Morning Ed: Transportation {2017.05.24.W}
Mexico is revving up Tesla engineering operations in Mexico.
Yes, but they’re vile canisters of germs that move a lot of people very far and very fast.
Once everything is in place, the pressure to take drivers out of the equation will be intense.
Trains can be made safer. Here’s how.
A good brief article on regulations for small planes making long trips, where it seems that the push and pull of industry and government has found a reasonable compromise.
Enough about self-driving cars. Self-driving scooters!
Though about taking down old urban Interstates, the wider context of city and Interstate makes for a good read.
If superheroes can have secret identities, why can’t royalty?
I’ve maintained from the beginning that any “driverless” system that requires the human to maintain their vigilance and attention to be able to take over at a moment’s notice is worse than useless for precisely the reasons outlined in that article. Our newer trucks have adaptive cruise that will maintain a “safe” following distance and it’s already causing problems with increased rear-end collisions. Drivers get complacent and don’t pay as much attention to the traffic ahead of the car that’s directly ahead of them.
That said I still believe there will be an incremental transition from fully manual to fully autonomous. It will start with fully auto on freeways and manual on surface roads and streets and gradually to more and more autonomous as the software gets better.Report
Furthermore, any article about anything less than fully automated cars that doesn’t address the problem of expecting eternal vigilance from mostly-non-drivers is simply an industry shill piece. This is, by the way, most of them.Report
Depends on where on the spectrum we’re talking about, and how the automated/human combination works.
We already have partially automated cars in that they change gears automatically and maintain a constant speed. We can add maintaining distance from the car in front of you without altering the dynamics. We can add a lot of things. It does seem to be that we reach a point where drivers are going to start drifting off if we’re not careful.
Some people swore up and down that was going to happen with the previous two.
The “how” matters in the way Rod sort of describes. Like, if it’s a matter of “You have to manually control the car in these areas, but these other areas it can be automated” then I think we’re good.Report
Will,
We’re already at the point where most people aren’t at peak efficiency while driving. Peak awareness is fucking hard. It entails basically simulating everyone on the road, and being prepared to react at a second’s notice. It’s tensing your muscles so that you have juuust the right connection with the wheel and the pedals.
I know a guy whose reflexes are reliably worse than your normal drunkkard’s (can’t catch a falling ruler). He’s the only person I know who actually drives sharp.
What we’re really talking about is exactly how bad it is going to get. That, and alerting. If you get a shock when they need you to take control (and enough beforehand), well, you’ll be doing a lot better than if you need to pay attention yourself.Report
I don’t have any difficulty believing we are less aware as we drive compared to yesteryear. But along with that came improved automobile safety records and less deaths, in part because less attentive humans with better cars do better than more attentive humans with less ideal cars.Report
And now I think we’re moving back down the curve. Daily I see people nearly driving off the road due to cell phones / texting / distraction. I see people sitting at green lights texting. People don’t understand / know how to merge. They drive 20 miles below the speed limit. They impede traffic by driving 50 in the passing lane when the speed is 55. Everywhere “the stupid” grows.Report
Remember all the publicity about how gun deaths have “overtaken” auto deaths?
That’s not because of a huge increase in the number of gun deaths.Report
A great chunk of auto deaths are alcohol-related. If we were actually interested in reducing auto deaths, we would be expanding car breathalyzer technology.Report
Auto safety has increase, and with it, people’s carelessness has increased, but there’s only so much that machinery can do. And I’m all for more auto deaths. It “depletes the surplus population” 🙂Report
Let’s just get rid of airbags and put a 12″ spike in the steering wheel…Report
We sort of did that with airbags to eliminate pretentious pipe smokers. Where does the pipe go when the airbag deploys?Report
I would totally accept that explanation, if I still didn’t see people driving around with small dogs running unrestrained in the front seats, or people who put their feet on the dash or out the front corner of the window.
People are largely quite ignorant of the power and force in a deploying airbag. The think it’s a pillow that magically appears in front of them.Report
There may be only so much that machinery can do, but we haven’t gotten to that point despite previous proclamations that we were at that point.Report
My uncle’s new car has a cruise control setting where it maintains a fixed distance from the car in front of it. It annoys me as a passenger, because the frequent acceleration/deceleration doesn’t feel right.
(but I’m not a fan of regular cruise control either in my area, because the car with gun it to get up any incline, which is not the most fuel efficient thing)Report
Report
Fair point. I think that lane centering, combined with cruise control that maintains distance, is the line not to be crossed. The issue with maintaining distance on the freeway is not that this requires any great attention. An experienced driver does this without it hitting his frontal lobe. The issue is that on long drives it is physically uncomfortable to keep that right leg positioned just so. The problem with traditional cruise control is that the car ahead of you is going one MPH slower, and you creep up on its ass. This is both dangerous and annoying. Adaptive cruise control is a clear improvement.
But add lane centering to this and there isn’t anything that the driver needs in the immediate term to pay attention to. This will inevitably lead to reading/sleeping/fornicating rather than watching the road.
I agree with Road Scholar that freeway driving will become fully automated. This is the low hanging fruit. We are already almost there, at least for decent weather conditions, and with the caveat that there are reports of false positives with potential for ensuing wackiness.Report
We have turnpikes because pikes in the flat lands of the Miidwest were getting people killed. Given the grid pattern of roads in much of the central US, it was easy to lay out roads that were straight as a pike for hundreds of miles. (As an aside, the word pike means both a straight pole arm and “to make straight”.)
And all along those straight early roads, cars would go flying off the pavement at high speed as their drivers drifted off to sleep because they had nothing to do. So they added recurring slight curves to the pikes to keep people awake at the wheel. Those are turn pikes, a pike with turns in it.
