I'm really mad at several college professors in the CS department at my undergrad who told us all "self driving cars are coming" and trusting them as authorities I quashed my many qualms and wasted a lot of time in otherwise important conversations about urban planning insisting that we account for the impact of this 'inevitable' variable.
"They will be safer than human drivers." tapped into my confirmation bias ... I hate drivers, after all. Of course every one could be bested by a machine. 1/
I should have trusted what I'd learned from first-hand experience programming & writing programs that processed images & that made decisions.
I should have noticed the parallels to the 'work' (self-perpetuating failure cycle) of the RAND corporation & compstat.
I knew the problem of making a car drive was hard. Messy. But, I dismissed this knowledge from my experience because authoritative people said "the smart boys in the valley have this all worked out"
@futurebird Well I work for a self-driving car company and while I hate cars and will probably end up moving to the Netherlands to get away from them, and primarily use my electric car go bike, I do think the numbers already speak to there being significantly superior to human drivers. they always stop at stop signs, they respect speed limits, they recognize painted bike lanes etc.
@cshentrup@futurebird "They respect speed limits"?? What kind of propaganda is this? At least one manufacturer has been caught explicitly programming their cars to speed and drive extremely aggressively because their customers wouldn't like the self driving feature if it didn't do that.
@cshentrup@futurebird Aren't cars predominately a USA problem?
Sure, cars are causing a lot of problems all over the world, no dilemma, but solving issues through more, hugely expensive self-driving cars is a very USA-style of rolling over the rest of the world indifferently when most of the problems could be solved by a good public transport system.
In short, why can we all just invest all that self-driving money in public transport, instead?
@elkarrde@futurebird oh good god no! The only country I've seen do a decent job is the Netherlands. Paris is also starting to get pretty good with bike lanes. they are dominant in most of the new world and a lot of Europe too. I mean visit Mexico or Canada.
@cshentrup@futurebird Yes, because it's usually the most convenient way of getting from point A to point B, outside of big cities - because the public transport infrastructure is not as developed as it should be. For example, Istanbul does a good job of public transport. It's been a while since I've been there, but I assume not much has changed.
Also, a lot of Eastern Europe drive cars worth less than 7k€ when bought, how do self-driving (and EVs) fit that bill?
@cshentrup@futurebird Okay, then how do I get from my village to my grandparent's village on the other side of the country in the mountains, with a lot of luggage? You can't charge an EV there.
Also, how do I get to a week of travelling all over the mountains with a lot of camera gear without owning a car?
Unfortunately, neither problem can be solved by public transport, as well.
@elkarrde@cshentrup@futurebird I don’t mean to be glib, but in the first case maybe take less luggage, or don’t go as frequently, and in the second take less equipment or don’t go. Human being lived for centuries without cars, I see no reason why it would be impossible to return (gradually) to living without them again.
@VirginiaHolloway@cshentrup@futurebird No worries, it's all for the discussion. Yeah, solutions exist, but my point is, cars as a form of transportation are not the problem, but over reliance on cars is. Personally, I've been in a city with my car twice this year, I've been taking a tram or a bus otherwise, my car was used only for trips to the seaside and my parents in the countryside. Could I have used a bus or train? Yes please, if that works as it should, but that's a different topic.
@elkarrde@VirginiaHolloway@futurebird That's not quite right. Even if you have alternative options, and aren't dependent on a car, you might be overly incentivized to use a car because we don't tax the negative externalities they create, and we subsidize them by providing free or discounted parking, and freeways instead of tollways.
@VirginiaHolloway@elkarrde@futurebird amen. not only will EV chargers be on the present soon, if people only use private vehicles for these rare cases, we're talking about 99.99%
@cshentrup@VirginiaHolloway@futurebird I really do hope that plan will work out wherever you live, but I have no illusions it'll be attainable in the next 30-40 years over here in my part of the world. 🤷♂️
On the other hand, public transport wouldn't need significantly more financial investment, just a bit more political will and less short-sightedness (at least here).
@cshentrup@VirginiaHolloway@futurebird Yeah, if they're running at all. In many places, infrastructure is already unable to sustain more than 2-3 EVs charging slowly. Local governments don't see an incentive to invest into something that needs a major infrastructure overhaul, and EVs are significantly too expensive for the majority, so it's a chicken-egg problem.
