Christopher Carr

Christopher Carr does stuff and writes about stuff.

Related Post Roulette

36 Responses

  1. Will Truman says:

    Re: Driverless cars.

    Trains aren’t driverless cars. A driverless car will be able to take me in nigh-infinite number of directions. Even a robust rail system will be more limited. The beauty of driverless cars is that it can work on our existing network with what I believe to be minimal alterations.

    There would be some institutional opposition, because unmanned driverless cars would put sectors of the economy out of business. Taxis, namely. I don’t think they have the pull to stop it, though. The biggest issue is going to be liability. I would expect juries to be even less sympathetic towards Toyota than they are towards a guy that got into an accident.Report

    • Kim in reply to Will Truman says:

      our existing network is a money sink, and is falling apart, and contributes to the deficit unduly. our existing network is inefficient and thus irritating.Report

      • Matty in reply to Kim says:

        Marty McFly travelled to 2015 and found all the cars where flying. In two years we won’t need roads.Report

      • Will Truman in reply to Kim says:

        All you have to do now is convince people to live efficiently and they do not actually like this thing which they have.Report

        • Kim in reply to Will Truman says:

          We have an ice cream flavor around here: Pittsburgh PotholeReport

          • Jaybird in reply to Kim says:

            You raise an interesting point. The roads in the nice part of the country are orders of magnitude better than the roads in the more corrupt parts.

            So people from (these states) will see driverless cars as the next step in evolution but people from (those states) will wonder if they’ll have to change software if they commute from NY to CT (or vice-versa).Report

            • Kim in reply to Jaybird says:

              There is this thing called the freeze-thaw cycle.

              Also, the other reason our roads suck is not because of corruption, but poor contracting rules. Nobody’s actually getting kickbacks, and I don’t even think the construction companies are overpaid. They’re just poorly paid to do poor work that needs to be redone.Report

          • trumwill mobile in reply to Kim says:

            In a city I used to live, they named a pizza after its busiest highway because it had “bumper to bumper toppings.”

            The interstate that connected where I grew up to the city was called the 401k because construction workers spent their entire careers “repairing” it front to back to front again.Report

    • James Hanley in reply to Will Truman says:

      Also, re: driverless cars. They won’t depend on any infrastructure beyond the current road system, nor will they be dependent on a common software. One of their great advantages is that they are network independent, so you can benefit from buying one–primarily in terms of increased safet and usable time (you can take notes on that biology lecture)–even if nobody else in your area has one.

      But most important of all, it’s a Christopher Carr post–hooray for that.Report

      • Mike Schilling in reply to James Hanley says:

        One of the selling points of a driverless car will be that it uses real-time traffic data to pick optimal routes. When there are enough of them, this will lead to horrific jams as they all pick the same one.Report

        • BlaiseP in reply to Mike Schilling says:

          Not necessarily. One thing we know about hive mind strategies — okay, let me take one step back here. An optimal routing strategy encompasses time of day. When I was back from the Army, I drove a Checker cab in Chicago. The run from 1800 North Wells to O’Hare at 0500 is very different from the same run at 0800. I hated the run, even though the money was good. But I evolved a strategy of going up Elston to Irving Park.

          It’s not difficult to impose hurdles and no-go points on a routing scheme in real time. Truckers do it all the time with bridge heights. Taking it one step farther, if enough driverless cars are taking a particular route, they’ll be providing feedback on optimality: I’ve been on this route and I’ve been moving at 2 miles per hour for ten minutes. That already creates a Hurdle in Google’s traffic mapping. Such a feedback mechanism will start re-routing traffic miles before the accident — or whatever has created the hurdle. At a statistical level, the routing software can say “It’s 0800 and the I-94 is already a horror story, don’t even consider it as a possibility”Report

          • Mike Schilling in reply to BlaiseP says:

            “It’s 0800 and the I-94 is already a horror story, don’t even consider it as a possibility”

            That’s after the system has already created the horror show, of course.

            The algorithms that are optimal for the system will include random factors to avoid a single bottleneck, much like ethernet includes a random backoff on collision. The algorithms that are optimal for an individual car will always pick the route that’s fastest right this minute. This is a good place for regulations to resolve the resulting prisoner’s dilemma.Report

            • BlaiseP in reply to Mike Schilling says:

              We can’t count on everyone being on the Driverless Network. The system didn’t so much create as tolerate the creation of the horror show.Report

              • Morat20 in reply to BlaiseP says:

                Algorithms ought to be fun. I see no other possibility than fully autonomous cars, at least until the old stuff falls off the road.