Adding lane centering and automatic following distance to automobiles is taking our highways back to the straight pikes, and we’ll rediscover why we switched to turnpikes.Report
I do enjoy a good folk etymology, and that is a doozie! The actual origin of “turnpike” is that they were toll roads, with the turnpike being the implement that blocked the road so as to require payment. See here for a good discussion. “Pike” is simply an abbreviated form of “turnpike.” In the eastern and midwestern American context a “pike” or “turnpike” probably started life in the 18th or 19th century as a privately owned toll road.Report
It’s sort of like how there used to be a NASCAR rule that you had to have at least nine yards of track width in any curve, and when race car drivers skidded during a turn but stayed on the track they would be said to have taken “the whole nine yards” to make the turn.
Or how it used to be that no hole or pocket in a car could be deeper than the average thumb (so that you wouldn’t have to be distracted by digging into a deep storage container) and that’s how we got the phrase “rule of thumb”.Report
Report
https://www.youtube.com/watch?v=xECUrlnXCqkReport
“Rule of thumb” in the usual meaning goes back much farther than that. I’ve read a number of antique fencing (swordsmanship) books. The phrase is used in one that dates to the 1690s where the author is explaining that he’s teaching a complete approach using a firm theoretical basis that would produce more consistent results than proceeding by “rule of thumb”. The phrasing suggested to me that he expected readers at the time to understand it to mean approximate or non-systematic.Report
I agree there are very few good articles on this subject. I can think of maybe three offhand.
First, I believe they’re seriously over-estimating the consumer demand. I saw a poll a few years back that asked if you’d be willing to pay an extra $2k for automation and the positive response was only about 20-30%. Interestingly, women were the least enthusiastic. I would posit that’s due to less interest, enthusiasm, and trust in technology in general.
Second, the real driver (heh) for this is going to be the trucking industry. Drivers’ work schedules are constrained by Hours of Service rules. For a solo driver this caps capital utilisation at about 35% – 40%. Automated trucks could easily double that number, so you can see the economic incentive.
Third, you aren’t going to see completely non-attended trucks running down the road any time soon. Driving down the road is only part of my job, and quite frankly, the easiest part, and the only part that anyone is actually talking about automating. For example, I spent the early years of my career pulling flatbed trailers. Basically a flat platform that you chain/strap cargo down to. Guess who does all that chaining, strapping, and tarping? (And undoing it all at the other end.) And then there’s loading, unloading, fueling, pre- and post-trip equipment inspections, etc.
So what’s going to happen is that the job of “truck driver” is going to evolve into “heavy truck operator” with less and less of that operation consisting of driving per se. There will be fewer trucks on the road because the operator will be able to go into a non-driving duty status and read/sleep/fornicate while rolling down the interstate, yielding that increased equipment utilisation percentage.
None of this can happen without both the technology and the regulatory elements cooperating.Report
Road,
2.5 cars per driving age American Adult right now. Nobody’s paying for shit. Bottom’s about to drop out of the market (particularly the used car market, as the leases come up).
I hear you on “the REST of my job” and definitely see automated trucks before anything else (Insurance Companies Fix Everything!)Report
Those numbers are probably too optimistic. Here is an article on affordability from which I take these numbers:
$20,806 = what the average American can afford to pay for a car
$30,000 = what the average American spends on a new car
$8,500 = the average projected markup for adding self-driving technology in 2025
$5,000 = the average projected markup for adding self-driving technology in 2030
$3,000 = the average projected markup for adding self-driving technology in 2035
Basically, most Americans cannot afford a new car, and fewer would be able to afford the initial thirty-percent markup for self-driving tech. The more people buy it, the cheaper it becomes, so the sad upper-middle class suburban commuter will be demanding tech mandates, roadway changes and modified liability systems to subsidize their lifestyle preferences.Report
This ties in with another question I have about driverless cars that I never see addressed. What is the failure mode? Whenever I ask this, everyone jumps in about accidents, but that’s not what I mean. I’m talking about a car with the far side of 150K miles on it, when random bits start to fail. To own an older car is to engage in constant triage. What can be taken care of in six months, what needs a trip to the shop this week, and what needs a tow truck right now? How does this translate to driverless cars? On the few occasions I see any mention of maintenance, the unstated assumption is to treat it like a new car under warranty, or even more, like a car with a three year lease. So what happens when the lease is up?Report
Why does it have to be different than now? You make repairs, risk it, whatever. Main difference I see is that at some point the cars may go on strike if certain repairs aren’t made.Report
My point precisely. I have a mechanic I trust. If I take the car to him for an oil change and he tells me about other work that should be done, I believe him. The ensuing discussion will cover priorities, from “we need to keep an eye on this” to “I will tackle you if you try to drive away before fixing this.” He considers it perfectly reasonable to include finances in the discussion. Three paycheck month coming up? Sure, it will wait until then. If it was “I will tackle you” situation and I didn’t have the money right now, he would front me, because I have been with him for years and he knows I am good for it.
If we are talking about a driverless car, how does this discussion work? I have a sneaking suspicion that the programming will err on the side of treating a burned out dome light as a critical failure that must be corrected immediately.Report
Richard,
If it’s being pitched to insurance companies, it’s going to depend on actuarial risk tables.
If it’s being pitched to consumers, it’s going to try not to be a pain in the ass for MOST customers. Please remember, we live in an age where we have 2.5 cars per driving age American Adult. Most customers buy/lease new and then add to their debt on the next car.Report
It’s possible that they will be overly cautious. It depends on what the liability and political incentives are. Given how popular auto inspections are in the absence of evidence that they do much to improve safety, my guess is that people will generally be supportive of a restrictive regime. Which, if so, so. I don’t expect it to get in the way of anything.Report
I assume this gets wrapped into liability reform. For this to work, I think Congress would have to create a strict liability program that overrides local product liability and negligence actions. The car manufacturer is going to be strictly liable for vehicular accidents, unless it can show some sort of overriding negligence on the part driver. I doubt failure to perform maintenance would be publicly acceptable unless that failure was the actual cause of the accident. But all of this points to the car computer system adjudicating fault.Report
PD Shaw,
at some point, you get “undrivable” (possibly drive only during daylight if a headlight is out). But yeah, that sounds about right.Report
Pay an extra 2K to own 100% of a self-driving car instead of 100% of a human-driven car? Of course not.
