> Levandowski was asked why Uber couldn’t just go test its vehicles somewhere else since California regulators clearly aren’t welcoming. He said that Uber’s engineers deserve to see their work displayed in the city where they work.
So it's okay to violate regulations because your engineers want to see their work in their city?
If they thought they were violating regulations, that would be true. But their official stance is that they are not violating regulations, so his point is totally valid.
And neither does the DMV... Which as stated in the previous thread on this, it seems they are exceeding the legislation that covers this. https://news.ycombinator.com/item?id=13181514
> it's okay to violate regulations because your engineers want to see their work in their city?
Other than the pointless flaunting of privilege in this ... I think keeping engineers very close to the project is usually a sign that not all is well with the project.
I read that as "We can't let this project out of sight of the engineers ... they need to be able to make changes on a day to day basis, from direct observation."
That these engineers would be moving in breakneck speed (ha!), to fix whatever they see in person rather than the usual separation of concerns in testing mission critical software.
Through those glasses, it is roughly an admission that it is in "patch everyday" phase of development.
While in general I agree that Uber is being reckless and arrogant, I otherwise really don't get your comment at all. Are you suggesting that quick iterations facilitated by proximity to that which you are testing and direct observation, especially at an early stage of development, is a bad thing??
If so, I've worked on several hardware projects (IC design) and IoT projects as a software developer and I couldn't disagree more strongly. That the engineers are close to the product so they can make quick iterations without massive amounts of waiting and telephone game confusion is a sign that they are doing a good job and don't have an idiot culture of engineering bureaucracy. IMO.
Perhaps that engineering bureaucracy (not sure what you're referring to, to be honest) in the context of safety-critical systems is not "idiot culture".
> Are you suggesting that quick iterations facilitated by proximity to that which you are testing and direct observation, especially at an early stage of development, is a bad thing??
And here we get to the heart of the difference between what is called "software engineering" and what is, in every other field, called "engineering".
There's a reason why Canadian engineers get the iron ring and legend says the first ones were made from the ruins of a bridge that collapsed (due to poor design). There's a reason why people describe aviation as a field where every regulation is written in blood.
Uber is moving into a field where no amount of "don't bring politics into this" can hide the fact that a "bug" in their product will potentially directly kill humans. Requiring people to A) actually be fucking careful and B) actually prove it is how this world works, and I don't think Uber's going to manage to skirt it with their usual regulatory arbitrage technique. The sooner they grow up and start acting like real engineers, the better.
I should have place my commas differently to emphasize that I have worked on many IC design projects as an EE then later on IoT projects as a software engineer.
No one replying to my comment has at all explained why during early iterations when the product is quickly evolving this separation is a good thing. Frankly it seems to me like (a) people are taking pot-shots are Uber and (b) none of you have actually worked on a hardware project.
There is a time when a strict separation between designers and testers is critical. That time is referred to as Qual, or qualification and it is for validation that a part or system operates correctly over all corners of the devices operating parameters. This is not the phase that Uber is in.
Again, I don't dispute at all that Uber are being arrogant and reckless jackasses. I'm just saying it has nothing to do with the fact that their engineers desire to work near to the systems they are designing since that is actually a sign of a smart and high functioning engineering team, not the opposite as people have suggested.
that is actually a sign of a smart and high functioning engineering team
Do you know what auto makers do in this kind of situation? They build a test track and a workshop at the test track and probably a larger facility adjacent to it, and situate the engineers there. They don't say "well, our engineers live in one of the densest urban areas in the country, let's just test our new car there to be close to them".
Uber should get a test track out in the desert or something and relocate some engineers to it. And maybe without the distractions of the SF office they'd manage to get some basic stuff tested, like "recognizes red lights" or "understands how to turn from the rightmost lane".
You still are arguing something I never even mentioned. All I was refuting was the assertion that their desire to be in the same location as that which they are testing is somehow a sign of their engineering disfunction. I have also said numerous times that they should not be doing this testing in dense urban areas in defiance of local laws.
