gosh, if only there was some kind of public system that could transport many people at once for a low cost that all those billions could have gone to.
Once again, people leaping at opportunities to apply tech to things because technology is a hammer and everything is a nail. Meanwhile all the horrible and unsexy things in life persist because bay area tech bros aren't forced to deal with them everyday.
If people in the US wanted to bicycle and support/use public transport, they would. They don't.
I wish they did. But they don't. You are welcome to try to convince them otherwise. But I'm glad people are also working on the problems we do have as they are rather than waiting to convince people of solutions they do not seem interested in.
Yeah, well, people want housing to be an investment and we've all seen where that got us. People want hour long showers and 1L big gulps of soda and NFTs. The will of the people doesn't count for squat, people are idiots.
I agree with your comment, there are lots of folks making real progress in this space and there are secondary industries that benefit from things like lidar which has gotten way better in large part to this work. Playing devils advocate though - Theranos had 800 employees at its peak, it's hard to think that at least a few of those people working there genuinely thought they were trying to solve the problem.
Standard commercial unit, no clue. It happens as you say, with sun shining directly in front, especially when passing under bridges or when another vehicle passes the sun (but not in your lane). The stark difference of direct sunlight and long shadows triggers it the most.
“Jerry, just remember. It’s not a lie… If you believe it.”
When a company asserts they will have self-driving cars in a year, and they aren't doing any on-road tests, is that assertion a deliberate deception? Or a willing self-deception?
In any case, people can work be part of a scam without knowing it's a scam. Do you think everyone at Bernard L. Madoff Investment Securities knew it was a big Ponzi scheme?
Tesla has been collecting money for a non-existent upcoming capability for how long now?
Or did they just redefine the meaning of “full self driving” to mean something other than the car fully driving itself?
Then there is Uber. You could read HN archival posts where significant numbers of folks here drank the kool-aid and waved away the obvious holes in the Uber business model with “of course this is just temporary… robot cars will be doing everything in two years”. Listen to a Ben Thompson podcast when he devotes dozens of hours fawning over Uber.
Full self driving is tantalizing for unethical business people because it demos so well. High minded talk about the trolley car problem and liddites shut down meaningful criticism.
The engineers may be collecting big paychecks working on it, but they are hardly innocent; they are no different than employees of any other unethical corporation.
I sense deflection and an attempt to change the point being made above your response.
It’s particularly unethical to promise a badly thought out goal that cannot be meaningfully achieved in the interests of drawing out financial support for a longer period of time.
I am reminded of the person who called their intimate partner “tomorrow” because they never came.
Aw, thanks for taking the time to write a personal comment. I'll try and be less impulsive next time and make a more substantial, less flamebaity contribution.
The technology has come a long way since the early days of the DARPA grand challenge. Fortunately, the technology will continue to progress, independent of your ability to comprehend it.
Your response is the playbook for these types of scams. Don’t believe your lying eyes, everything’s great.
Perhaps I’m simple, but I’m pretty sure the people who dropped $10k on full self driving “in a year” or “in the future” in their Tesla model S in 2017 got fleeced. Tesla sold access to a product that doesn’t exist, and never delivered it. Uber sold a business model to stupid investors that doesn’t exist to justify breaking other laws.
That doesn’t mean that the long running Google work in the field, etc is a scam. That doesn’t mean that working on self driving is unethical. Ripping off people is.
There is room for fraud within a scope of offering undiscovered technologies, but the undiscovered technologies themselves are not fraud. Theranos isn’t comparable to self-driving cars.
So imagine I sell a future-looking feature which is incompete and unreliable, even causing an order of magnitude more fatalities per usage (e.g. mile) than the alternative. And continue to sell and offer it under a confusing name for years. What would have to happen for it to cross over from naive aspiration to fraud. I.e. for Tesla FSD to become like Theranos?
For me that line was crossed when Tesla offered FSD on public roads, to the general public, and continuing to include 'full' and "self driving" in the name. Every day it continues more non-Tesla owners are put at risk while the government is asleep at the wheel.
I raise Tesla as an example of a scam similar to Thermos in that it's crossed from naivete to fraud. And considering the stuttering progress of self driving by Uber and even Waymo (in adverse conditions) it may well be self driving as a concept is premature. So attempts to market and sell it as is (beta or not) are fraud.
There was a time when many within the AI community thought that every problem could be solved using neural nets, tuning and lots of training. Elon truly believed them, and believed himself that if you trained a large enough network on trip recordings and/or sensor data, the network would magically come up with correct driving inputs. It seemed to work well enough on closed courses, so everybody, including Elon, thought that within a year's time we'd have figured out the kinks. That's when he made the mistake of promising FSD. Turned out, what worked in one city/country didn't work in another. And then we had inconsistent road markings, missing lane dividers, lanes going into walls, pigeons... So they had to get back to more 'conventional' machine vision combined with some neural to recognize objects. It turned out driving is a hard problem that humans happen to be very, very good at. And here we are, many years later, and the problem isn't close to being solved - and not for lack of trying.