The whole potential of self-driving cars is to pay less because you own less than 100% of a car.
Pay only 37% of the price of a human-driven car, because you own a 1/3-of-a-car share of a 3000 household, 1000 vehicle self-driving car co-op, each of whose cars is 8-10% more expensive than a human-driven car? (And save hundreds a month on downtown parking, since the car drops me off at work and then heads out to pick up the next co-op member, or find a place it can park for free)
That’s more like it.Report
But how many members of that co-op have roughly the same 9 to 5 work schedule? The idea will be a lot less enticing when you are trying to get to work, but can’t because all the cars are already checked out. We might also contemplate what this will do to rush hour traffic, as the cars that did the early morning drop off are all heading back out to the suburbs to pick up the late morning crowd.Report
Rich, if we assume a 30 minute commute, then you can probably get 3 trips in from 7am delivery to 9am delivery.Report
I think one car per three members would likely do alright. Allowing for some folks who work evenings, work weekends, telecommute part time, live centrally and work further out so the next pickup can happen close to that dropoff, etc.
But fair enough, maybe the figure is one car per 2.5 members, not one per 3. I really don’t think it’s likely to need to be much higher than that.
And while I grant that single-occupant four-wheel motor vehicles remain terribly space-inefficient compared to walking, mass transit, cycling, telecommuting, or literally anything else ever invented – they’re still better than a downtown that’s half parking.
More space devoted to destinations, and transport to destinations, thanks to less space needed for storing great honking metal boxes next to destinations, thereby decreasing the distance between destinations, would also allow more people to walk between some of their destinations.Report
Co-ops at that scale seem unlikely; I don’t see the advantage of setting up a small co-op over ordering a car as needed (or buying into a larger subscription like Zipcar).
In general, I think that there aren’t enough comparisons to taxis and other car services in these discussions. Today, these are generally good enough to rely on in large cities, but not a great option outside them, and even in well-served areas lots of people choose to drive their own cars. Self-driving probably changes the picture quite a bit (labor is, as I understand it, the most expensive part of operating a vehicle), which might expand the areas in which these are viable; however, lots of space will likely remain outside the reach of reasonable service and lots of people will likely still prefer having their own car at their beck and call.Report
Fair enough, maybe it won’t take the form of co-ops, but simply displace cabbies and make cab fares much cheaper. Maybe automobile companies will take out the middleman and displace cab companies.
Whether users’ payment is per trip or per month, it comes out to the same ultimate result – the crossover point where not personally owning a car is cheaper and as or more convenient than owning one, moves up to more automobile trips per month, and down to smaller communities, than where it sat before.
More people will avail themselves of the services of shared cars, and fewer will choose to individually own cars.
Less space will be needed to store individually owned and operated cars during the 95% of the time when they are not moving, most notably in high-rent areas like downtowns; meanwhile the smaller numbers of cars in active circulation will spend more of their time in motion.
Many housing, recreation, and work locations will gradually be de-coupled from parking locations, and many existing apartment building and workplace parkades will probably be converted to other more productive uses.Report
Uber has automatic driving cars in my city.Report
There will be fewer truckDRIVERS PERIOD because the……
FIFY (and I’m not really happy about it)Report
I highly recommend this book.
Digital Apollo: Human and Machine in Space Flight
It covers the stability, automation, and human interface questions from World War I to landing on the moon.
As a sampling of the book, here are some tidbits.
During WW-I and into the early 20’s, there was a great stability debate in the aeronautical community. Should a high performance aircraft be stable or unstable? The advocates for unstable aircraft held the upper hand because an unstable aircraft is more maneuverable and responsive (better fighters). But the debate passed them by because improved aerodynamics in the 1920’s gave aircraft greater speed and endurance, and unstable aircraft couldn’t be flown long distances because they simply wore out the pilots. Another reason for stable aircraft was that higher speeds and longer ranges burdened the pilots with increasingly complex navigation tasks, and increasing overall aircraft complexity required the pilot to also be a system manager. If he had to devote all his attention to keeping the horizon level in the windshield, then he couldn’t do all the other required tasks.
Fast forward to the X-15, and they’d created an aircraft that couldn’t be flown manually because it went through too many flight regimes, which meant its stability behavior changed so much that a human couldn’t adequately compensate for all the different “feels” it would have. So they built an analog computer that represented the ideal X-15, and then used comparators to null out the difference between what the real aircraft was doing and what the ideal analog aircraft was doing. The pilot flew the on board analog control system and the X-15, with the pilot on board, followed along.
But Apollo had to be digital, and the vehicle’s control was so complex that a human couldn’t possibly fly it. Not even Neil Armstrong could fly it. So they delved deep, very deep, into how a human and a computer could work together in the optimal way to accomplish a mission. After Apollo, engineers took an Apollo Guidance Computer, reworked its software, and put it in a fighter aircraft to see if a computer could solve the stability issues inherent in supersonic flight. It worked, and so we shifted into the computer age of flight controls.
The book addresses a lot of questions on what you should automate, what you shouldn’t, and why, with deep looks at human factors and how people behave, including their psychology. For example, the Lunar Module could land at the pre-selected landing site in fully automatic mode. But it was never used because every single time there was a rock in the way, but a clear area just a little bit further away. Every single time. So on every mission, the pilot had to put his hands on the controls and land it “manually”. There was never any chance that a military test pilot on live TV would land on the moon in automatic mode no matter how good automatic mode was. The mode was a waste of core memory that was there to make the programmers happy.Report
Robot Cars: Dear god I will rue the day. When all the robots are following the law, it’ll be hell. Want to change lanes to go around the accident? “i’m sorry, that is illegal. Double yellow line”. So we sit and wait ….forever. .What I find curious is we have the technology now to not have people at work. I can effectively do my job from anywhere. I can skype meetings, and do. I can email/call/skype/text people. Other than the manufacturing folks and the facility folks, no one in my entire company actually needs to be in the building
Trains: They’ve been talking about positive train control for a while now. At least the last two train wrecks.