Maybe they should relocate everyone! Total tangential point!
As much as I'm not a fan of Uber, I have to agree with the whole sentiment expressed by Uber in the article, which is that they are doing the exact same thing Tesla is, which is having a person behind the wheel and using the software to assist, so if Tesla isn't breaking the law, how are they?
According to this[1], Tesla has a CA DMV autonomous vehicle testing permit, but the question is whether that applies to this and to Tesla's autopilot, or whether it's for future work towards an actual autonomous mode.
That said, if Uber does want an actual autonomous more eventually, not getting that permit, and thumbing your nose at the authority who gives them out, is probably not a winning strategy...
...
Okay, just looked into a PDF[1] linked form the other article, and according to that, autonomous clearly includes what's going on here, and with Tesla's autopilot. I think Uber may have made a big mistake.
Well, that by itself isn't actually what's in question. The question is what autonomous means, but the PDF linked there does define that, and what Tesla autopilot and Uber's cars are doing clearly does constitute autonomous under their definition. I'm left to conclude that either Uber messed up big time, or rep speaking in that article doesn't know what they are talking about, or it's the beginning of a purposeful disinformation campaign.
> (what) Uber's cars are doing clearly does constitute autonomous under their definition. I'm left to conclude that either Uber messed up big time, or rep speaking in that article doesn't know what they are talking about, or it's the beginning of a purposeful disinformation campaign.
Yeap. I have no idea why they choose not to apply for a permit. Money or resources cannot be the issue for sure.
Tesla isn't doing autonomous city driving, far from it. What is deployed in customer cars right now is little more than highway lane keeping and collision avoidance.
For city driving, they have engineering vehicles that are properly registered and report disengagements and accidents like everyone else.
Tesla cars do not work in residential streets because it is far more difficult to navigate autonomously than the highway.[1]
This is what Uber is doing. Frankly, Uber is reckless and someone is going to die just so that their 2nd tier engineers can feel goey inside. They are running through red lights, cutting off cyclists with their half-baked products.
I wonder if these vehicles are programmed to pull over when cops flash their lights/sirens? Do they accept tickets? Can they be towed without the vehicle rebelling (har!)?
Interested in how enforcement by the DMV would work beyond working with Uber at the corporate level.
They have drivers on board which probably don't want to deal with the consequences of not reacting to police.
Is the possibility that the car could be driving autonomous enough to pull them over, and how to actually prove that was? (Presumably having self-driving hardware fitted isn't illegal, just using it might be?)
According to the DMV, having self-driving hardware is enough to require a permit (the permit that Uber doesn't have / claims they don't need).
If the police want to sidestep that argument, I'm sure the police could find something like "failure be in control of the vehicle" to cite the drivers with, if they see that they're not touching the wheel.
The whole ethos of Silicon Valley and the reason why it exists is lack of regulation allowing for huge amounts of innovation. What Uber is doing is perfectly reasonable and unsurprising, especially given that they don't believe they're breaking the law.
> The whole ethos of Silicon Valley and the reason why it exists is lack of regulation allowing for huge amounts of innovation
Bill Hewlett's oscillator or Moore/Noyce's processor or the Macintosh have unsurprisingly little need for regulation compared to an autonomous car on public roads.
We don't know if it's different or not. It may well be the same. But Tesla has a permit [1] and Uber couldn't be bothered to apply for one even after receiving a warning from the DMV. This is after Uber publicly announcing that they were testing self driving technology [2].
I didn't downvote, but parent, please explain why you are using an offhand comment, and not the actual justification that Uber provided for for why they are not breaking the law.
That's what bugs me. I'm with you on that, but it seems the overwhelming majority of people just don't care about such "externalities" or ethical issues if their lives can be made a little more convenient or they can save a buck.
Its noticeable for me how visceral the reaction to disobeying government is for lots of people. But i think its important to have actors that do it.
