Morning Ed: Automation {2017.02.27.M}


Will Truman

Will Truman is the Editor-in-Chief of Ordinary Times. He is also on Twitter.

Related Post Roulette

146 Responses

  1. Avatar Kolohe says:

    this is a heck of a video, allegedly showing the view from a Daesh drone dropping a grenade bomb on an Iraqi ammo truck.

    (Allegedly, because sometimes the Taliban and AQ tried to pass video of themselves being on the receiving end of kinetic military action as their own work)Report

    • Avatar Oscar Gordan says:

      The drone flies like a quadcopter, so it could be an daesh toy. The bomb doesn’t look homemade, so a knowing eye should be able to guess at make & model. I’m a little surprised how well the bomb fell given no obvious targetting reticle.Report

      • Avatar Kolohe says:

        It’s all a bit too good, like when the Professional Russian was doing Call of Duty(?) viral marketing.Report

      • Avatar Joe Sal says:

        The drone doesn’t ‘bounce’ on bomb release. Also considering the housing diameter of a bomb that could do that, it would have to be a sizeable drone, well beyond toy size.Report

        • Avatar Oscar Gordon says:

          It was hitting an ammo truck, wouldn’t take much bang for the boom. Still, you are right that the drone would have to be huge to not bounce noticeably after dropping that.Report

  2. Avatar LeeEsq says:

    First link makes sense. The more specialized, custom made or artistic goods like fancy jewelry or relatively unique pieces of furniture and clothing are still going to need human input.

    Obsolete teachers: A lot of learning is best done in a small groups like lab experiments for science classes or discussing history or literature or learning how to speak another language. Teachers are still needed for these. I’m not really sure why there is the obsession with auto-dictates.

    The Cowen and Metz articles are related because they are about the potential negatives of automation and AI. I really don’t think that anything can be done to halt the advance of automation and AI because the benefits for the wealthiest are simply too attractive to be resisted. We can mitigate the effects of any disruption they cause by ensuring the productive gains are distributed more fairly than during the Industrial Revolution. The real debate on automation and AI are between liberal economists and policy makers that want to ensure the gains are evenly distributed and the more adamant free marketers that believe nothing needs to be done.

    The first xerox ad was fun.

    The Eli Whitney article shows that a lot of people really don’t care about knowing history accurately but see it as series of stories. Any fact that gets in the way of the desired narrative will be ignored. On a previous Slate article one of the commentators noted that until recently history and movies were how people retained any history they learned in school.Report

    • fillyjonk fillyjonk says:

      Yeah, but online learning often overlooks the lab component, or pushes for virtual labs. This is the bugbear my department is currently debating.

      Personally, I HATE virtual labs and would resist doing them if I were a student. And as a faculty member, I have had a few students come through my classes who did virtual labs for their lower levels, and they are the ones I have to go through the whole litany of glassware with (“This is a beaker. It is for holding liquids. It is not accurate for measuring liquids; for that, you want a graduated cylinder, which is this…”)

      But I can totally see those who are into the idea of “the most at the cheapest” figuring whatever is lost in virtual labs isn’t worth keeping.

      I have 12 years before I can make full retirement and every day that passes I wonder if my job is going to last for that long. And that’s sad, because I really love teaching (when it goes well). And labs *are* the most fun part of teaching.

      But the whole “but online education” thing has been going on for years; we are regularly subjected to lectures, I am not sure why, as to how MOOCs or online classes are the wave of the future, and there will be something like ten “superstar” professors in the whole nation, and the rest of us will work as minimum-wage graders, at least until a good grading algorithm is developed. As I said, I’m not sure why we get that anti-pep-talk at the start of every fall semester but we do. (Perhaps to make us grateful for the jobs we currently have? Even with no pay raises for 10 years and additional duties we’re told to take on? idk.)

      I worry about this at my parents who are retired professors and they laugh off my concerns, but they’re not out in the workforce and I don’t think they fully see how AI is going to change every aspect of employment….Report

      • Avatar aaron david says:

        I think the lab aspect is what is important and will keep the profession going. I can see schools farming (heh) out the in class periods while keeping the lab going as a truly practical way to increase education levels while keeping costs managed and class sizes down. If it frees up more profs to be lab professors (who also research, a necessary part) and keeps the amount of class space open for labs, this might not be a bad thing, outside the social sciences.

        Like you, my father loved teaching and loved his labs (the space was various orchards), but could care less about the politics, which made retiring quite easy when the time came.Report

      • Avatar Brent F says:

        I was at a talk by the Dean of Science at my alma mater on the weekend, talking about how they are pushing to lead in the MOOC field. The way he seems to view it is that the big value of MOOCs is to offer the courses that other institutions can’t, which is why they are leading with putting palentology courses online as its a popular subject (people love dinosaurs) most universities don’t have a palentology department and can’t offer undergrad courses in it. The other area he thought it had legs is offering the standard 100-level lecture online as an alternative to the packed classroom of 200+ pupils.

        You can see what they are doing here:

        So maybe long term it doesn’t replace the professoriat, but makes it more efficent. Spend less time teaching mass lectures that anyone in the field can do, more of the teaching 10-40 students in a more hands on fashion for labs and seminars. Then an explosion of the number of courses effectively availible as you can connect professors in specific areas they specialize in to students interested in the area.Report

        • Avatar Saul Degraw says:

          What was your eating club at Princeton?

          A friend went with her dad to the dad’s college reunion for Princeton and I am just wondering if you were at the same.Report

          • Avatar Brent F says:

            Um, I’m not sure how you got going to Princeton out of my link to the University of Alberta’s online courses. That’s where I went for undergrad and grad and I headed to UBC for law. Both excellent school, particularly in the fields I took at them but not having the cachet of a Princeton.Report

        • fillyjonk fillyjonk says:

          I think MOOCs are a bad fit for us, as

          a. Our drawing card traditionally has been small, high-personal-attention classes
          b. We have an “underserved” student body, as in, people who don’t have a lot of family experience with college, and so “how to college” is part of our instruction
          c. Some of our more-rural kids have crummy internet access

          And yet, we keep hearing how this is going to save us, and I am just very skeptical. It feels to me like we’re trying to do what “everyone” else is doing instead of trying to focus on what makes us unusual.

          I dunno. We try to stay inexpensive but state appropriations keep getting cut (we are down to about 30% of our budget from state sources, from much higher when I started this gig) and I think people are flailing in fear that we’re gonna lose the rest of that.Report

          • Avatar Brent F says:

            Yes, that sounds like exactly the wrong way to adapt to “disruption.” The focus should be on what you do that adds value and there is a market for. It sounds like you really aren’t in the low-instructor engagement, mass lecture game anyway so its not your part of the market anyway.Report

            • fillyjonk fillyjonk says:

              Add in that the things I listed above are part of the reason I took this gig, and are a big part of why I stay here: the ability to see someone go from a scared freshman who doesn’t quite know what they’re doing to a senior who has med school/grad school acceptance and knows how to write a proper paper is far, far more satisfying than sitting in my office and clicking buttons to “interact” with students I’ve never met.Report

    • Avatar Road Scholar says:


      The Eli Whitney article shows that a lot of people really don’t care about knowing history accurately but see it as series of stories.