Scooter: I’ll say the same thing I told my Dad when he was in a power wheel chair. Put a cow catcher on the front and a horn. As to the video, loose the damn blinking light and, I assume, annoying back up sound, cow catcher, and some Kevlar shielding. Replace blinking light with 360 degree rotating SAW filled with armor piercing and incendiaries rounds and you’ve got yourself an urban assault / pacification skooter of DOOOOM!Report
Yup. Just this week, on my morning commute, there was a two-car accident on a windy stretch of two-lane (one each way) highway through a wooded area. The crash was fresh enough when I got there that fire and police had not yet arrived, but old enough that traffic was backed up perhaps half a mile. A utility vehicle of some sort had positioned itself strategically, with its yellow lights flashing, and the utility guy was directing traffic. He was doing a good job, alternating directions and keeping traffic flowing as well as possible, while keeping room for the emergency vehicles once they showed up. He had absolutely zero legal authority for this. He simply was doing what needed to be done, and the drivers tacitly accepted this. It worked as well as could reasonably be hoped for.
No one has ever explained to me how a fully automated car would handle this. The closest I have seen was one car company (Nissan? I’m not sure now) that admitted it couldn’t, and was setting up essentially a remote call center for humans to intercede. This does not inspire confidence.Report
Rich,
pull off, park, wait. Otherwise, use a manual override.Report
Google cars don’t have a manual override. That’s what makes them fully automated.
Winding two-lane roads through woods don’t really have a “pull off, park” option. This is where the accident occurred: https://www.google.com/maps/@39.3522061,-77.0158795,3a,60y,212.61h,69.72t/data=!3m6!1e1!3m4!1sbKo34-yJpwn0BTPpGuNaQw!2e0!7i13312!8i6656
In any case, if the car doesn’t have the capability of crossing that double yellow line, because that would be illegal, then the “pull off, park” answer means sitting stopped for a couple of hours until the accident scene is cleared out. Given that my car with its meat driver was delayed between five and ten minutes, this seems a sub-optimal solution.
Had both lanes been blocked, the meat-driver solution would have been to make an illegal U-Turn and take an alternate route. We are routinely informed that driverless cars follow all traffic laws to the letter. In the real world, the system only works because people don’t actually do this. I assume that something will give in our automated future, but no one seems to talk about what that something is and how it would work.Report
Rich,
That’s got enough room to pull off, at least on one side (be difficult to get back on, but…). Still, point taken on country roads.
Google cars had BETTER have capability to let the ambulance go ahead of them (that’s generally “pull off a lane”)… if they don’t, they’ll be in trouble for traffic laws.
U-turns anywhere are legal in my state unless posted. We may see traffic law modifications in order to allow driverless cars to “drive safely” within the law.Report
“When all the robots are following the law, it’ll be hell.”
Oh look, there’s a pedestrian on the side of the road. Hey, they kind of vaguely moved towards the road. The law is that you yield for pedestrians so I better SLAM ON THE BRAKES RIGHT NOW. Since I’m a robot car I don’t care about the humans spilling coffee all over themselves. Whoops, I got rear-ended because I stopped short in the middle of the road for no apparent reason! Fortunately the law is that rear-end collisions are the fault of the following vehicle, so the accident is considered to not be my fault.
******
Of course, if it’s illegal for your car to not be a robot, things get a lot simpler.Report
DD,
Or you could just let CAT drive. CAT has very good visual processors, if he is still an asshole.
(Cat is currently hiding in Japan, after accidentally deleting months of someone’s work.)Report
The onboard computer will be aware that it is being followed, how closely, at what speeds, etc. Likewise it will be able to match the pedestrian’s movements to literally millions of hours of network training, asking the question “how likely are they to step out on the road” versus “how likely are we to get rear-ended” versus “how quickly can I stop safely” — and among those variable it will choose an action.
The end result is, it has to make the same decisions a human does, weigh the same factors, without missing any details or getting tired or having a “panic response,” etc.Report
Sensors will get you so far, but computers aren’t great at abstract thought:
1)you have 5 tumbleweeds entering the roadway
2)you have a tornado 800 yards ahead
3)hailstorm
4)the bridge is outReport
Honestly, I expect things like tumbleweed and dust storms will be a big problem for the early generations. But once some event happens a few times, if it proves to be a real nuisance, the engineers can start training the networks for it.
An anecdote, one of the early test models had a problem with riders of fixed gear bicycles. See, the system had been trained to recognize cyclists, but only on “freewheel” setups. A typical freewheeler will stop at an intersection and put their legs down. “Fixed gear” folks, however, I guess often keep their feet on the pedals and kinda do a short forward-back cycles to keep balance while waiting for the intersection to clear.
The cars, of course, had no idea what was happening, because the bike kept “lurching” forward as if it was going to go into the intersection. So they waited. Forever.
Until they trained the network on that. Now it can sense that behavior and make a correct decision.
Likewise for accident avoidance. This becomes a “global” (more “within local area,” if taken literally) control/optimization problem. But just as a person can direct traffic, a computer can. Of course, this would require coordination, communication, and a number of other solvable problems.
My point, the problems are real. They are also solvable.Report
It will be interesting to see. The problem I have with reality is it is a parametrically intense environment to model real time. Then it is a completely another thing to recognize the correct action when faced with something abstract, and that is what reality often throws at you, something abstract.Report
Joe,
Depends on the computer, really. When the AI starts running scams and figuring out new angles to make money via pornography (I remind you exactly how many porn websites there are out there), then I start to believe that even a hobbled version will be able to do something decent during a hailstorm. And we have AIs that are that intelligent now (and offering selected versions of themselves for use as driving AIs)
The bridge is out is an easy one, so is the hailstorm.Report
“The onboard computer will be aware that it is being followed, how closely, at what speeds, etc.”