Argentina had an Uber-copy cat that played very nice with regulators: from day one they were working with officials and lawyers and were totally legal. They started to pick up demand and the gov. suddently passed a law specifically banning a way to ask cabs that only affected that company. They were crushed by the state one day to another, when they played as nice as they could have.
AFter that I realize that Uber's strategy is necessary. The guys that played nice got their assess kicked.
There's different degrees to disobeying government. There's disobeying regulation that protects rent-seeking entrenched dinosaur industries. And there's disobeying rules that protect pedestrians from being run over by red-light speeding autonomous vehicles.
I can choose whether uber is a suitable replacement for a regular cab for myself. But I can't choose whether I would or would not like to share a crosswalk with one of their self-driving cars.
You're right. If anything, the DMV needs to put greater oversight on Uber, especially after the red light speeding accident. Paying $150 and submitting itself into the permit application process is only the beginning.
Necessary for what? What is the end-game of Uber and does that really align with the best interests of their drivers and/or riders?
Uber is revenue unsustainable and driving for a monopoly status on delivery/taxi.
I don't think the ends justify the means here... especially if there's literally casualties involved.
I'd say reckless disregard for traffic laws is pretty universal - go read up comments on a story about texting while driving - 90% are screeds against those who text while driving, recklessly endangering others on the road.
For many kinds of changes and progress. Disobedience is one of those things that ends up measured by results.
> What is the end-game of Uber and does that really align with the best interests of their drivers and/or riders?
What do you mean? Uber literally makes all their money by providing a service that people willingly consume. They are really providing lots of value.
> Uber is revenue unsustainable and driving for a monopoly status on delivery/taxi.
Maybe, but does it matter? An Uber monopoly is a lot easier to fight than a state monopoly. To the very least, there are several companies competing in the space already. And if Uber goes bust, the only losers are the shareholders.
> Uber literally makes all their money by providing a service that people willingly consume. They are really providing lots of value.
Uber is not making money. They are burning through investor cash in excess of $2B per year in the false hope of breaking the backs of the incumbents so Uber can achieve monopolistic rents far in excess of status quo.
Riders "willingly consume" Uber because of this investor subsidy which conceals the true cost of carriage. When the easy money dries up, it's game over for Uber unless they can go full autonomous sans driver.
Uber is running out of runway and they know it. That's why they're flaunting the DMV in the rushed hopes of getting this Hail Mary to market.
What are your thoughts on the tens of thousands of San Franciscans who do renovations on their homes without getting permits from the city? Are they disgusting as well?
That depends -- do those renovations have a chance of harming other people? If you add stuff to your house that's not up to code, sell it to someone, and that person dies as a result, then yes, you're disgusting. If you use your wealth to change building codes so you're not responsible for that person's death, you're even more disgusting.
You do realize that the permit is just a cash grab. They don't inspect the car, do any sort of analysis, etc. It's just paperwork. That's all. You really support that kind of mindnumbingly stupid bureaucracy?
So Tesla is doing this thing where they're running in the background and logging data of decisions that they would make vs. what actually happens.
I'm wondering if Uber did the same thing at all. Because at the moment, it sure doesn't feel like it. Anecdotally I've heard many stories from friends in the area encounter autonomous Ubers committing traffic violations like running red lights.
If a few people get killed, Uber won't care, and neither will their defenders, any more than they care about hiring drivers who are medically unfit to drive, hiring rapists, or anything else Uber do.
>Asked if Uber was trying to avoid disclosing accidents involving its self-driving cars, as permit holders are required to do, Levandowski denied this
...and now we see why. I'm not sure why journalists attempt to pin "intent." Enough with this "gotcha" bullshit.
Simply inform the public that one of the consequences of Uber's decision to circumvent the law is that they don't have to report accidents which then provides an advantage over its competitors.
Literally the only people who care are SF fresh college grads. The public doesn't care about many things. They only care about their wallet, their family, and maybe the environment.
Keep thinking that informing the public will actually change things. You need a different method.