      My best friend growing up has always had a fantastic memory. This dude can reel off names, dates, and places like IBM’s Deep Blue. Me, not so much. The specific facts part of history is something I always struggled with. But history is more than just a card catalog full of facts. It’s also understanding the broader forces at play. Is it really essential to know the precise role that Himmler played in the Reich to understand the broader narrative of the Holocaust? Does having an encyclopedic knowledge of of all the names, places, and precise dates actually lead to deeper understanding? Sometimes, for specific events, sure. Most of the time ,probably not.

      Anyway, my friend with the eidetic memory became a lawyer. Hey, so did you! Maybe there’s a connection? Maybe you could be a bit more understanding of those of us that lack that talent?Report

      • Avatar LeeEsq says:

        There was an interesting book from the mid to late 1990s called “Historical Thinking and Other Unnatural Acts.” The thesis of the book is that the ability to think like a historian and treat history as something more than past, present, and future requires a special sort of psychology that a lot of people do not have. While I agree that history is much more than a “card catalog full of facts”, I also think getting the basic facts right is necessary for doing abstract historical argument correctly. When you don’t know that Eli Whitney was white, your going to sound dumb when arguing the cotton gin’s role in American and world history.Report

        • Avatar Richard Hershberger says:

          I also think getting the basic facts right is necessary for doing abstract historical argument correctly.

          Testify, brother! There are a lot of bad baseball history books out there. Baseball history books come in three types by author: sportswriters, amateur enthusiasts, and academics. In all three cases there are good ones and bad ones and mostly mediocre ones. For your truly spectacular crapfests you have to look to the academics. The three linked books are bad in different ways, but they share a lack of concern with getting the basic facts right.

          My sense is that this isn’t a condemnation of academic history in general, but of academic sports history in particular. Two of those three books are by guys whose academic area has nothing to do with baseball. If you are an academic and a baseball fan, there is a natural temptation to combine the two: a fun way to pad your C.V. If you are conscientious and self-aware, you also realize that you would have to learn a whole new area to do it right, and will likely be put off by this realization. If you believe that your Ph.D. makes you omni-competent, then you are off to the races! On the other hand, the third link is to what seems to be a book form of the guy’s doctoral thesis, so this is his area. The research put into it is, to be blunt, half-assed. Yet here he is. My guess is that his thesis committee lacked the background to realize the half-assedness of the research. This is itself cause to pause for thought.Report

        • Avatar PD Shaw says:

          School history textbooks are full of unimportant minutia. Part of the problem, discussed in James Loewen’s book, “Lies My Teacher Told Me,” is that the school boards that approve history textbooks compare the alternatives based upon the inclusion of some favored figure. (They don’t have time to read multiple textbooks) A school board in Vermont might decide which book to assign based upon the extent of the coverage of favorite son Chester Arthur. (Loewen dosn’t have a problem with regionalized history, Arthur might provide a useful entry point to a variety of topics to Vermonters)

          Loewen’s book does not mention that any history books falsely stated that Eli Whitney was black, and I’m pretty sure if any did, he would have. I’m also skeptical about the claim in the link that half of students in America were taught that. (Lies I learned from Facebook)Report

          • Avatar Kim says:

            There’s nothing worthwhile about history that isn’t better covered by economics.
            The wheels of justice may grind slowly, but economics tends to grind up pretty quick if you get enough sand in the gears.Report

  3. Avatar notme says:

    Nothing about that award ceremony last night? Surely there is something to be said about liberals siting around stroking each other an whining about Republicans.Report

  4. Avatar Richard Hershberger says:

    Obsolete teachers: meh.

    The linked article is an example of a time-honored genre. It presents a fantastic new teaching technique that will revolutionize education. The article is followed some time later by another article describing how it isn’t working that great after all. This in turn is followed by an article from the originator of the technique complaining that everyone else is doing it wrong. At that point the cycle is complete and we move on to the next fantastic new teaching technique that will revolutionize education. There are only so many possibilities, so if you wait long enough, things cycle through a relatively short list of revolutionary techniques.

    When I was in fourth grade my school used a revolutionary math program (not in the computer sense of “program”: this was in the early 1970s.) I don’t remember the details, but there was an elaborate filing system of cards with problem sets. We took a skills assessment test at the beginning, to identify where to begin. You would have your problem set for your precise level. When you completed the problems you would move on to the next card. My recollection is that students took turns managing the card file. They may have even graded the work.

    I loved it. Another student, Alfred, and I were the most advanced. As nerdy types, even in fourth grade we were natural allies, so we turned this into a friendly competition, racing through the program. The next year I was in a different school district–an occupational hazard of being a military brat–and to my regret I never saw that program again. In retrospect I realize that it worked great for me because I was highly motivated and because fourth grade math involves (or did then) a lot of long multiplication and division, which is to say it is more about practice than learning new concepts. I have my doubts about how well it worked for less motivated students, or how well it would work for, say, trigonometry.

    Which is to say, I think that actual face-to-face interactions with a real live teacher is still necessary. Computers are certainly part of the mix, and this likely will expand, but plopping a kid in front of a terminal and expecting magic isn’t terribly likely.

    What might change this is if (or when) kids grow up interacting with AIs the same way they interact with real people. Frankly, focusing on how this would affect teachers’ job opportunities would be to bury the lede spectacularly.Report

    • fillyjonk fillyjonk says:

      the math thing sounds like SRA for reading, which I remember.

      I kind of liked SRA, even though I realize now it lacked any discussion between students because we were all on our own levels.

      And yeah, the human-interaction thing is one reason why I am an online-learning skeptic: people are already bad enough at humaning with other humans, we don’t need to isolate them more.

      That’s not to say I don’t think the (perceived) cost-savings and flashiness of online education is going to lead to some universities closing or going 100% online, or greater pressures for faculty to teach that way.

      My dad tells me “Oh, this is just like IETV in the mid 1970s” and I hope he’s right (in that IETV went away) but I’m enough of a pessimist to fear he isn’t.Report

    • Avatar Kim says:

      The main problem with people learning things is motivation. That’s why literacy rates have skyrocketed — we need reading a lot more now for Fun Stuff.

      If you want something that motivates kids to learn, well, we’ve got schools like that (some schools where either you graduate or you die. Stuff like that).Report

    • Avatar PD Shaw says:

      The reading program in the linked article reminds me of my 8th grade “reading lab” from over 30 years ago. There was a room with what it appeared to be large viewmasters through which lines of story would be scrolled at a speed that could be manually adjusted, and afterwards a quiz taken on the screen that would produce a composite score for speed and comprehension. Reading lab was once a week and maybe it excited some kids, but the stories were boring, and I learned that I could crank-up the WPM, focus on the nouns and verbs and get a high composite score on multiple stories in an hour without doing much reading.

      Perhaps this helped more challenged students, I have no idea. My kids attend or attended an “Apple” middle-school where just about everything integrates laptops with the coursework, but the least tech aspect of it is reading. They are all supposed to be carrying a book of their choice at all times, and be prepared to talk about it if stopped in the hall. People learn to be better readers by reading things they like to read.

      All that said, the article is really about people learning to read English as a foreign language, which generally swamps resources of a public school. This is no doubt a good thing, and would be better if the national government paid for all aspects of it.Report

      • fillyjonk fillyjonk says:

        Tachistoscopes! We did that in seventh grade. I hated it because there was that pressure to read and digest the SINGLE line you were offered before it clicked on to the next one. It felt to me like a very unnatural way to read, and it also reminded me of that old Charlie Chaplin movie where he got stuck in giant cogwheels.