What difference does that make? If the following car hits me the following car is at fault. The. End. “yeah but no but yeah but no but it’s different because” if the following car hits me the following car is at fault. Pay attention and don’t ride my tail.
Likewise it will be able to match the pedestrian’s movements to literally millions of hours of network training Sure it will, just like the speed limit should be set to the 85th percentile speed but somehow it’s always 10 mph below that number because reasons. Law trumps code. The law says stop so you stop if you think there’s a pedestrian entering the roadway.
“it has to make the same decisions a human does, weigh the same factors, without missing any details or getting tired or having a “panic response,” etc.”
Stepping on the brakes to stop the car when a pedestrian enters the roadway isn’t a panic response, though. It is, in fact, what human-operated vehicles are supposed to do but invariably don’t.
People keep saying “oh what if there are pedestrians, how will the car decide whether to run them over or get into an accident” and the answer is obvious–get in an accident. Cars have crashworthiness standards, occupant protection devices, and laws that (in theory) define an environment wherein all participants can safely handle the unexpected.Report
This isn’t actually that straightforward, legally speaking. Remove automation from the equation. Talking about plain old regular cars, suppose you are driving down the freeway at normal freeway speed with normal freeway distance between you and the car ahead. The he suddenly slams on the brakes on whim, and you hit him. What then?
I live in a “contributory negligence” jurisdiction. That means that, in theory, if the accident is partially your fault–even just one percent your fault–you can’t collect. Most of the country is “comparative negligence” jurisdictions. This means that if you are partially at fault, the court allocates fault between the parties and prorates damages accordingly.
So in our scenario of the guy ahead of you slamming on the brakes on a whim, there would be a pretty good argument to be had that he was partially at fault. The implications of this argument depend on where you are.Report
“prorates damages accordingly”
Any bets that the law will eventually favor ‘the code’.Report
In some states, at least, you’re still at fault.Report
Years and years ago, a NJ state trooper told me that their normal practice in the situation where someone pulled out and an oncoming car struck them was (a) if the oncoming car hit the rear of the car that pulled out, the driver of the oncoming car was ticketed for failure to have their vehicle under control; or (b) if the oncoming car hit the side of the car that pulled out, the driver of the car that pulled out was ticketed for failure to properly yield the right-of-way.
Of course, around that same time, New Jersey magazine ran a story about how to be a NJ driver that included tips such as “Never make eye contact with the driver of an approaching vehicle. If you don’t make eye contact, then you can pull out and it’s up to them to avoid you.”Report
You all are acting as if the law won’t adjust to the reality of AI. The point is, this is a fairly basic decision theory problem, and indeed there will be cases where someone is going to get hurt and the AI will have to decide who. But that is no different from humans. The difference is, the AI will make decisions more quickly and more accurately.
There is a psychological barrier, where humans apply their own faulty reasoning, but computers certainly can provide a system closer to optimal than human drivers.
On the other hand, yes, talk radio will be full of bizarre conspiracy theories, just as our Kim will have “friends” who have the inside scoop from a rouge engineer. A certain level of political crazy is baked in. So it goes.
Give it time though.Report
Ha, the law is a AI institution.Report
“You all are acting as if the law won’t adjust to the reality of AI.”
You could use the same argument that the law should adjust to the reality of human behavior. Nonetheless, we persist.
And what, exactly, is the car going to do other than stop? If you’re going to suggest that the car might run an optimization problem and conclude that it should take the risk of hitting a pedestrian, then you’re intentionally getting yourself stuck in a trolley problem. Because now you have to explain why it’s okay to build and deploy a car that will intentionally not take every possible step to avoid running someone over.
“computers certainly can provide a system closer to optimal than human drivers.”
Why is “never run over a pedestrian no matter what” not an optimal system?Report
One thing to keep in mind, and I believe it’s important: If our legal culture prevents it from happening here, some other country with a different legal culture will let it happen there. If fear of plaintiff’s attorneys prevent it here, other countries are less litigious. If our laws err too much in a particular direction to allow it to happen, it doesn’t follow that every other country will have the same problem.
And once these things exist and are on the roads in other countries, we will want them. The political pressure to free up the logjam will be really intense. We’re not going to let Canada have these awesome things that we don’t get because of lawyers or a philosophy puzzle.
So the barriers here are going to be technical. (I do think “humans and robots sharing the road” are going to be among the biggest.)Report
“We’re not going to let Canada have these awesome things that we don’t get because of lawyers or a philosophy puzzle.”
Remember that whole discussion about Epipens, and how there’s like eight different kinds in Europe but we can’t even get two kinds here?Report
Ah, but we have one kind here! And for most that’s enough. And it’s a product most people don’t want it need.
Self-drivibg cars? A lot if people are going to want that. Entire industries are.Report
DD,
Because at some point, the pedestrian is at fault. At some point, you can say “99.99% of the time, the pedestrian doesn’t BOLT into traffic.”
If you do not say this, then cars are unable to drive more than 5mph whenever a pedestrian is within 10 feet of the road (which is roughly “I can bolt into traffic”).
Truly, a man walking down the sidewalk is UNLIKELY to turn and walk straight into the road, right in front of an oncoming car. You do not design a vehicle to slow to “can’t hurt him” speeds, simply because if he does walk into the center of the road, well, that was dumb. You may attempt to not hit him if he does go into the road (humans would too, naturally), but you don’t play the .0001% game.Report
“Because at some point, the pedestrian is at fault.”
Oh, the pedestrian’s actions are often the reason that an accident happened. But there’s a difference between “why the accident happened” and “at fault”.
“If you do not say this, then cars are unable to drive more than 5mph whenever a pedestrian is within 10 feet of the road (which is roughly “I can bolt into traffic”).”
Well…yeah.