At least, I hope the next block of the nearby nuclear power station will get a green light soon, for the future needs.
@cshentrup@VirginiaHolloway@futurebird Our government approved the expansion, but the other one owning the plant said they'll need a referendum, so let's hope it'll get greenlighted.
@cshentrup@VirginiaHolloway@futurebird Lucky you, no driveways here, just a public parking. I'd need to throw a cable from the 4th floor to a car possibly parked a few hundred meters away. 😅
@cshentrup@VirginiaHolloway@futurebird Never been to Japan, dunno, honestly. I'd prefer underground garages with charging stations, but realistically that's just a pipe-dream.
This is where I think plug-in hybrids (PHEVs) get completely overlooked. Most people can do most of their driving on a 30 mile battery that charges on an ordinary 110V feed overnight. And then if you need to do longer driving, the gas engine is there. 80% reduction is gas usage with 10% of the battery minerals of a Tesla and more flexibility.
@MichaelTBacon@cshentrup Hybrids are pretty popular over here, yes - 220V charging over the night, for those who can, city driving is done on the battery, and the petrol engine kicks in. It's far from ideal, but it works.
Range-extender hybrids are a good solution, also, where the petrol engine (or any other engine, fwiw) has no mechanical connection to the wheels and is used just for producing electricity. Unfortunately, all available models got discontinued (Opel Ampera and BMW i3).
Yeah, it's annoying that they're not more available, but there's still a fair few models available in the US. Our 2021 Pacifica van is a true range-extender hybrid. The gas engine sits quite unless you need a boost of power or until you've exhausted the battery, and then it's exclusively supplying juice to the electric motor.
@leadore@cshentrup@futurebird Bad choice of words, it's 5 in the morning; I was trying to refer to an absolute reliance on cars in the USA, and strictly limited public transport options. 🤷♂️
@mastodonmigration@futurebird why don't you give away all your money until you're as poor as the poorest human? I mean if we're going to talk about charity.
The problem with the technocrat solution is that it always starts with "technology will save us", often new technology. For example, I've yet to see a driverless car solution which is better than a mass transit solution -- and we already have working mass transit solutions.
So we pour billions into researching driverless cars, when we could start building better mass transit now.
@javierg@futurebird@mastodonmigration@eyrea@cshentrup I wish technology and brain cells could be applied to innovation planning solutions (in addition to mass transit) that could lesson the number long distance commutes so many people need to take. Does going from a 2 hour round trip commute in a gas-powered car to using an EV / self-driving car really address the bigger picture?
@SusanHR@eyrea@javierg@mastodonmigration@futurebird that guy who understands nothing about economics from mastodon.social again? validating my decision to block that entire server. when some thread i see with lots of people arguing with someone with demonstrably no ability to learn from a debate, 85% it’s some mastodon.social techbro in the dunce corner
@mastodonmigration@cshentrup@futurebird it's almost like this guy named neoliberal is a technocratic asshole? Who would have guessed, with a name like that.
@futurebird@mastodonmigration it's not about access to information. it's about being smart enough and determined enough to learn the right answers. most elected officials, people who have to make policy decisions for a living, don't even know basic economic terminology like hypothecation or dead weight loss. so of course the lay public won't know these terms. of course we want experts running things.
I agree that you want experts, but selecting leaders on the basis of expertise alone implies a poverty of general technical knowledge such that the most expert people aren't doing research or formulating new theories, but rather working the government.
It also takes the public out of the process of selecting leadership so it is anti-democratic.
@futurebird@mastodonmigration No it doesn't take the public out of the process. the public is necessary for alignment. to improve aim you want deliberative processes like election by jury.
@futurebird did you see the part where he elides education and access to information? In the context of his performance last time this is just bad faith.
@cshentrup@futurebird Speed limits are set according to human reaction times and psychology, though, in those instances where they are even set for safety at all (as opposed to noise control or flow management).
@cshentrup@futurebird there are so so many different things that lead into setting speed limits, and the percentile method and the NHTSA standards are just a couple of them. But these remain fundamentally human-centered methods.