IMHO finding a solution to self-driving isn't so far from finding a way to scale intelligence enough to reach AGI. I'm sure we'll get to self-driving before we get to AGI, but solving self-driving will be a major stepping stone towards AGI. Hence Elon is now aiming square at it:
https://www.nextbigfuture.com/2022/01/elon-musk-must-lead-ne...
Last night I was thinking of road markings. Or sings really. Which at high level are somewhat unified across certain areas. But then you have fun list of regional variety or additional details that really do matter, but throw entirely new wrench in the works.
Like simple sign of banning lorries and vans. But then having a text under it allowing vans... Not to even mention the hell that is parking and stopping prohibited signs.
While I don't own a Tesla nor have I ever even been in one, I've seen plenty of videos of Tesla's full self drive in action, and it is doing incredibly well!
Driving is not a 1+1 operation. Real life is messy and their system seems to handle that mess pretty damn well for many many scenarios.
I feel like people attack full self driving as if it's supposed to be better than humans on day 1, meanwhile the definition of better is impossible to agree upon.
>That's different from a human driver how exactly?
We have many decades of experience with human drivers and generally understand their shortcomings and how/when they fail.
Self-driving systems tend to fail due to weird edge cases of their systems that humans can't perceive. Tesla's FSD will navigate tight, winding neighborhood roads with ease then it'll try to steer you into a construction zone on a mostly empty interstate. Weeks later they'll push an update that handles the interstate well but regresses in the neighborhood.
That is to say, we have many decades of experience with human drivers running into walls and humans, and have largely given up on the problem and instead focus on making the car crumple instead of the thing it hits.
Human drivers have known weaknesses and known workarounds. Have a super dangerous stretch of road? Add loads and loads of bright orange signs for miles before with flashing lights. Still doesn't work? Close the road and redesign it according to what humans can reasonable handle.
AIs are still so raw those weaknesses change on a regular basis and so opaque that no one is even trying to develop workarounds.
Wouldn't it be better to get traffic directing AI for local spots?
It could specialize for a specific intersection, and assist cars passing through it.
Scam or not, I've said this before and I'll say it again: self driving is a side show.
Electrification? Yes, important. Getting fewer people to own and use cars altogether? Totally. But is self driving going to get us there? Think about it, all the reasons that people cite for wanting to own a car today will still be there. I don't see many people going "well now that it drives itself I won't need it in my driveway."
Especially because for some fucked-up reason, the connection to "my car" is much more emotional than it should be. I'm from Germany and the attachment people have to these things is insane. There's actually some interesting psychoanalytical work about this.
Yeah it's like a parasite how much attachment there is to cars
People have lots of money and passion sunk into them
Children's die cast cars, children's play mats that have road networks on, city builder games that are just car centric, car TV shows, children's movies
There's lots of things in our cultural fabric that make it hard to imagine anything else
I suppose it's quite similar to the promotion of guns, but here in the UK we heavily restrict access to guns
A lot of people on this site will spend hours and thousands of dollars perfecting their desktop configuration or their mechanical
keyboard setup because it’s a token of their pride and joy and a manifestation of their livelihood. I’ve met people with the same level of love and dedication for their cars, it’s just an incredibly important part of life in North America.
And there's a good debate to be had about the negative externalities of IT equipment, and short product release cycles!
Personally I've taken to buying more second hand, and making sure my systems are upgradeable. But still, silicon and other IT stuff has insane footprints wrt water, CO2, and other pollutants...
That being said, we can look at both. My point here was just that self-driving will not magically make the world better. It will, if it ever works, make cars more convenient for the people inside. But it will not address the big problems with cars, which are all about the people outside and the environment.
>the connection to "my car" is much more emotional than it should be.
Who is anyone to say how anyone "should" feel? Or "should" do anything? If you had said "more emotional than it could be" you'd have a point.
Cars provide people the ability to travel almost anywhere whenever they want. They also provide shelter and become personalized spaces. I think it is natural to have an attachment to something with those kinds of roles. They speak to very primal instincts and needs.
I have a pretty strong attachment to one of my bicycles. Irrational? Sure. Human? You bet. The same goes for my best (for me) trumpet.
When the economics of not owning a personal car make sense, they will become less important and there will be less attachment to them in this way, like folks in NYC have towards cabs.