See, you’re thinking about this as “well this is how it’s always been so this is how it should just keep being“. And I’m saying “the reason it’s always been that way is that it hasn’t been possible to force people to do things any differently”. And I’m looking at things like this and thinking “what if you told the San Francisco city government they could legally mandate that all vehicles in town have technology installed that made it impossible for them to run over pedestrians”.
Besides, if you aren’t driving and you aren’t paying by the minute, then why do you care how fast the car’s going?Report
v,
With the whole greyballing thing, knowing a rogue engineer is in the realm of “I’m just telling you about the things you read about in the paper.”Report
It is, in fact, what human-operated vehicles are supposed to do but invariably don’t.
The result being thousands of needless deaths a year.
To be clear, are you presenting the introduction of automated cars that are better than human drivers at predicting when a human is entering the roadway, and taking the necessary action to avoid killing them, as a bad thing?Report
“are you presenting the introduction of automated cars that are better than human drivers at predicting when a human is entering the roadway, and taking the necessary action to avoid killing them, as a bad thing?”
Not at all. This is a caution about mixing human drivers and robot ones. As has been pointed out, robot drivers can’t not follow the rules. The two ways to solve this are A) a sufficiently-detailed rule set, or B) strongly limiting the variables in the environment.
People keep wondering how robot drivers will react to the complex roads with all their different situations. I think it’ll be the same way that modern cars deal with gopher holes and drifting sand; by traveling only on prepared roadways that lack these things by design. If the only cars on the road are robots then there won’t be human-robot complexities to worry about.Report
Damn…. i don’t want to tangle with a gopher that can dig a hole in asphalt. But poor maintenance and harsh weather like snow will be an obstacle for robo cars. Granted Anchorage is an outlier but our roads are often covered with snow for weeks. Anything that needs to see road lines or curbs is useless.Report
Google drones fly down potholes here. They don’t come back.Report
DD,
Robot drivers can indeed not follow the rules. You can, if you’re crazy, let CAT drive. CAT is an asshole, and actually managed to infiltrate an art museum’s online presence.
Yes, it is possible to make a robot car that will do normal decisions. Rulesets aren’t the only way to make an AI.
We have enough processing power right now to let a general AI take over a car (with, again, some limits on its processing).
Yes, at first, it’ll be “run trucks at night on highways” (because, do the easy stuff first, duh!)Report
The secret identity link goes to an article on why techies need to stop trying to improve tv and film.Report
What do you expect from a secret identity link?
It’s not going to tell you where it is, that’s for sure.Report
The environmental lawfare chickens have come home to roost in suburban Maryland.
(I am in favor of this project, but you live by community activism and the EIS, you die by community activism and the EIS)Report
The strange thing about the article is that there’s no mention of the Environmental impact… it just references the Judge stating that he doesn’t really like the 5-ridership cases that underpin the business case for the project.
The environmental connection appears to be that by basing his ruling on his assessment of the business case, the judge was able to thwart an environmental approval necessary to secure Federal Funding?
Possibly its a terrible article written by cub reporters… but as its written I can’t help but wonder why exactly we have judges evaluating business cases? A shitty business case is a shitty business case, but that oughtn’t, under ordinary circumstances, be subject to Judicial authority – unless the authority is scrutiny of business cases… which the article fails to describe why a range of ridership figures – which range from 0% to 25% feeds from Metro – could possibly serve as a reasonable basis for a ruling.
I mean, in the good old days we used to at least have an uncommon local slug or a snail to put on the masthead of defiance. (Not that I particularly care whether the Purple line is built – as long as the Silver line gets completed, that is.)Report
IIRC, and this has been ALL OVER NPR the last two days, the judge rejected the old ridership volumes since Metro’s ridership has tanked with the whole smart track maintenance efforts. The ridership volume drives the need for the line, which is in doubt, and makes the calculus of the environmental impact different.Report
Yes, it’s down but does anyone really expect ridership to stay down? It seems to me that ridership will pick back up once maintenance is done.Report
It depends. It depends upon whether or not safe track actually makes a difference. It depends on whether or not DC suceeds in pressuring metro to AGAIN sacrafice maintenance to keep the trains running late nights on the weekend so bar hoppers in Adams Morgan can get home. It depends on whether Metro manages to an instill a mindset of safety. It depends if the various states agree to fund Metro. It depends if riders don’t get fed up with all the delay or and actually see service improvement.Report
Metro’s being digging themselves into a hole for a long, long time. I’m not willing to go as far as ‘death spiral’ as some others, but even after Safe Track and Back2Good are done, the system is still going to have reduced service and higher costs than they had a few years ago (and remember how good MetroForward worked out)
The system added 5 new stations in an entirely new service area and they *still* had an overall decline in ridership.
Biking is actually soaking up a fair amount of short distance (within 5 or 6 miles) ridership, and Uber&Lyft etc are filling in the gaps on some of the longer distance ridership.
Plus, for the foreseeable future right now, gas is cheap, and monthly parking isn’t that expensive (and is also subsidized the same as Metro for fed employees)Report
Yes, they have been. I grew up on the Orange Line in the heyday of the Metro and have watched it’s decline with some sadness.Report
A decent review (but from a point a view with a definite and unapologetic agenda) of how this evolved can be found on this blog (links are in reverse chronological order
As I understand it, any EIS has ‘impact’ and ‘benefit’. The ‘benefit’ side of the calculation uses ridership projections, but the people that objected to the project made the case that the ‘benefit’ was being miscalculated, because the projections overestimated ridership, as the underlying data depend on Metrorail and bus ridership, which has seen a downturn since 2009 (eta – for numerous reasons, safetrack just being the latest, and all of them because Metro has been grossly mismanaged for decades with all critics thru that time being called transit haters or racists or both until it’s now an unignorable crisis)- and thus the purple line ridership numbers need to be re-calculated with a new Metro baseline.
People are making all kinds of accusations at the judge; my only problem with him is that he did sit on this ruling for over 9 months, which is just not right. S*** or get off the pot.Report
With more precision the ruling says the the Federal Transit Administration needed to address formally, within the context of the EIS public comment process, why they weren’t changing the purple line ridership projections due to the decline in Metro ridership, as some had argued they should have during the public comment process.