@futurebird
I don't trust self-driving cars because they were programmed by people who have obviously never, ever watched one single Terminator movie. Or Maximum Overdrive.
@mzedp@futurebird May I ask if you have a favorite reading resource on why you advise against it? I had completely forgotten about that hype, curious if the hype abatement has just been cost-related, or if there is something more fundamental to why it isn't viewed as a viable solution?
@janeadams@futurebird Basically, if energy was free and you could produce H2 cheaply enough, it might make sense.
But neither of those ifs is realistic.
And then, when you actually start looking at who's pushing this H2 stuff... it sort of becomes apparent that this is just the fossil fuels industries dying hope of there being some miracle clean fuel they can substitute oil for to preserve their business model and keep their dominating socioeconomic position.
@mzedp@janeadams@futurebird Renewable energy is basically free these days. The problem is being able to store it. Hydrogen is already cheap enough, and will get cheaper.
In reality, you are just a Luddite spamming conspiracy theories. You are totally stuck in the past.
@mzedp@futurebird You've been brainwashed by BEV propaganda. In reality, fuel cells are also electrochemical systems. You are the same kind of guy as those who writes elaborate diatribes against new technology. Something that has happened many times in the past.
@futurebird I’m certain self self strong cars are coming but people have been way too aggressive in their timelines. It’s clearly doable but it’s also clearly a lot of very hard work. I’m waiting for them to deal with all the fun stuff. Snow storms, surprise deluge, and horse in the road.
The professors that I'm mad at were saying it was 5 years away 20 years ago.
We may see some limited applications in trucking, and on specially approved roads. (which would become pedestrian free spaces, I suspect, and this has ... implications)
But, I don't expect to see a car parking itself in a grocery store parking lot any time soon at all.
We'll have human drivers running the cars via the internet (so you can outsource the driving) first. (combined with a computer system )
@futurebird@norgralin
to me, the "self-driving" trucks on specially approved roads thing is yet another example of tech re-inventing rail, and doing a rather bad job of it.
@futurebird@norgralin 20 years ago 5 years was ludicrously optimistic for #selfdriving cars given the precursor to #waymo only started in 2009.
Even now, the tech is still not quite there and we are nowhere near significant wide scale adoption in any road transport sector.
I still think it will happen. I don't think it's worth doing urban planning for because it is far off and should be viewed as a complementary technology to existing mass transit options that have more explicit planning reqts.
@futurebird@norgralin I was 100% convinced that if nothing else long-haul trucking was going to be over as a human-employing thing, but Waymo threw in the towel on it this year, which genuinely shocked me. If they can't make it work, don't hold your breath on anyone else in the next decade or two.
(That said, auto-parking is one of the thing that actually did come out of this -- a lot of higher-end cars will auto-parallel-park themselves. It's neat!)
@futurebird having self driving cars and EVs fix all the problems with North American cities was so tempting for folks. It was easy to believe because it would be so simple compared to what is actually needed (mass transit, denser neighborhoods, more walking & biking, less car dependency, etc).
It’s unfortunate that folks in power are still holding out, but so important for people to keep pushing for the bigger solutions in the meantime
@futurebird Self-driving cars are absolutely possible, but they will have to carry so many sensors and cameras and computing power that they will cost millions. And they'd have to be networked, sharing sensory data and getting additional intelligence from some computing centre where some huge AI model goes through all the traffic data in real-time. Not very efficient, I think. Besides, since in many cities parking is more expensive than fuel or electricity, they'll just drive around all empty until they can pick up a passenger.
@futurebird the frenzied high of tech optimism in the 2010s affected nearly everyone, myself included. We’ve all been getting a (probably much needed) reality check
@futurebird Amen. I keep having this with LLMs. I know what they are, I understand the limitations, but I nonetheless get sucked in by the enthusiasm of “really smart people.” Two hours of wasted time later, I find myself saying “is that all there is?”
@futurebird In the mid-2010s, car cos presented their plans & outlook for autonomous cars at an econ conference.