        I’m still not sure how it helped anything. I was a good reader – an avid reader – but my comprehension tested more poorly with the stupid tachistoscope things because of the stress they gave me and also because I couldn’t look back to check something like I would with, you know, a real book.

        I wrote about it on my blog once and someone commented that a similar technique was used to train plane-spotters in WWII. I could see THAT being useful because being able to identify quickly based on a glance would be important for knowing “friend or foe” but it seemed pretty stupid as a reading tool.Report

        • Avatar Kim says:

          A friend of mine with dyslexia reads a whole page at once — because he has to. He can’t count on reading any particular word properly, so it’s all context.Report

    • Yep. My mother was given a new curriculum every year and had to attend training programs and then the very next year some other miracle technique would come along. In the summer she and I would assemble these huge collections of kidney beans and paper clips and bowtie pasta for the various math projects, and organizing chapter books for the reading program. The school district had to have spent an unGodly amount of money every year buying the newest curriculum series. And now anyone can pick them up on eBay for a couple bucks and my husband gets huge stacks of old textbooks at the county dump every day for recycling.

      I had the cards too and I also loved them (ah yes, filly, SRA’s, I’ve tried to remember their name in the past and couldn’t, thanks). But unfortunately my teacher insisted that instead of moving at your own pace, you had to start off at the lowest level and do every. single. card before being allowed to progress. There were LOTS of cards to get through and while I got a high score on all of them, it felt like busywork, too easy. I remember finally making it through all the easy cards with one week left of school and then finally I was allowed to go to the next level where I had to work a lot harder and realizing that I’d just wasted the last year doing things I already knew and probably should have been in that box to start with. :/Report

  5. Avatar Richard Hershberger says:

    I am skeptical of the claim that driverless cars will reduce traffic congestion. History has shown that when it comes to road infrastructure, if you build it they will come. Start from a major city. Build a freeway from it out into open desert. Wait ten years, and you will find the freeway full of cars of commuters going to work from the housing developments that sprang up.

    Let us stipulate that driverless cars will be generally practical sooner rather than later, and that they will rapidly achieve such prevalence that they optimize traffic flow. Does this mean that traffic jams are a thing of the past, like disco and Bund rallies? Or does it mean commutes of longer distance (and therefore more cars per mile) because people will tolerate longer commutes, and therefore live yet further away from their jobs?

    If the young ‘un’s preference for higher density living sticks, even once they have kids, that will be a real game changer. Driverless cars will be part of this only to the extent that they encourage the trend.Report

    • Avatar Kolohe says:

      The biggest advantage of everyone being in a driverless car is rational and fair queuing when it comes to a merge. You don’t have the cars racing forward in the lane that’s about to end, getting over at the last moment, holding up the thru lane traffic, which in turn causes more people to skip ahead, more slow down of the thru lane, etc.Report

      • Avatar Kim says:

        Think bigger: The biggest advantage is no more need for red lights.Report

      • Avatar Damon says:

        My state already has a rational merging law. The merging vehicle must yield. This is generally “enforced” by the community of drivers. Is someone going to program a vehicle’s algorithm to account for 50 states various laws on such things?Report

        • Avatar Kim says:

          Absolutely. Right now, of course, the automated cars run on nonpublic roads.Report

        • Avatar Trumwill says:

          My state already has a rational merging law.

          Leaving aside that said state has among the worst drivers in the nation, a universal merging-vehicle-must-yield may not be the most efficient outcome anyway.

          Is someone going to program a vehicle’s algorithm to account for 50 states various laws on such things?

          That’ll be one of the easiest hurdles to jump.Report

        • Avatar Kolohe says:

          And we know that human beings follow traffic laws more carefully than any other set of laws.Report

          • Avatar Damon says:

            True, but the law says I don’t have to yield to the merging car…and if there is an accident, the merging car is at fault. Period. Sorta like getting rear ended. Baring some weird circumstance, the guy who rear ends your car is at fault.

            Damon don’t yield for merging traffic.Report

            • Avatar Kim says:

              You ever crashed somebody doing that? Into concrete filled barrels, say?Report

              • Avatar Damon says:

                I’ve never seen barrels, filled with concrete or not, on, near or along side merge lanes. Most people stop. Of course, sometimes you have the occasional idiot who refuses to slow down and get behind you, insisting to speed up and get in front of you as they quickly run out of room.

                And then there are the fools who are almost past the off ramp and realize they want to use it and get peeved when you don’t yield to them….who could have slowed down from 65 to get behind me. Nah..they just rode with their high beams on and honked at me.Report

              • Avatar Morat20 says:

                Those barrels are filled with water, by and large. Sometimes sand.

                Not concrete.Report

            • Avatar Kolohe says:

              Getting into an accident makes one’s daily commute much longer, even if one isn’t in any way at fault.Report

    • Avatar Kim says:

      Right now, we are at 2.5 cars per driving-age adult. We have reached peak auto.Report

    • Avatar Road Scholar says:

      Or does it mean commutes of longer distance (and therefore more cars per mile) because people will tolerate longer commutes, and therefore live yet further away from their jobs?

      There was a study some years ago on commuting that found that regardless of the mode of travel — car, train, bus, bike, walking, or combinations thereof — on average people preferred a commute of about 30 minutes. Not only did they not like longer commutes, they also didn’t care for shorter ones.

      The authors speculated that the commute served as a kind of transition from a “home” mindframe to a “work” mindframe and vice-versa and thirty minutes was about right for that purpose.Report

      • Avatar Trumwill says:

        I once had a commute under 20 minutes. It was really, really weird. Uncomfortable, in a way.

        I think commute tolerances might grow longer if people don’t have to pay attention to the road. But there are definite limits there. Even when the cars have wifi and all that.Report

        • Avatar Kim says:

          Scranton to Manhattan.Report

          • Avatar Marchmaine says:

            …is a great idea for phase 1 hyperloop.

            I know someone who does Scranton to Dulles… but he’s a pilot and only does it once every 10-14 days… but still a little crazy.Report

        • Avatar Mo says:

          I have a 10 minute commute, my only complaint is that it’s too short to listen to a podcast. That is the entirety of my complaints.Report

        • Avatar Troublesome Frog says:

          I went from 65 minutes each way to working from home and the difference was jarring. The most obvious change was weight loss since I had an extra 2 hours a day and could do things like “exercise” and “cook food” which I’m told are things that are good for human beings to do.

          I’d be willing to go back to commuting maybe a total of 75 minutes a day, but that’s about as far as I’d push it. The quality of life increase from an extra 5-10 hours a week is too much to give up. It’s basically like having every Friday off. Think of the extra stuff you’d get done if you only worked 4 days a week.Report

      • Avatar Richard Hershberger says:

        On the other hand, with true driverless cars–not merely glorified cruise control–this dynamic might change. By true driverless cars, I mean ones where you can spend the drive time in one of the Big Three activities: reading, sleeping, or fornicating. Once these become possible, the commute switches from chore to down time. At the same time, the correlation of real estate prices to proximity to major population centers won’t change. People already will accept a longer commute in exchange for more house. Go another thirty minutes out and you can live in manorial splendor!Report

      • Avatar PD Shaw says:

        I’ve had a 15 minute commute for most of my working life and cannot imagine a situation where I wouldn’t feel a 30 minute commute was worse. There may be other factors, but I just wonder if people average about a 30 minute commute time and feel better about their commute by believing its perfect, just like the size of Donald Trump’s hands.Report

        • Avatar J_A says:

          For a decade I had a 20 minute commute, and I agree it worked great to both focus in the morning (start to think about what I needed to do that day), and to relax and decompress.