                That’s not a problem — driverless cars can deal with idiots on the road better than cars being driven by people (by and large) and certainly are going to be better than the average driver before they’re allowed out on the streets.

                The sad part is simply the real benefits of it can’t come into play until you can utilize at least some sort of traffic control — swarm logic or whatever — and reallocate lanes and roads by demand.

                Rerouting around roadblocks or jammed roads can be done driverless, as long as there’s some sort of city traffic system setup it can query for road speeds.Report

              • James Hanley in reply to Morat20 says:

                Morat, I’m in agreement with your overall point, but is avoiding congestion the “real benefit”? It seems to me that the real benefit is the increased safety.Report

            • Morat20 in reply to Mike Schilling says:

              The current system creates horror shows. *shrug*.

              At least driverless cars would pervent rubbernecking, which is like half the slow-down at the scene of accidents.Report

      • Thanks! It’s nice to post. I’ll have to find more time for it!Report

  2. Matty says:

    We already have driverless cars. They’re called trains.

    Got to disagree there a train is different to a car in ways that have nothing to do with who or what drives (and in fact most trains have human drivers). A train runs a fixed route between places that are not your house and are unlikely to be your workplace, a train is a multi-passenger mode of transport and a train has a timetable it is supposed to follow. None of these things is true of a car nor would they be true if something like the google car became commonplace. We’d retain door to door at a time of your choosing transport but small children and drunks would be able to join in, which is sure to enrich everyone’s life experience. Yes I’m serious have you ever been trapped on a train or bus with either of the above?Report

  3. Kim says:

    Math’s due for a reorganization too. Possibly it will be the same one as physics, and then quantum mechanics will be really easy to solve.Report

  4. Kim says:

    I predict Gondolas. Flying through the air! (on wires).Report

  5. Jason Kuznicki says:

    More data will just make the Fermi Paradox even more paradoxical.

    I disagree. For example, we now know that term fp is very large. The reason the universe is quiet is not the lack of planets.

    We may be able to fill in other gaps very soon. The Drake Equation helps break the problem into chunks, and some of them are soluble.Report

    • George Turner in reply to Jason Kuznicki says:

      The Drake equation is silly because it doesn’t tell you anything until you know everything. It’s a way to make an unkown seem potentially knowable absent any actual data or sampling, and you could use the same method to make it seem plausible to estimate the population of Orc’s in Middle Earth or the number of mermaids that speak Latin and live in giant clam shells (which is found by multiplying the average mermaid population density of each ocean by the area of the ocean, times the percentage of mermaids who speak Latin, times the percentage of giant clams that house mermaids, time the number of giant clams).

      It’s the same method as early speculations of how many angels could dance on the head of the pin (angel size, angel spacing, and pin-head area) or Galileo’s scientific estimate of the required volumetric size of Hell.

      P = a*b*c*d*e*f*g*h*i*j is uselss if even one of the terms is unknown, since the unknown term could be anywhere between zero and near infinity. When you’re down to one remaining unknown the equation reduces to P=k*u, where u is unknown and thus P is still unknown. A simpler version is that the number of aliens in the galaxy is given by Pop=Lp/La, where Lp is the number of laser pistols in the galaxy and La is the average number of laser pistols per alien.

      The Drake equation stays around because getting a better handle on any one of the terms feels like progress toward the answer. It’s not. It’s only progress toward defining that particular term (like the number of probable planets, or the number of Earthlike planets). Until we find an actual alien civilization, the equation doesn’t rule out the answer of only one civilization in the entire universe.Report

      • Kim in reply to George Turner says:

        meh. you make guesses. some of them are intelligent.
        You won’t ever know the truth.Report

      • George Turner in reply to George Turner says:

        Well, the problem with the Drake equation is that you’re still just guessing about something with more to base it on than you had when you started, but the fact that it’s an equation makes it seem “sciency.”

        And more direct method of guessing the number of intelligent species in the universe is to just guess at it directly. All we’re really doing with the fancy equation is rejecting guesses because they seem to high or too low. It’s the same way we’d guess at how many people were really angels sent here by God and carrying out special missions. Obviously one in a million is too few, but one in a hundred is too many. There are ways to narrow the calculation, such as figuring out how many people seem to be Godly, and then multiplying by the fraction of Godly people whose existence seems divine or mysterious, times the number of events that demand a supernatural explaination, etc.