Their interest in selling to transit systems/local govt (last mile support), rental car cos & ride-share cos + high prices seemed to mean that the ownership model was not intended for individuals. Sure, wealthy folks will buy their toys, but most people won't own them.
I expect autonomous cars to be like infrastructure & haven't seen a reason to change my view on this since.
@futurebird I made a questionable car buying decision under the assumption that self-driving cars would be here imminently; now I'm wondering if they'll be here in time for my retirement (because obviously the USA will still be car-first by then 🤢🤮)
@isotopp@futurebird Peter Norton, Author of #Autonorama will be the keynote speaker at the Winter Passenger Rail Gathering on December 9 in Centralia, WA and online. Hosted by @summit@aawa and @aanw Details and registration at https://gnwprs.org We hope you can join us!
@futurebird
I expect the best use case will be driver and machine learning working together. People are fine drivers but have trouble with attention. Computers are good with attention but can't do interpretation as well as people. Together I expect we can make things safer. Heck, the alarms on my fil's new car are already keeping him and everyone around him safer...
@futurebird FWIW I know a few of the ppl who were involved in the earliest skunkworks project inside google that turned into Waymo, and I think this was a classic case of being so hyped that you've solved the (legitimately really difficult!) first 90% of the problem that you completely discount how hard the second 90% is gonna be. Paved with good intentions and all that.
Didn't help that Mountain View is probably the easiest place on planet earth for prototyping this stuff.
The STS reading I was doing before my PhD program kept me from biting on it otherwise I probably would have. I hate drivers too, but I was learning that complex tech systems, like our roads, are complex social networks of humans and non-humans. Replicating the human role in that was going to be very hard, and some prior attempts had already failed along cybernetics lines.
Musk's awful decision to pull LIDAR was the final straw. I went from "I dunno" to "this will absolutely fail."
Regardless of the driving tech, it will be a very long time (possibly never) before a fleet of driverless cars is anything other than a colossal waste of resources and cause of pollution compared to a functional mass transit system.
@futurebird I used to do research in an adjacent field, and in 2019 I went to a conference about self driving cars. Talking to the grad students and professors in this space, I was struck by the fact that none of them expected level 5 self driving cars. They were surprised when I would bring it up. It was like they were ignorant of the rhetoric going on in the public sphere.
@futurebird I’m laughing about this because I just went to a conference on self driving vehicles in June… we’re 90% of the way there, which means we have 90% more of the time and money left to spend on this project.
Someone write a ballad about a contest between driver John Henry and a self driving car to see who can crash more. Ends with John Henry triumphantly crashing 100 times and dying with the steering wheel in his hand while the self driving car is still wobbling from its 99th crash.
@futurebird I mean, they're not wrong. But there's the matter of time scales. There is a long long history of overconfidence in the speed of development of technology -- especially AI. Here it is, two decades after the projected time-line, and we still don't have a glitchy talking computer on the way to Jupiter. But we do at least have rudimentary talking computers on our Star Trek communicators.
@futurebird Everyone who comes up with this:
If its so easy, why aren't trains, that move on controlled tracks and planned with timetables, self driving since years?
Saudi Arabia is trying to create a monopoly of transportation, A LOT of people are influencing for them. Billions of dollars lost, but if they destroy transit by bribing city councillors, form monopolies of transportation itll be fine. Fine means firm control is important.
The arguments I had with people about self driving cars just 5-10 years ago.
The whole thing kind of makes me mad because it's so much time and energy on something that doesn't really solve any major problem, it just seems like it would be neat to some people.
Besides the privacy and freedom arguments that rarely seem to come up.
You really think they're just going to let you go anywhere you want any time you want?
@futurebird Don't be too mad at them. At some point most of us believed the guy was an actual genius who will change the world in no time. It felt great to have another nerd at the top, "one of us". We learned our lesson and are collectively much more critical, pragmatic, realistic.
@futurebird The "they are safer than humans" argument is a prime example of cherry picking a limited set of criteria to extrapolate a fallacious conclusion. It is easy to cite several ways in which these robot cars work well so long as they are functioning properly. To conclude this makes them "safer" ignores all the edge conditions that actually make them death traps as this week's blocking of emergency vehicles demonstrates.