          It turned into. 30-45 minutes commute, and anything beyond 30 min it just added stress, going in or going out.Report

        • fillyjonk fillyjonk says:

          I have a ~5 minute commute. The tradeoff I pay for that is that “good shopping” (e.g., not walmart) is an hour’s round trip away.

          Some days I feel like “it would be so nice to live in a town with a decent grocery store.” Other days, I really like being able to run home for lunch when I don’t have afternoon classes.Report

        • Avatar dragonfrog says:

          My commute is about 30 minutes by bicycle. I think a much shorter one would be worse, because riding a bicycle is such a pleasure – My commute could be around half that time if I drove, but then I’d have to spend 1.5 hours to get an hour of cycling into my day.Report

        • Avatar Will Truman says:

          My commutes, ranked from best to worst:
          1. Three miles each way, no interstate, 10 minutes each way. (20 minutes probably would have been better!)
          2. Thirty Miles each way, 80-90% Interstate, 30 minutes to work, 40 minutes back.
          3. Fifty miles, fifty minutes each way, 95% Interstate
          4. Sixty-five miles, fifty-five minutes each way. 95% Interstate
          5. Fifteen miles, forty minutes each way. 10% Interstate

          6. Forty miles, 105 minutes to get to work, 75 minutes to get back. 90% Interstate.Report

        • Avatar Chip Daniels says:

          Mine is a 15 minute walk.
          Compared to the old 90 minute car/train/walk, it’s glorious.Report

      • Avatar veronica d says:

        @road-scholar — My commute is about 30 minutes (except for those days the MBTA goes pear shaped, AKA, days that end in “y”). But anyway, yeah, that’s about right. 30 minutes feels about right.Report

    • Avatar dragonfrog says:

      I think density will be encouraged if something like the true cost of driving is borne by the driver.

      Right now my property and income taxes build and repair roads and highways whether I drive on them or not. The cost isn’t meaningfully built into fuel taxes (here they’re something like 5c on the dollar of infrastructure cost imposed by burning it) and what that in turn does to the cost of goods from distant places etc.

      Anytime I hear someone rhubarb about how “my registration and fuel tax pay for the roads so people shouldn’t be allowed to ride bikes on the road unless they pay a tax too,” I chuckle to think how bitterly they would complain if they became right tomorrow.

      To the extent driverless cars become collectively owned cars and the capital cost of buying one becomes borne in proportion to how much one uses it, that will likely encourage minimizing driving.Report

      • Avatar aaron david says:

        I think density will be encouraged healthcare saving will be achieved if something like the true cost of driving medical care is borne by the driver patient.

        Careful what you wish for.Report

        • Avatar dragonfrog says:

          I’m not wishing, I’m predicting.

          I mean, I do think that self-driving cars would make car-share co-ops more powerful, and that if it works out that way that would be a good thing in that it would make more options available to more people, including
          – living in more dense walkable neighbourhoods, retaining a lot of the advantages of car ownership without the costs of car storage.
          – making retail areas more dense with stores and less about parking lots, improving walkability without losing customers who reach the store by driving.

          Anyway healthcare is different enough to be almost entirely irrelevant in this case, in that (1) how much we drive, while not entirely self-determined, is much more under our control than whether we get cancer or are bitten by a malarial mosquito, and (2) the social benefit in medical care is more often in the same outcome as the individual benefit (people get treated earlier, have fewer chronic conditions from delayed treatment, spend less time infectious), whereas with driving the social and individual benefits are often out of sync.Report

        • Avatar Stillwater says:

          That’s an interesting analogy, Aaron. I think it stretches the limits of credibility, myself, since the hidden costs TFrog mentions relate to infrastructure. What’s the medical care analogue of infrastructure? Things like switching from paper to electronic record keeping? I think that’s already happening, yes? Hospitals? Not sure what you mean here.

          Going the other way the analogy is strained as well, seems to me, since the analogue of medical being borne by the patient would be that vehicle costs are born by the owner/driver. Which already the case, too, seems to me.

          Healthcare is its own unique problem from a market and public policy perspective, seems to me.Report

          • Avatar aaron david says:

            My point wasn’t that commuting and medical care are on the same level, its that the logic that making costs visible is always a good thing. Many people here feel that HC is a nonmarket type good, but at the same time rail against diffused costs in transportation. My point was that the same logic can be used in other situations by simple substitution.Report

      • Avatar dragonfrog says:

        To clarify – I’m just thinking about the capital cost of the car, not of the infrastructure it drives on. I think it would be good to have more, even if not all, of the cost of the infrastructure directly tied to its use. But that’s not what I’m predicting above.

        Of course today a lot of the capital costs of car ownership are actually marginal – depreciation, maintenance, etc. But they’re not experienced as marginal – few people actually put money into a kitty for replacement tires, fan belts, oil changes, suspension, with every tank of gas they buy. Everything is good and then “Oh no, what’s that awful squealing sound?”

        And I don’t know of any auto insurance companies that insure on the basis of distance driven – a car that drives 50 km a week costs as much to insure as one that drives 3000 km a week, even though the companies pay out more for accidents from cars being driven a lot than cars that barely get used.

        A co-op that charges a low flat membership fee and a covers most of its costs out of per-distance fees would make the marginal costs, truly felt as marginal costs, probably changing people’s economic decisions – even though little has really changed other than the realism with which cost breakdowns are experienced.Report

        • Avatar dragonfrog says:

          (OK, turns out I didn’t know of per-distance auto insurance mostly because I haven’t been shopping for auto insurance lately. It’s become available around here in the last year or two – but even so, 70% or more of the insurance rate is fixed even if you only drive around the block once a year).Report

        • Avatar Michael Cain says:

          And I don’t know of any auto insurance companies that insure on the basis of distance driven – a car that drives 50 km a week costs as much to insure as one that drives 3000 km a week, even though the companies pay out more for accidents from cars being driven a lot than cars that barely get used.

          State Farm used to have mileage bands. They were pretty crude. IIRC, I got a discount when I retired and my mileage dropped below 7,500 miles/year. In some states, State Farm now has a program that uses your phone and an in-car dongle to monitor some aspects of your driving, including miles, and provides discounts for low mileage (among other things). Progressive also has a dongle that goes in the car, and they offer discounts for low miles.Report

      • Avatar DensityDuck says:

        “I think density will be encouraged if something like the true cost of driving is borne by the driver.”

        Hah! People have been trying to find the Holy Grail of “true cost borne by the user” since forever.

        “Right now my property and income taxes build and repair roads and highways whether I drive on them or not. ”

        Really? Got some numbers for that? According the numbers I’m seeing that’s not the case. Maybe it’s different where you are, but I’d be surprised it if were hugely so.Report

        • Avatar dragonfrog says:

          To your first point:

          It’s true the infrastructure costs won’t be magically allocated to drivers on their own. It might happen, but not without additional measures. I’m talking about only a “weak” version of that:

          If the way self-driving cars pan out is largely as co-op cars that spend close to half their time driving, rather than as individually owned ones that spend 99% of their time parked, then I predict the likely billing model will be almost entirely per distance traveled – even the insurance would be mostly rolled into that. They’ll be cabs without cabbies, basically.