> I have a pretty strong attachment to one of my bicycles
Hey there fellow bike lover! Mine is a 1979 Peugeot racer and I bought it with literally my last money for the month as a student just because it was so freakin' pretty! :)
Yes you have a point, it's difficult to judge what emotional attachment is "good" or "bad". But in a pretty objective way, cars have a lot of negative effects that e.g. bikes just don't.
I also think it's kind of sad that we haven't found a way for people to feel physically safe without encasing themselves in literally a metric ton of metal. Sadder still that for many, cars provide shelter. Those roles could and should be better filled by other means, and would require less subsidy from society than car infrastructure demands. Plus, it would just be the decent thing to do.
Besides, for many people (I'm seeing this in my own family at the moment) the idea that cars are cool and interesting is injected constantly, over and over again, everyday, it's really hard to do anything against. For starters if anyone knows good "matchbox" style bicycle toys, let me know, because I think I've exhausted the market after just two or three birthday gifts.
Wow. Me too. I had a 1980 or 81 Peugeot Super Vitus racer. So gorgeous. Regardless of functionality (and it was great), it was worth it just as industrial/mechanical art.
I'm totally with you regarding your opinion on cars and how arbitrary our attachment to using them is in many situations. But after bicycle commuting for almost 25 years and seeing how uninterested people are, even "progressive" childless 20-somethings in a very walkable, rideable, progressive town with a good climate, I'm more apt to meet people where they are at and fix the problem we have (emissions from burning gasoline) rather than try to force people to redefine the problem and choose the solutions I prefer.
Jim Keller seems to think self-driving is a solvable problem. I find him more credible than Levandowski and an article that is written in a way that is not even-handed or objective at all. It starts with a premise and tries to convince rather than weighing evidence in a way intended to be objective. The article literally starts with anecdotal evidence, as if that's relevant. It's wording and construction makes its intent clear; it's an emotional appeal.
I could be convinced that working on self-driving isn't worth it with a proper weighing of evidence. But this article ain't it.
Not a scam. A harder problem than originally thought. Also there are several companies working on this. So they’re in on the scam together? What a joke of an article.
How about trying to be a little more honest about it and people might listen, or forget that this article isn’t completely biased against the tech.
64 comments
[ 3.1 ms ] story [ 139 ms ] threadOnce again, people leaping at opportunities to apply tech to things because technology is a hammer and everything is a nail. Meanwhile all the horrible and unsexy things in life persist because bay area tech bros aren't forced to deal with them everyday.
When a company asserts they will have self-driving cars in a year, and they aren't doing any on-road tests, is that assertion a deliberate deception? Or a willing self-deception?
In any case, people can work be part of a scam without knowing it's a scam. Do you think everyone at Bernard L. Madoff Investment Securities knew it was a big Ponzi scheme?
So you see, it's true. From a certain point of view.
Or did they just redefine the meaning of “full self driving” to mean something other than the car fully driving itself?
Then there is Uber. You could read HN archival posts where significant numbers of folks here drank the kool-aid and waved away the obvious holes in the Uber business model with “of course this is just temporary… robot cars will be doing everything in two years”. Listen to a Ben Thompson podcast when he devotes dozens of hours fawning over Uber.
Full self driving is tantalizing for unethical business people because it demos so well. High minded talk about the trolley car problem and liddites shut down meaningful criticism.
The engineers may be collecting big paychecks working on it, but they are hardly innocent; they are no different than employees of any other unethical corporation.
I am reminded of the person who called their intimate partner “tomorrow” because they never came.
You slaver.
You may not owe capitalism better but you owe this community better if you're participating in it.
https://news.ycombinator.com/newsguidelines.html
Perhaps I’m simple, but I’m pretty sure the people who dropped $10k on full self driving “in a year” or “in the future” in their Tesla model S in 2017 got fleeced. Tesla sold access to a product that doesn’t exist, and never delivered it. Uber sold a business model to stupid investors that doesn’t exist to justify breaking other laws.
That doesn’t mean that the long running Google work in the field, etc is a scam. That doesn’t mean that working on self driving is unethical. Ripping off people is.
> That doesn’t mean that the long running Google work in the field, etc is a scam. That doesn’t mean that working on self driving is unethical.
Seems like you're still grappling with your own stance on the issue.
For me that line was crossed when Tesla offered FSD on public roads, to the general public, and continuing to include 'full' and "self driving" in the name. Every day it continues more non-Tesla owners are put at risk while the government is asleep at the wheel.
Of course a counter example would be just as relevant since it could prove the concept isn't entirely a scam.
Not a scam, juat humans being human.
IMHO finding a solution to self-driving isn't so far from finding a way to scale intelligence enough to reach AGI. I'm sure we'll get to self-driving before we get to AGI, but solving self-driving will be a major stepping stone towards AGI. Hence Elon is now aiming square at it: https://www.nextbigfuture.com/2022/01/elon-musk-must-lead-ne...