        Over on a Star Trek forum I needled all the people who derided religion as mystic nonsense since it had no actual observable proof of any divine spirits or supernatural beings, yet insisted that it’s an absolute certainty that other intelligent alien species exist – with no actual, observable proof. It’s driven by the same thing, the feeling that logic and math dictates that what you believe exists must surely exist, even though there’s not yet any evidence of it. Thus I don’t group the Drake equation with science, especially since the method trivially extends to estimating the galaxy’s average number of warp-core breeches per year.Report

        • Kim in reply to George Turner says:

          I may not have proof of other intelligent life. But I do have evidence.
          (naturally, such intelligent life does not exist in OUR universe).

          [still all points well taken! and well put]Report

        • Jason Kuznicki in reply to George Turner says:

          It’s certainly possible to make too much of the Drake Equation. (“There are exactly fifteen technological civilizations in our galaxy…”)

          But I think it’s an admirable exercise in chunking — breaking up a complex problem into the parts we can solve and the parts we can’t. Of course, we will have to do some significant updating of priors if we ever encounter another technological civilization. And they might have to as well, depending.Report

        • George Turner in reply to George Turner says:

          Well, the equation is useless until you get a sample size of two (because at least one of the key terms remains a complete unknown until then). Once you get a sample size of two you don’t need the equation at all, since you’ve got two data points in a given number of sampled stars, telling you the final answer (within some statistical bounds). Having a second data point gives you the answer even if you don’t know the other terms in the Drake equation, or that there ever was such an equation. “Number of planets of planets fitting x, y, z around a star fitting h and j with life forms of class k or m? Who cares? The answer is two civilizations per 32,500 stars.”

          It’s an equation that is useless until you have enough data, after which it remains useless.Report

  6. mclaren says:

    Precisely. The Dark equation is pure pseudoscience, because it’s an equation in which every single variable is unknown.

    The assertion that “we now know fp is large” falls apart because the most we know is fp in a small nearby region of space within our own galaxy. There are 36 galaxies in the Local Group alone (up from just 12 a few years ago), and there are at least 100 billion galaxies in the observable universe.

    Moreover, according to Guth’s inflation hypothesis, the actual size of the universe is estimated at 62 billion light years in diameter. Only the observable universe is 13.7 billion light-years in diameter, but since cosmic inflation caused the early universe to expand exponentially in size early after the Big Bang, the actual size of the universe is most likely much larger. Moreover, Alan Guth has pointed out that the actual size of the universe may be infinite.

    From the Wikipedia entry on “Observable universe”:

    “Assuming dark energy remains constant (an unchanging cosmological constant), so that the expansion rate of the universe continues to accelerate, there is a `future visibility limit’ beyond which objects will never enter our observable universe at any time in the infinite future, because light emitted by objects outside that limit (the object outside of the limit expand at the faster speed than speed of light) can never reach points that expand away from us at less than the speed of light so therefore would never reach us. (A subtlety is that, because the Hubble parameter is decreasing with time, there can be cases where a galaxy that is receding from us just a bit faster than light does emit a signal that reaches us eventually[6][7]). This future visibility limit is calculated at a comoving distance of 19 billion parsecs (62 billion light years) assuming the universe will keep expanding forever, which implies the number of galaxies that we can ever theoretically observe in the infinite future (leaving aside the issue that some may be impossible to observe in practice due to redshift, as discussed in the following paragraph) is only larger than the number currently observable by a factor of 2.36.”

    So we just don’t know. We don’t even know how big the universe is. We don’t any of the variables in the Drake equation. It’s pure pseudoscience, a completely worthless waste of time at this juncture, like cavemen speculating on the size of the band gap in a semiconductor.Report

  7. Fnord says:

    Trains are driverless? I’m pretty sure that the vast majority of trains, at least, have someone up front operating the thing. Sure, the ratio of passengers to drivers is much higher than for automobiles, but that’s actually part of the “can’t go where ever you want” problem. Even if we had rails everywhere we currently had roads, trains still wouldn’t be as flexible as cars (they’d be more like buses, still following set routes).Report

    • Murali in reply to Fnord says:

      All subway trains in Singapore are driverless. They are fully automated now.Report

      • Mike Schilling in reply to Murali says:

        The people mover at the Dallas airport is driverless too. I saw a child run out of the car’s door at the last second, and there was no one to stop his family from being whisked away from him. (The kid was too intent on whatever he’d run after to notice. His mother was pretty upset, though I’m sure they got him back safely.)Report

  8. My point about trains being driverless was not that they actually do or do not have drivers but that riding a train (or a bus or a taxi) already provides all the convenience of a driverless car without being nearly as cool.

    My point is really that I don’t see much for driverless cars outside of cool factor and marginal gains in leisure/productivity for the already super rich.Report