          At that point, you see the full cost of driving rolled into the distance driven, and can decide based on full information whether to make the drive. And more often than under the current system, where many of the marginal costs aren’t experienced as such, will probably decide not to make the drive.

          Right now, it’s easy to decide to drive to Costco to save $22 on groceries. Even if I log my fuel purchases against odometer readings so as to accurately deduct fuel costs from drives I think will save me money (which lots of people don’t), I might figure I only burn $6 of gas, so I’m still $16 ahead on the trip compared to going to the more pricey store that’s in walking distance.

          But every kilometre I drive I’m also bringing closer the day when I need an oil change, and new tires, and a new fan belt, and ultimately a new car – but those costs aren’t usually calculated and experienced as marginal per-distance costs. If I did that, I might well find I’m better off economically leaving the car parked and buying the more expensive groceries.

          To you second point:

          The PDF you linked to doesn’t, to me, make obvious what funding sources produce what percentage of the money.

          For example, page 4 notes there are user fees, property related charges, and “subsidies”, and gives a breakdown – not among those three categories at all, but among federal, state, and municipal sources.

          I did find scattered things about individual municipalities, for example in Halifax, NS property taxes pay for about 77% of the cost of streets, motorist-specific taxes for 23%.Report

          • Avatar DensityDuck says:

            “For example, page 4 notes there are user fees, property related charges, and “subsidies”, and gives a breakdown – not among those three categories at all, but among federal, state, and municipal sources.”

            Try reading the whole thing, particularly page 12.

            If you want to get grumpy about “your taxes” going to build roads that other people drive on, then sure, you can find the occasional quarter-percent of sales tax that goes into the transportation fund (which is then spread out between all transportation projects, including railways and mass transit). But the exceedingly large majority of transportation funds come from user fees.

            “Even if I log my fuel purchases against odometer readings so as to accurately deduct fuel costs from drives I think will save me money (which lots of people don’t), I might figure I only burn $6 of gas, so I’m still $16 ahead on the trip compared to going to the more pricey store that’s in walking distance.”

            The standard deduction for mileage is fifty-four cents per mile, which is considered to be an amortization of all fuel, maintenance, and depreciation costs across the entire year. (You can track actual costs and deduct those instead, but then you don’t get any depreciation writeoffs until you actually sell the vehicle.) And this is for business use, which assumes a high duty cycle (regular use for a large portion of every day); personal use, which is point-to-point and generally only two or three trips per day, would likely have a lower rate if actually determined.

            Meaning, if CostCo is less than twenty miles away it’s still cheaper to drive there than it is to walk to the expensive store.Report

            • Avatar dragonfrog says:

              I’m not saying you’re wrong about how California funds roads. I can’t tell from the document, is all – it’s page after page of boxes and arrows without dollar figures from which to derive percentages.

              Page 12 is specific to state & federal highways, so maybe not the most relevant (I don’t know a lot of people who do their intercity highway travel by bicycle, nor do I know a lot of people who get all angry about others riding bicycles on the highway. It’s city streets where that’s the case).

              You misunderstand though – I don’t get grumpy about “my taxes” building facilities I don’t use. My taxes pay for parks I’ve never played in, streets 99.99% of which I’ll never see, pools I’ll never swim in, libraries I’ll never borrow from, universities I’ll never attend, ob/gyn clinics for which I lack the relevant tackle. I am fine with all that.

              I object to the argument that I shouldn’t be allowed to ride my bike on the streets, or that infrastructure shouldn’t be build specifically to support safe cycling, because “drivers pay for the roads” – in part (though not only) because it’s not true. Not where I live it isn’t.

              Drivers cause darn near 100% of the wear and tear on municipal streets, but everyone pays for them out of property taxes. I’m not saying I don’t want to stop paying for roads, I’m saying:

              1) I want drivers to understand I’m subsidizing them, not the other way around – just the same way I’m being subsidized by every non-swimmer in the city every time I go to a municipal swimming pool, by non-museum-goers every time I go to a municipal museum, etc.

              2) I think drivers would complain even more bitterly if tomorrow the fuel tax and vehicle registration costs were changed so that drivers really did pay the full cost of the demand they put on the infrastructure.

              I don’t have the exact numbers for Edmonton, AB, but it’s not going to be far off of the 77% property-tax figure I found for Halifax.

              Per this it sounds like US drivers cover a bit more of the cost of US streets and roads than Canadian ones do – but they’re still heavily subsidized by non-drivers.Report

              • Avatar dragonfrog says:

                see also

                “As one would expect, there are genuine differences in payments versus expenditures from region to region in the US and from country to country, independent of differences in accounting frameworks. Cameron’s (1994) accounting for Southern California in 1991 suggests that tax and fee receipts related to motor-vehicle use easily exceed public-sector expenditures (when bus and rail receipts and expenditures are excluded from both sides of the ledger), but Komanoff and Sikowitz’s (1995) accounting for New Jersey indicates that there receipts are only 77% off expenditures. Hanson’s (1992) accounting for Wisconsin in the early 1980s indicates that state and federal user fees are only about half of roadway expenditures by all levels of government, and Ryan and Thomas Stinson’s (2002) accounting for the seven-county Twin Cities metropolitan area of Minnesota in 1996 shows that user taxes and fees (which exclude general property taxes, general state aid, and special assessments) provide about 60% of all revenue used for highways.”

                It looks like (1) you’re probably right about California, and (2) California is very much an outlier in this.Report

    • Avatar Oscar Gordon says:

      I’ve explained this before, but if you talk to traffic engineers, they’ll tell you that, in high volume traffic, the upstream effect of one irrational driving decision can not be understated.

      That guy who wildly swings across 4 lanes of traffic to make his exit, or get to the far left lane, and causes people to slam on their brakes, he causes a ripple braking effect that can propagate upstream for, when conditions are right, literally miles. Basically, anytime a person slams on their brakes in high volume traffic, that action causes everyone upstream to brake hard as well. Driverless cars, especially if they can talk to each other, even a little bit, will pretty much eliminate that.Report

      • Avatar Richard Hershberger says:

        Sure, I’ll stipulate to that. My point is that historically, if you build a new freeway there is a short period of a couple years of commuting nirvana, until traffic volume expands and you are back where you were before. Why should we believe that our driverless future will be any different? It will increase the capacity of any given road, but there still is a point where congestion slows everything down.Report

        • Avatar Oscar Gordon says:

          This will be true for every mode of transit we develop. Busses get full, trains get full, there is only so much sky for your flying car to operate in, etc.

          The biggest benefit will be removing that actions of drivers that are irrational/ego motivated.Report

          • Avatar Richard Hershberger says:

            Fair enough. But this means that any happy talk about congestion being a thing of the past should have an asterisk to a footnote qualifying that this happy state will be temporary.Report

          • Avatar switters says:

            I’m guessing that everything else being equal, dirverless cars will reduce congestion. But all that means in reality is, that the infrastructure we currently have will support more vehicles than it currently does, so the point in time at which enough busineses or people demand upgrades to infrastructure to get the politicians attention will simply be delayed x amount of time. So well get a little period of better than usual traffic, followed by business as usual.Report

            • Avatar Michael Cain says:

              I also think that congestion will be reduced, for various reasons. Vehicles purpose-built for single person commuting will be smaller, so more vehicles per lane-mile. Following distances will be decreased, so even more vehicles per lane-mile. Assuming at least rudimentary communication between vehicles, on-ramp and off-ramp behavior will be improved on average*. Speed irregularities will be reduced in unexpected places: my last commute, there were about eight weeks per year when the interstate went up an incline directly into the rising sun; everyone slowed down some, but when someone slowed down too much you got the kind of traffic jam that someone described elsewhere.

