” moment that those cars experience — and they experience one every 15 of those miles or so — and without the driver in the car the car would frequently be a total loss.

            I’m sorry, they’re not there yet. They may never be.

            And as far as traffic-as-continuum-flow goes, applying fluid mechanics is a trivialization of the problem. If traffic were fully-developed laminar flow in an enclosed pipe, it might work, but it’s not. It’s turbulent, it’s discrete, the article itself argues that it is in some ways analgous to a dilatant fluid, for heaven’s sake.

            People with pathologically orderly minds who like to imagine smoothly-functioning systems where they are in charge and solve all known problems are driving this. Not the tech.

          • Andrew Allison

            I seriously doubt that this is true. First, let’s elimiate the non-AV problems, i.e., an AV being the victim of a human-operator error. Second, in a really “oh s**t” situation, the chances of the AV’s human backup being able to respond faster than the AV are slim-to-negligible. BTW, Tesla’s accident/mile figures are also significantly higher than those for HVs.

          • Jim__L

            Andrew, I’m getting this straight from the people that do the testing of AV’s.

            I’m reminded of a story about a bunch of Dominican Scholastics in a monastery arguing about the ideal number of teeth for a donkey to have. A poor Franciscan pipes up and says, “I rode here on a donkey. Why don’t we go and look to see how many teeth he has?” The Franciscan’s suggestion was dismissed out of hand.

            AV’s aren’t poor little victims of big bad humans here — they’re
            feckless automatons being given a job too big for them to handle.

            AV’s are phenomenal in theory. In practice, they’re just not there yet, and they may never be — there are a huge number of very smart people working on them, and it just hasn’t happened.

          • Andrew Allison

            Jim, what you’re suggesting is that Google is lying about the performance of it’s self-driving vehicles, i.e.. that the autonomous miles are not, in fact autonomous. I really, really hope that you are mistaken. WRT to donkey, the published, as opposed to the anecdotal evidence is as I wrote.

          • Jim__L

            Well, what I’m saying is that the testers aren’t saying the same things as whatever you’ve read. Which, I admit, is unlike Google.

            I think that a lot of people (including the author of this piece and the commenters here, and possibly those who write the reports you cite) really, really, really, really, really want self-driving cars to succeed.

            I think that may be blinding them to the limits of the tech.

  • jeburke

    Talk about chasing unicorns. This fascinating fad will crash and burn, so to speak, when the first “self-driving” car kills a family of six. It’s not about statistics. It’s about being able to hold someone accountable.

© The American Interest LLC 2005-2016 About Us Masthead Submissions Advertise Customer Service