*edit for typos.
Like simple sign of banning lorries and vans. But then having a text under it allowing vans... Not to even mention the hell that is parking and stopping prohibited signs.
There are just so many edge cases with driving we take for granted and little standardization for local areas.
Driving is not a 1+1 operation. Real life is messy and their system seems to handle that mess pretty damn well for many many scenarios.
I feel like people attack full self driving as if it's supposed to be better than humans on day 1, meanwhile the definition of better is impossible to agree upon.
We have many decades of experience with human drivers and generally understand their shortcomings and how/when they fail.
Self-driving systems tend to fail due to weird edge cases of their systems that humans can't perceive. Tesla's FSD will navigate tight, winding neighborhood roads with ease then it'll try to steer you into a construction zone on a mostly empty interstate. Weeks later they'll push an update that handles the interstate well but regresses in the neighborhood.
That's a separate issue you're conflating.
Human drivers have known weaknesses and known workarounds. Have a super dangerous stretch of road? Add loads and loads of bright orange signs for miles before with flashing lights. Still doesn't work? Close the road and redesign it according to what humans can reasonable handle.
AIs are still so raw those weaknesses change on a regular basis and so opaque that no one is even trying to develop workarounds.
Electrification? Yes, important. Getting fewer people to own and use cars altogether? Totally. But is self driving going to get us there? Think about it, all the reasons that people cite for wanting to own a car today will still be there. I don't see many people going "well now that it drives itself I won't need it in my driveway."
Especially because for some fucked-up reason, the connection to "my car" is much more emotional than it should be. I'm from Germany and the attachment people have to these things is insane. There's actually some interesting psychoanalytical work about this.
People have lots of money and passion sunk into them
Children's die cast cars, children's play mats that have road networks on, city builder games that are just car centric, car TV shows, children's movies
There's lots of things in our cultural fabric that make it hard to imagine anything else
I suppose it's quite similar to the promotion of guns, but here in the UK we heavily restrict access to guns
Personally I've taken to buying more second hand, and making sure my systems are upgradeable. But still, silicon and other IT stuff has insane footprints wrt water, CO2, and other pollutants...
That being said, we can look at both. My point here was just that self-driving will not magically make the world better. It will, if it ever works, make cars more convenient for the people inside. But it will not address the big problems with cars, which are all about the people outside and the environment.
Who is anyone to say how anyone "should" feel? Or "should" do anything? If you had said "more emotional than it could be" you'd have a point.
Cars provide people the ability to travel almost anywhere whenever they want. They also provide shelter and become personalized spaces. I think it is natural to have an attachment to something with those kinds of roles. They speak to very primal instincts and needs.
I have a pretty strong attachment to one of my bicycles. Irrational? Sure. Human? You bet. The same goes for my best (for me) trumpet.
When the economics of not owning a personal car make sense, they will become less important and there will be less attachment to them in this way, like folks in NYC have towards cabs.
Hey there fellow bike lover! Mine is a 1979 Peugeot racer and I bought it with literally my last money for the month as a student just because it was so freakin' pretty! :)
Yes you have a point, it's difficult to judge what emotional attachment is "good" or "bad". But in a pretty objective way, cars have a lot of negative effects that e.g. bikes just don't.
I also think it's kind of sad that we haven't found a way for people to feel physically safe without encasing themselves in literally a metric ton of metal. Sadder still that for many, cars provide shelter. Those roles could and should be better filled by other means, and would require less subsidy from society than car infrastructure demands. Plus, it would just be the decent thing to do.
Besides, for many people (I'm seeing this in my own family at the moment) the idea that cars are cool and interesting is injected constantly, over and over again, everyday, it's really hard to do anything against. For starters if anyone knows good "matchbox" style bicycle toys, let me know, because I think I've exhausted the market after just two or three birthday gifts.
I'm totally with you regarding your opinion on cars and how arbitrary our attachment to using them is in many situations. But after bicycle commuting for almost 25 years and seeing how uninterested people are, even "progressive" childless 20-somethings in a very walkable, rideable, progressive town with a good climate, I'm more apt to meet people where they are at and fix the problem we have (emissions from burning gasoline) rather than try to force people to redefine the problem and choose the solutions I prefer.
I could be convinced that working on self-driving isn't worth it with a proper weighing of evidence. But this article ain't it.
How about trying to be a little more honest about it and people might listen, or forget that this article isn’t completely biased against the tech.
Even After $100B, Self-Driving Cars Are Going Nowhere - https://news.ycombinator.com/item?id=33106739 - Oct 2022 (107 comments)