              To be honest, though, based on my experience with one place that implemented it, the single best thing to do to improve congestion is keep the big trucks off the main urban highways at rush hour.

              * There’s a research topic for someone. What are the rules for emergent behavior** that, if every vehicle followed them, allow multiple lanes of traffic to work together to create a place for a merging vehicle?

              ** Eg, birds flocking, fish schooling, and herds stampeding all follow the same three simple rules to avoid collisions, with different values for the small number of coefficients.Report

              • Avatar veronicad says:

                Well, on the whole it is a “very NP-hard” problem. Add to that we need a resilient solution, cuz even if you find a super tip-top optimum, your solution must still do well if a tree branch falls across some major road — cuz tree branches will fall, or something. It’s always something.

                Flocks of birds are a different problem, since they mean to move in unison. For traffic, you have competing needs. For example, busses should be prioritized, but how much, versus the person who payed the “optimal service” price? — and here comes politics. But it will invariably require backing off some good solutions to leave wiggle room in for unexpected events, but how much backing off? Who decides?

                Anyway, I can think of “solutions” to all these problems, but in my experience, simple solutions to big problems usually don’t work. We’d need learning algorithms, and this problem is really actually hard.

                I suspect this won’t work half as well as its proponent.

                We should still probably do it. In the long run, it will work better than what we have now, but only by degrees.Report

        • Avatar Troublesome Frog says:

          Even if the total amount of congestion sadness each drive encounters remains constant because we have a fixed tolerance for that sadness, the total number of people we’re able to get in and out of any given city should increase dramatically in the process, which is good news overall.

          I think that like everything else, there’s some elasticity/incidence thing going on. Of the total benefit, some of it will be enjoyed in the form of reduced commute times and some of it will be enjoyed in the form of increased living/commuting flexibility (which increases commute times to negate some of the benefit). Overall, it still sounds like a win, and I’m inclined to believe that there are actually enormous efficiency gains to be had and split between volume and speed benefits. I live and drive around a pretty horrible part of the world for traffic, and even with the number of cars battling for space, most of the real problems are caused by human stupidity.Report

    • Avatar North says:

      What strikes me about driverless cars is less congestion questions than questions of ownership. Paying for private transportation is pricey right now and a big driver of that cost is labor, not capital. If driverless cars are perfected the price of grabbing a cab from point A to point B in an urban environment will plummet. That makes not owning a car and merely paying for rides a much more economical option. That in turn could speak enormously to questions of parking space in downtown and yeah maybe even congestion though the fact that instead of being parked the autotaxi is returning to the street for more passengers strikes me as making congestion a wash.Report

      • Avatar Richard Hershberger says:

        I’ve never figured out how this model of commuting is supposed to work. Most people commute to and from work at roughly the same time. This is why we have rush hour. It is easy to say people should space out when they commute, but there often are structural reasons for the 9-to-5 schedule that aren’t going to simply disappear. If everyone is going to commute in a vehicle they don’t own, this will only work if there are enough vehicles to meet the rush hour demand. Regulated utilities built up to meet peak demands, but I don’t see Uber, et al., doing this. Rather, I see them instituting peak pricing, which they will probably call reduced off-peak pricing, and lecturing us about modifying our work schedules, which is all very lovely for those with jobs where they can do this, but not terribly helpful to the rest of us.

        Then I get to the office and the car I rode in on goes its merry way. Where? We have this vast over-capacity for the working-hours demand, so presumably it will go park somewhere. Where? Will Uber, et al., have their own parking facilities? OK. So now we have an extended rush hour, as all these now-empty cars go to the parking lot in the late morning and then return to the offices in the late afternoon. Come to think of it, they’ll probably need parking facilities out in the ‘burbs, too. Otherwise there will be a lot of empty cars on the road after dropping off commuters at their homes.Report

        • Avatar North says:

          Sure, but now you’re talking about stuff that computer nerds and algorithm builders get all gooey and happy talking about. Also you’re conceptually leaping from our current commuting system to a potential end state without considering the midway steps.

          To your points, yeah probably the auto-taxi’s would surge around rush hour with a great flow of cars moving to the residential regions from the job centers then back again. I don’t mean to suggest complete elimination of self driving; just a reduction. Consider how dramatic the impact would be if every automated taxi eliminated a half dozen driven commutes. All that unused parking space could probably easily allow room for some kind of central garage.
          Some of those cars would probably dock themselves but a lot would probably disperse, and cruise for passengers. It’d be a really different world, especially in cities, and easier than ever to not own a car. That would be a really big change.Report

          • Avatar PD Shaw says:

            Do computer nerds and algorithm builders have any interest in making elevated commuter trains driverless? Because it seems like a mode of transportation on a defined route with limited risks to pedestrians and other means of commuting would be comparatively very easy.Report

            • Avatar Kolohe says:

              Some of them are. Unions and intertia prevent them from being more prevalent.Report

            • Avatar veronicad says:

              Vancouver’s SkyTrain is driverless, I think? (Maybe I’m wrong, but someone told me it was.) Anyway, it seemed to run pretty well. On the other hand, driverless cars/busses will revolutionize traffic flow in a way that driverless trains doesn’t. This would be a society-changing kind of thing, perhaps on par with cellphones (“hand computers”) or the automobile itself.


              Driverless trains, from a society viewpoint, don’t really differ much from trains with drivers.

              After all, I spent a week in Vancouver last month. I rode the train around. I actually don’t know if they had drivers or not. I paid my fare, rode the train.

              On the other hand, the fact that Vancouver didn’t have Uber or Lyft was damn annoying.Report

            • Avatar North says:

              What Veronica said. Hell, pondering it there probably wouldn’t even be garages. People would probably buy cars as investments and have the things working the streets for them via some kind of uber service whenever they weren’t using them themselves. Thus the garage would be peoples garages and driveways.Report

              • Avatar Michael Cain says:

                People would probably buy cars as investments and have the things working the streets for them via some kind of uber service whenever they weren’t using them themselves.

                Anyone plotting out the future of self-driving cars in the US that isn’t including “keep the aging Boomers in their suburban houses as long as possible” is missing the boat. There are millions of vehicles that meet their owners’ needs four mornings and two evenings per week, and available to generate a few thousand dollars of extra income per year.Report

              • Avatar Kim says:

                Um. Right. How many millions of refugees do we need to have before we don’t want the boomers in those Nice Large Suburban Homes anymore?

                Compare with the population of Miami.Report

    • Avatar DensityDuck says:

      “I am skeptical of the claim that driverless cars will reduce traffic congestion. ”

      Of course it won’t. What it will reduce is how much we care about it.

      If my car has an autopilot then my “commute” is only as long as it takes to drive to where the autopilot can take over, and after that I’m at the office.Report

  6. Avatar Kolohe says:

    Denzel Washington did portray Eli once, so the confusion is understandable.Report

  7. Avatar Damon says:

    Robots: The Forbes article is pretty much on point. At least until the machines develop a modicum of adaptability/intelligence, a human have an advantage in some jobs. For painting cars, not so much. All the former paint guys at the plant I used to work in, after robots were installed, ended up doing “masking” jobs. Robots can’t put paper and tape on cars to block off non painted areas…yet….