So it’s very lawfareish.
eta- like, this is the very petard that people who never granted deference to Bush era EPA decisions (and went to SCOTUS over them) are hoisted upon.Report
Marchmaine: This is a NEPA case. The key language in the article is the bit where the judge is quoted as saying that the project proponent failed to take the requisite “hard look” at the effects of another project. “Hard look” is classic NEPA terminology.
Kolohe: Close but not quite. NEPA analysis starts with a “project”, in this case being building the Purple Line. If the project clearly will have minimal impact, the lead agency then prepares an EA (environmental assessment) which demonstrates that the project will have minimal impact, and then the agency adopts a FONSI (finding of no significant impact). Alternatively, if the project is likely to have an environmental impact, the agency prepares an EIS (environmental impact statement) followed by the adoption of a ROD (Record of Decision).
A fair EIS must consider a range of reasonable alternatives, in light of the project purpose and in light of the cumulative impacts of other projects. A EIS must also respond fairly to comments submitted by the public. So if the agency received a comment to the effect that its ridership projections were widely inaccurate and the ridership projections were a core component of the identification of the project purpose, the agency cannot simply ignore that study. The agency must explain why the study is flawed or the agency must have some other defensible basis for choosing to rely on its preferred studies.
Transportation EISs are a huge pain because the planning window is so long. Agency studies are frequently commissioned years in advance, so by the time that the EIS is released to the public facts on the ground have changed. Laying the groundwork for a successful defense takes substantial advance planning by agency leadership, its NEPA compliance team and its attorneys.Report
It sounds to me like the judge found that the transit authority had not adequately considered alternatives to building new light rail. This seems like classic NEPA stuff. The Federal Agency is required to evaluate whether this is the best approach to its objectives in light of environmental concerns through an environmental assessment:
EDIT: Looking at Kolohe’s first link, this is an environmental impact statement, not an environmental assessment. I think the basic purpose of the study is the same, just more detail is required of EIS.Report
NIMBYism in action. One reason why European infrastructure projects get built faster is that European politics gives project opponents fewer ways to gum up the works.Report
It appears there are the stirrings of a revolt over Brown’s gas tax and an effort to put it to a referendum. I don’t think him calling the opponents “freeloaders” helps, but then again it’s gov moonbeam.Report
I don’t know that a referendum brewing for an issue is so much an indication of a revolt as it is a normal Wednesday in California.Report
Colorado Gov. Hickenlooper signed a bill creating a commission to study passenger rail along the Front Range. The General Assembly passed the bill funding the commission with bipartisan support. Based on my almost 30 years here, there’s some sort of tipping point that’s been reached when Colorado Republicans are willing to even study some form of transportation that’s not roads.Report
How long do you suppose it will be before they claim that they have always been for passenger rail, any claim that they didn’t is yet more proof of liberal media bias, and it is the Democrats who are the true opponents?Report
I’ve argued for a while that, at least in some states, there’s a gap between the state-level Republicans and the ones who go off to Washington. I haven’t looked at the voting details on this bill, but suspect it was passed by urban Democrats and suburban Republicans. A few years back Texas created a water infrastructure bank; my contacts there told me that it was passed largely by urban Democrats and suburban Republicans in order to fund projects that would take water from rural East Texas areas for the Dallas and Houston metro areas.
A strip 40 miles wide centered on I-25, running from New Mexico to Wyoming, probably catches 80% of Colorado’s population. If not today, it will in another ten years. It’s an obvious thing: Heinlein casually dropped a reference to the “ribbon city” from Pueblo to Denver in The Moon is a Harsh Mistress in 1966. He cut it short on the north end, but when he lived in Colorado, it was in Colorado Springs. Along the highway 287 corridor a few miles west of I-25, from Denver’s west suburbs 60 miles north to Fort Collins, there are few open spaces and you’re never out of sight of a development.Report
Here in my red county in a blue state the local politicians run the gamut from sensible Main Street Republicans to the barking mad county commissioner who rails about UN plots to take away our liberties. (Sadly, I am not exaggerating for effect.) I have little problem with the Main Street crowd. For most of the stuff that matters to local government, they are fine. The barking mad crowd is a minority, but a large enough group that there is constant danger of their gaining temporary control of something that matters. At one point two of our five county commissioners were crazies. That is far too close for comfort.
Right now in the next town over one of the town councilmen is barking mad crazy. The rest of the council–conservative Republicans all–are currently working at getting him recalled. I gather that this is mostly because he is a total pain in the ass, though that’t not the official reason.
Then there are the guys sent to the state legislature. They used to make a great display of ideological purity, refusing to deal with the Democrats, making them dead weights. I haven’t heard this talk lately, so perhaps they have decided that the possibility of getting something done also has its uses. But there isn’t a one of them that I would trust not to go full crazy, if they saw this as the path to advancement.Report
This is a welcome change.Report
Ah ETOPS, or Engines Twin Or Passengers Swim.Report
To go back to the thread yesterday about libertarians. Fine, I’ll accept the majority of libertarians aren’t evil mustache twirlers who only care about low taxes and hate anything passed after Teddy Roosevelt was sworn into office.
If that’s true, why is the Foundation for Economic Education writing things like this? I mean, where’s the libertarian foundation not writing silly things or focused on inane things that actually gets funding? Like, there’s Radley Balko who’s fantastic, but even he left to join the Post.
https://fee.org/articles/birth-rates-are-at-historic-lows-and-here-s-why/?utm_source=sumome&utm_medium=facebook&utm_campaign=sumome_share
“They were aspiring adults and given as many responsibilities as they could handle within their range of competence, which was always shifting in the direction of more and more.