@futurebird Another fallacy is the "not their fault" argument. Human drivers avoid accidents all the time that would not be their fault. Take one case where there are 100 accidents, all not the robot car's fault, and another case where there are 10 accidents with 5 the being the driver's fault 5 not their fault and 90 avoided. Which is safer?
Bottom line, figures don't lie, but liars sure can figure. And these guys are big fat liars.
@futurebird One more. Apparently these robot cars don't work if they are disconnected from the internet. This is literally a Star Wars episode where all the droids are disabled by one strategic strike. We saw what happened when 2 robot cars blocked traffic. What happens when 1000 go offline all over the city at the same moment?
This is dangerous and irresponsible and the lunatics perpetrating this on us should be held personally liable for any death or injury.
@mastodonmigration@futurebird Another reason that robots can’t drive better than people is because the entire car-going infrastructure is designed with human drivers in mind, not automata. If an engineer were tasked with designing roads that were robot-friendly, the rules, markers, interface with pedestrians, and other standards would be very different (see: rail).
Because there is a heavy reliance on human cognition (and to a large extent, the complex rules of local culture) to navigate the infrastructure, it is very difficult to exceed the performance of a human driver. In many cases, human drivers must simulate what happens inside other human drivers’ minds as they negotiate the driving environment together; there is no way a machine can begin to do that today.
tl;dr: we made roads for human drivers, not robots. If we want robots to exceed humans, we must make roads made for robots.
@drahardja@futurebird Absolutely, all the statistics they are gathering are collected on very prescribed sections of roadway during favorable conditions. Toss in inclimate weather, construction and other common anomalies and the house of cards crumbles.
@futurebird
Early on in the research for it, I had a friend who was working with the most sophisticated tools available to Harvard, and over and over, he kept showing that a car cannot tell the difference between a child and a balloon, and given the choice between destroying one, saves the balloon 60% of the time. I am absolutely gobsmacked that they're just out there ruining everything right now and no one is stopping it.
@futurebird do they have to take the state drivers’ test? And pass the written exam? Parallel park between cones while the examiner makes hostile notes? Distinguish between signs that are uncommon and out of context? Yield to emergency vehicles? If not, why not?
@futurebird Whenever self-driving cars get mentioned in the context of urban planning, I start internally screaming "JUST BUILD TRAINS WE'VE HAD THEM FOR A CENTURY AND THEY'RE GOOD also more bikelanes PLZ" but this screaming is internal so no one else can hear it but me.
Been thinking the importance of knowing when we've been tricked or taken in and (this is the harder part) the importance of talking about it.
Imagine if every victim of a scam, every person who was deluded by bad ideology was open about what they learned? If we just all decide that if you can realize you were tricked that's almost as good as never having been tricked at all.
Everyone falls for something at some point. Changing the dynamics of shame can inoculate the future.
Some scammers use trust by association with victim A to get in with victim B and so on, in a chain deception.
By the time a friend group has been well and truly scammed, they finally compare notes and everyone ends up feeling awful about getting their friends scammed.
@futurebird@unicorndeburgh ngl though, it's hard even for those of us who at least try to make an effort to be more open minded in general (and I won't lie and say I don't fail sometimes!) to admit we were wrong sometimes -- especially if someone completely hoodwinked us, making us feel even dumber. I can understand why people as a whole just double down rather than admit it. But it's sad and harmful to the whole human race as you so eloquently pointed out.
Related to that: I saw an article a while ago about how not wanting to be seen as a sucker can influence people's behavior. That can both lead people to believe in rumors or conspiracy theories, and also prevent us from dropping those beliefs later.
Adding shame on top of that would make it even harder to admit when we're wrong.
@futurebird@unicorndeburgh
This (openness plus self-monitoring plus non-blame culture) is how aviation tries to work. It's a reaction to the tension between commercial pressure and the need for safety.
@kentindell@unicorndeburgh@futurebird I wish I wrote down my thoughts about the possible futures of self-driving cars from around a decade ago. I thought there would be 5 phases, each of which would take between 10 and 50 years, but I don’t remember what exactly the 5 phases were. But we’re still in the first phase, limited experimental use, and that old guess is looking optimistic
Add comment