    The ZME article seems a bit of a stretch for some jobs. We are seeing waitress and such jobs being effected with the terminal on the table, but I for one, refuse to self check my ass at the grocery store-mainly because I refuse to conform to how they want me to check out. Scan item, place in bag. What if I want to scan all my items and place them then…you group certain items? BUZZZ not allowed. F that.

    Xerox: We’ve come a long way. One of the employees in my office didn’t even know what a 5 1/4 inch floppy was, and apparently, there’s only 1 dedicated fax machine in the building….

    Driverless Cars: A large group of people could telework. I sure as hell could. Why is this rarely discussed? And does anyone have an issue with using tax payer money to build public roads and then having congestion pricing applied to the road? I sure as hell do.Report

    • Avatar DensityDuck says:

      ” Robots can’t put paper and tape on cars to block off non painted areas…yet….”

      The solution, of course, is to not have non-painted areas. It even becomes an upgrade; the base model has matte-grey plastic trim, the “appearance package” includes “body-color trim panels” (which are just the plastic ones with paint on them).

      ” What if I want to scan all my items and place them then…you group certain items? BUZZZ not allowed.”

      I’m not sure what you’re talking about because I have never encountered a self-checkout kiosk that didn’t permit you to “group” items.Report

      • Avatar Damon says:

        The masked areas were already painted color that mimicked “tinted glass” and had to be covered up before the car color was applied. This was back in the 90s.

        The checkout machine doesn’t like it when you scan the item and put it on top of the bag carousel. It wants you to scan it and put it in the bag. The F? I got meat and eggs at two ends of the conveyor and I want to group them both in one bag and put my ice cream in another and the produce all in one bag. Machine no likey.Report

        • Avatar DensityDuck says:

          “The checkout machine doesn’t like it when you scan the item and put it on top of the bag carousel. It wants you to scan it and put it in the bag. ”

          Huh. None of the checkout machines I’ve encountered require that something go in the bag.

          I’ve seen setups where the scale had a bag rack built into it that took up a lot of space, but you could still put things directly on the scale.

          I suppose that someone might build a self-checkout kiosk where the scale is attached only to the bag rack, but that’s only speculative–I haven’t seen that anywhere.

          “I got meat and eggs at two ends of the conveyor”

          Or maybe you could put them on the same end of the conveyor, performing your sorting prior to loading the belt rather than post-scan.Report

          • Avatar Damon says:

            In the self checkout units of my grocery store, the scale is directly below the “user interface screen”. Next to that unit is a circular bag carousel. I don’t know if the system doesn’t like stuff being put on top of the carousel or only likes stuff put in bags, but when I take stuff from my basket to the carousel and put it on top of the carousel, it gets annoyed.

            As for grouping my items in my cart. I’ll shop my way, you shop yours. The self serve is allegedly there to speed my exist from the store. If they aren’t doing that, they are failing. They can figure it it. Frankly, this is why I don’t use the self checkout except for less than 20 items and even then rarely.Report

        • Avatar Troublesome Frog says:

          The self checkout machine at our local Safeway has bugs in it that are such classic beginner coding bugs I could probably go directly to them in the source code. For a while, my wife and I were trying to figure out the most efficient way to perform our transaction to get through without having it freak out and hang up the process. My wife discovered an order of operations that is not only very speedy but also leaves the machine in a broken state for 30 seconds after the transaction is over. It babbles to itself, goes into an error state, then returns to the start screen. Pretty cool.

          It varies from retailer to retailer, but it’s clear that most of those were definitely lowest bidder projects.Report

  8. Avatar J_A says:

    “But… if it’s robots either way, why make stuff in China to begin with? (Raw minerals, I guess…)” (I forgot to block quote hehe)

    There’s two things of note here

    It’s been decades since China stopped being a cheap labor location. Manufacturing that relies on and requires cheap labor (difficult to automate) like garment making moved years ago to cheaper places like Vietnam or the Philippines, and moved again (when these places became too expensive) to “cheaper” locales like Myanmar and Bangladesh (*). If China manufacturers want to remain competitive they have to automate too.

    Why stay in China? Well, for a start, a lot of high end manufacturing has moved back from China to the USA already. But don’t forget that China is actually very far from here, but very close to the largest markets in the world (markets that were negligible way back then). So, just like it is economic to move manufacturing back to the USA because of transportation costs, it makes sense to serve Asia markets from China (plus, with the new rail links, Europe markets are easier to access also).

    (*) plus the capital cost (equipment, etc,) of a garment factory is very cheap, and easy to move around to a new locale.Report

  9. Avatar Mo says:

    Another reason to stay in China, much, much looser environmental standards.Report

  10. Avatar Will Truman says:

    Well this is frustrating. Right after this gets published, a piece on automating the judiciary.Report

    • Avatar Chip Daniels says:

      Which is kinda my take on how this wave of automation is different.
      Its not the physical jobs that are being automated, its the mental ones.

      Anything that requires logic or analysis is at risk of being made into an algorithm. And a shockingly high percentage of what most professionals do, is logic and analysis.

      I think it would be supremely ironic if the only people whose jobs are not automated are poets and theater majors.Report

    • Avatar Burt Likko says:

      I must swallow my reflexive recoil from such a thing and concede that this may be a useful tool to judges contemplating bail, sentencing, early release, and expungement decisions. With that said, it is a tool, not a comprehensive solution. The tool will only be as good as the algorithms embedded within it and the information plugged into it; those algorithms and inputs may well contain all sorts of biases that are inherent in society and thus may amplify rather than reduce those biases. Besides, real life people in real life situations are simply more complicated than a computer program could possibly be.

      So I might not be bothered if the judge hearing my case looked at the results of this tool and considered it as one of many factors. But I would still want my case heard by a human judge and a human judge to make the final decision.Report

      • Avatar Kim says:

        When the AIs are better at contract law than you are… I worry a lot more about that than them being judges.

        It’s perfectly legal, AND you’ll fail to notice it, and what is the judge going to do then?Report

        • Avatar Francis says:

          Well, there really is no reason to have contracts drafted by hand any more. An expert system should be able to ask a series of questions and crank out the document.

          But it’s going to be quite a while before the State Bar of California (for example) shifts the responsibility for mistakes from the attorney to the programmer.

          ps: I’d love to learn more about legal AIs. Any links you can share?Report

          • Avatar Burt Likko says:

            It would have to be a sophisticated system, but yes, I generally agree with this proposition. The rudiments of it, at least for the sales of physical goods, can be found in Article 2 the Uniform Commercial Code.Report

        • Avatar DensityDuck says:

          You keep talking about how AIs can write contracts that don’t have loopholes.

          Here’s the thing: No contract, as written, has loopholes. Loopholes aren’t found, they’re made.Report

          • Avatar Kim says:

            Will you listen? The AI can write contracts with loopholes that you cant spot.

            To take a non-legal example:
            $35 for 50 DVD spindles. Looks like a great deal, doesn’t it? (this was done at the time when dvds were routinely $50 for fifty). Not so much a good deal when the guy gets … 0 DVDs. Plenty of spindles though!