There was no Department of Labor and Department of Human Resources to “protect” them from living full lives. Kids in those days were regarded as valuable because they were tangibly productive. They worked, gained skills, and produced for their families or otherwise worked for businesses here and there. They were assets. As they gained skills, discipline, and a work ethic, they could become ever more valuable to their custodians and communities. This is a major reason why people wanted them. And the kids, in turn, were socialized to be grateful to their benefactors whether at home or work.
And notice from the story of Anne that a main job of kids in those days was to care for people in their aging years. So kids were valuable on both ends of the life spectrum: as co-workers when the kids are young and then as helpers as their custodians age.
What’s different today? Now kids are mostly a financial cost and defined as such, because the law, educational system, and welfare state make it that way. Oh sure, people still love their kids. Emotionally and spiritually, we speak piously and beautifully of the infinite value of their lives. “Report
Let me try to understand the question before I answer it: If the majority of libertarians aren’t evil, why did this organization publish a particular paper?
Did I get that right?Report
No, why are the only major libertarians organizations either nutbars like this or over at Reason, disproportionately focused on finding wacky college students protesting almost as much as Rod Dreher is?
Like, if libertarians actually believe in getting rid of corporate welfare first, removing chains before removing benefits, and all the smart things James Hanley has said on this site, where’s that think tank? Because whether you agree with the policy or not, places like the Center for American Progress, Brookings, the Heritage Foundation, and the like push forward policies that the majorities of those ideologies support.
If you by the work being put out by libertarian think tanks, the most important things in the world are making sure 9 year olds can work in factories and making fun of wacky college students.Report
Sounds like you want the IJ.Report
The same reason prominent liberal commenters spend time pointing at libertarian outliers who point at liberal outliers, I suppose.
Cato Unbound, for example, is discussing “Contemporary Perspectives On Religious Liberty”.
Like, if libertarians actually believe in getting rid of corporate welfare first, removing chains before removing benefits, and all the smart things James Hanley has said on this site, where’s that think tank?
Here’s the Cato Institute’s webpage dedicated to corporate welfare.
It took me 10 seconds to find on Google.
If you by the work being put out by libertarian think tanks, the most important things in the world are making sure 9 year olds can work in factories and making fun of wacky college students.
I’m thinking that you’re wrestling with a strawman here.
For one thing, the essay that you originally opened with was talking about the set of responsibilities given to children and comparing what was normalish in the past with what is normalish now and regretting that the attendant costs for having a child have skyrocketed. (And if you’d like me to find you an essay from the left complaining about prolonged adolescence (perhaps not using those words, of course), I know for a fact that I can find them for you.)
You’re making sweeping generalizations about an entire political philosophy based on a handful of outliers when there are other examples that are doing the things that you’re asking “why aren’t they doing this?”
It’d be like me pointing at college students rioting and asking “why don’t the left actually try to help people get health care instead of breaking Starbucks windows?”Report
Maybe because those wacky college students draws eyeballs? Maybe, like me, they recognize that these kids will one day be adults, and god help us if even a small percentage of these kids keep their wacky ideas about cultural approbation and RACISMMMMMMMMReport
Again, if I look, I just know I will find silly liberal brain droppings posted on popular left websites that get considerable funding (ETA: I can start at HuffPo – the crap that shows up there at times…).
Crap like the stuff at your link annoys me, but it’s a symptom of the digital age, when content must be produced on a regular basis in order to keep people clicking on posts and proving the site relevant (and generating ad revenue). Thus writers can publish all manner of silliness and still get paid.Report
Its generally easier to identify drivel from other sides than your own because the drivel of the other side confirms your priors while drivel from your side does not.Report
Exactly. I ignore drivel that seems to come from my corner unless it’s like woefully bad*, because there is always drivel. I really only perk up if said drivel is trying to turn that drivel into policy goals.
*The FEE article isn’t entirely off base either. As economic conditions improve, the need for high numbers of offspring declines. No need for large families to work farms, or small businesses, or in factories (pre-child labor laws). As advanced education becomes more common, families are started later, so getting more than 2 kids becomes tricky as fertility declines with age. Etc. The posts cause is ideologically leaden, but the effect is well documented and known.Report
The FEE article isn’t wrong but I’m reminded of Justice Holmes dissent in Hammer v. Dagenhart when he wrote “But if there is any manner upon which civilized countries have agreed – it is the evil of premature and excessive child labor.” Some things are left better unsaid even if technically true. FEE might be right on the economic reasons why people have fewer children but pointing that out makes you look like terrible.Report
One of the points the FEE article makes is that we seem to have decided that the definition of ‘excessive’ here is ‘any’, or functionally close enough to it to not matter.
However, that strikes me as besides the point. The population of parents having fewer kids are the middle and upper classes, i.e. the SES that can best afford to not have kids working, or rather, the SES where the meager financial inputs of child labor to the household adds so little as to be noise in the budget.Report
Its generally easier to identify drivel from other sides than your own because the drivel of the other side confirms your priors while drivel from your side does not.
Huh? The reason partisans don’t notice drivel from their own side is precisely because if confirms priors. The argument goes both ways, and obviously so, at least if the term “drivel” can be defined in non-partisan terms. Seems to me that facts, evidence, and sound argument plays a role in this.Report
But this isn’t HuffPo where any idiot can sign up for a blog – I’d say, find me insane off the walls stuff like this on think tanks the Brookings Institution or the Center for American Progress or hell, even the Heritage Foundation.Report
Re: Downtown highways. It’s the same circular argument every time:
1. This neighborhood is terrible! Ever since the highway cut us off from our neighbors and belched carcinogens into our lungs, things have gone downhill. People have moved out and those who remain are plagued by crime and decay!
2. Should we get rid of the highway and replace it with a park. It will re-connect the neighborhood and make it more livable?
3. Heck no! If this is a place people actually want to live in, those of us who can only afford to live in neighborhoods nobody else wants to live in will be priced out.
4. So you’re saying that this neighborhood must remain dreadful to protect the dwindling population of people who live here?
5. No, we’re getting poisoned by the pollution and isolated from resources because of the highway!
6. So we should tear down the highway and make things nicer?Report