            Perfectly legal lexical trickery. (and yes, done by an AI)Report

            • Avatar DensityDuck says:

              There’s…not actually anything special about the AI there? It’s just another riff on selling an “Xbox Box”, or a photograph of an iPhone.Report

              • Avatar Kim says:

                Nothing special, sure. Other than an artificial intelligence came up with a plan of action and executed it — and made money in the process.

                But that was years ago. It’s gotten better since then.Report

              • Avatar Chip Daniels says:

                The 3 Laws of Robotics.

                No loopholes there!Report

              • Avatar DensityDuck says:

                I prefer Clarke; a sufficiently-advanced technology has an ability to create loopholes that is indistinguishable from magic!

                The best way is to invent a superset and declare yourself a member of it, with the contract as a subordinate set that only applies in specific circumstances. Failing that, argue over the specific definitions of the terms–like, what exactly the meaning of “state” is in this particular instance.Report

            • Avatar Francis says:

              To be blunt, (a) I don’t believe you and (b) it’s highly likely that the seller would be held to perform anyway. Ambiguity is construed against the drafter and courts are very reluctant to enforce yellow-dog contracts that are drafted to deceive.Report

              • Avatar DensityDuck says:

                Yes, it seems more like Kim’s example of an AI-written loophole requires an AI judge for it to work.

                I mean, I guess that if the AI were intentionally attempting to deceive and hit on the strategy of using deceptive language, that would be cool. I think it’s about as likely that the AI found the line item listed as “DVD Spindle” and then someone who desperately wanted the AI to have created a loophole said “a HA, that means the PLASTIC THING THE DVD COMES ON, not THE ACTUAL DVDs, EAT IT MEATBAG!”Report

              • Avatar Kim says:

                That assumes that things go to court. What’s FAR more likely to happen is that the seller is going to say, “Send it back, and we’ll refund your money” — except the shipping for that many spindles makes the transaction not worth it.

                I suppose you could take this to court, but the seller is likely to say: “It said 50 DVD Spindles. That’s what we provided. It wasn’t even priced at a reasonable pricepoint for you to think it was 50dvds on a single spindle.”Report

              • Avatar Francis says:

                The point, dear Kim, is that I very much doubt the existence of an AI that has a whole language processor so sophisticated that it can deliberately (knowingly? I lack the necessary adverb) generate contracts with ambiguous sales terms.

                And while there are plenty of bad, mediocre, lazy and sloppy lawyers out there, not all of us are. Get back to me when an AI (not its lawyer/supervisor) can slip an intentionally ambiguous term into a multi-million M&A agreement.Report

              • Avatar Kim says:

                The best loopholes are the ones that the other side thinks are points they are winning.
                Example: US Military asked for a “combat-realistic” video game — as realistic as possible. Video Game Designer delivered… something that was totally realistic — and because of that, unfun and not terribly useful for Military Recruitment. (Yes, I do know someone who trolls the US Military for fun and profit).Report

              • Avatar Francis says:

                Dear Kim: Above you wrote: “The AI can write contracts with loopholes that you cant spot.”

                I challenge you. Burt has my email address. Please send me a copy of an AI-written contract with loopholes. You’ll also need to provide me some assurance that the contract was written by the AI and not some dopey lawyer.

                Presumably after I’m done with my review you can introduce me to the lead programmer.Report

              • Avatar Kim says:

                There are multiple reasons why this won’t be happening.
                (If that means that you conclude that I’m a liar, so be it. It’s probably safer that way — you’re not supposed to believe me.)

                For one thing, you aren’t offering nearly enough money. ;-PReport

              • Avatar Francis says:

                I don’t think you’re a liar.

                I think you’re a fabulist.Report

      • Avatar Chip Daniels says:

        A lot of discussions about automation assume that the human will be entirely removed from the loop.
        But from what I’ve seen, the human is kept, just in a greatly diminished role.
        Sort of like how cashiers went from being a semi-skilled job to a completely entry level task.

        I can envision where doctors, lawyers, architects and engineers face a trend of de-skilling, where the AI takes on more and more of the logic and analytics, leaving the human to make the final decision from a limited array of menu choices offered.

        I see this as not only posing a macro economic challenge of how to sustain jobs, but also a cultural challenge of the sexes.

        If we accept that men tend to value and excel at skills of physical strength and quantifiable analytics (Steelworker! Engineer! Argh argh argh!) and women tend to value and excel at skills of personal interaction (Nurse. Teacher. Therapist), then we can see that AI poses a direct threat to men’s economic dominance.

        For example, the male doctor will see his job split into the analytics and the human interface, where a friendly non-threatening female will interact with the patient, following the menu choices and suggestions from the computer to provide a diagnoses and suggested course of treatment.Report

      • Avatar PD Shaw says:

        This passage in the article does not inspire: “The real-life judges made decisions quite different from what the algorithm would have recommended. In theory, this could happen because judges have important information that the algorithm doesn’t—but in reality the algorithm just works better, especially when it comes to identifying the riskiest cases.”Report

      • Avatar El Muneco says:

        To tie in to another thread, I’ve seen almost this exact same comment a number of times before, only it was about automating the strike zone.
        Including the weight of concern and the importance of keeping humans in the loop…Report

  11. Avatar DensityDuck says:

    I am really not sure what to make of that article about the automated traffic stops. It seems to zing back and forth from smug allusions to Big Data to sneering ironic sarcasm about police officers’ behavior.

    ” of the 85 people the Guardian claims were killed by cops in December 2016, only six died as the result of a traffic stop, implying fewer than a hundred a year. ”

    Well. That’s a hundred people who probably wouldn’t be dead right now. Lindenberg seems pretty content to chalk them up as acceptable casualties in the big Game Of Life. I’m all for a realistic assessment of risk but maybe telepresence for a traffic stop is not so huge a cost as to make it uneconomical to avoid a hundred dead people (and any amount of rioting and general societal fear of police).

    “The lack of data means that while it’s possible a traffic-stop robot would be an improvement over the way police currently handle stops, it could just as easily make things worse.”

    People have yet to be killed by looking at a TV screen, except in that one Cronenberg flick where James Woods turns into a VCR.

    “the robot has a few shortcomings compared to humans. For example, it doesn’t have a nose, meaning any cop who used it would be unable to smell marijuana after a stop for a broken brake light.”

    Ah yes, here’s the real core of the column, an anti-cop screed. One gets the sense that this guy likes the idea of cops getting shot, and if a few black dudes get wastes, well, they were probably guilty of something.Report

  12. Avatar Joe Sal says:

    I have to eat crow for a little while. My predictions of a y-axis shift away from right wing authoritarianism has appeared to be the exact opposite narrative than what is occuring for now. Trump is drawing some of the camp away. Kokesh published this a little over a week ago, that roughly parallels a few thing I see happening:

    This could mean we are not de-escalating authoritarianism soon.Report

  13. Avatar Chip Daniels says:

    I know notme posted this in today’s post, but it belongs better here.

    Wendy’s Installs Automated Kiosks

    So the working class guy who saw his manufacturing job go to Mexico takes a job at Wendy’s only to see that job automated.
    Maybe he can become a truck driver?
    Or maybe get real ambitious and become a lawyer?
    And watch those jobs vaporize as well?

    Winding back to Jaybird’s query to me yesterday-
    What is society’s moral obligation to the steadily growing army of un and under employed people?Report