It is very reasonable, and good journalism, to ask about motives.
If it were reported as you suggest, Levandowski would probably complain of being the victim of innuendo. Providing the responses are accurately reported, asking about motives helps to protect the innocent while possibly stripping the guilty of cover, at least if they end up contradicting themselves.
If one of their autonomous vehicles kills somebody while running a red light, it could well be the end of Uber. Kalanick should probably be fired at this point.
Oddly enough, the answer might be yes. Tesla is under investigation by the NHTSA for their fatal autopilot crash some months ago, in addition to a homicide investigation by the Florida Highway Patrol.
I really doubt anything criminal will be brought against Tesla, but it at least sets a precedent for an investigation. Considering Uber's complete refusal to follow regulation, it seems you could argue they're being negligent and if someone dies that might be interpreted as manslaughter (I am not a lawyer). At the very least, I'm sure it will be a PR nightmare and a huge court case that would put a chink in their armor.
Explain why you think Tesla's situation is different. Their argument is that their cars are like Teslas is all ways that matter legally.
Why isn't Tesla finished after their extremely negligent death?
The first automobiles killed many people and have continued to do so in ever greater numbers. I think everyone knows we're going to have some deaths moving to computer controlled cars but that the price is worth it.
Does anyone know why Uber didn't get a $150 permit?
They've been working on autonomous vehicles for awhile, obviously they knew about the permit requirement months if not years ahead of time. Are they trying to set some sort of precedent, or are there hidden fees or restrictions people aren't talking about that Uber is trying to avoid?
I'm going to be going against the echo chamber here, but I traveled around Europe and South America for five months recently (digital nomad)... The common question was "...why do you think the US/Silicon Valley has the technology and opportunities that it does...". My honest answer... Less regulation and an attitude of do what it takes to be successful in the early days. The hustle.
I think people underestimate how cut-throat technology and business is, especially in the market Uber is in. Uber is doing what is takes to be successful. Ultimately you have to ask yourself, who is going to have the best and brightest minds, lawyers, and lobbyist.... Uber or the Government?
It's interesting to me to read the comments here; so many references to 'running red lights' as if it was a common thing and not one reported instance where Uber says the driver was controlling the vehicle and there was no passenger. Was there some development there I missed and it turns out none of the cars recognize traffic lights?
I thought well of Levandowski, whom I've met, until this.
This is typical of Uber's "blame the driver and avoid liability" approach. If Uber signs up for DMV's testing program, Uber posts a $5M bond and accepts liability. With Uber's plan, the driver has to pay, if they're not an employee. Uber's manuvering here is probably so they can have contractors drive their self-driving cars without Uber taking on liability.
Uber already tries to dump as much liability as possible on the driver.[1]
Tesla doesn't sell "autopilot" any more.[2] The original system was way under-sensored and responsible for at least four major crashes. (I've posted links to the videos before.) Newer Tesla cars come with better sensors, but without the software to use them. Tesla hopes to return with full self-driving, but so far, they're still testing and issuing press releases. Tesla's crashes are still in litigation in several countries, and an NTSB investigation is in progress.
- Agreeing that they are doing testing, would probably make it more difficult to move to "production" - when do these testing requirements end?
- The California regulations are quite light, but other cities might come up with more complicated requirements. Agreeing in California would set a precedent and make it harder to avoid complying in other cities.
- There's requirements for the test driver. These might be difficult for Uber if they want to use semi autonomous cars in normal business. There is for example requirements for specific training, driver has to be employee or contractor of the company etc.
- If they agreed to apply for testing permit, it might be that the permit is declined or later suspended.
- As others have stated, not having to disclose the accidents and other statistics might allow them to avoid bad publicity and keep stuff hidden from competitors.
- Looks like one part of the requirements is to enroll into "Employer pull notice program". I don't think that applies just for test vehicle drivers. Would this have some implications for company like Uber?
https://www.dmv.ca.gov/portal/dmv/?1dmy&urile=wcm:path:/dmv_...
I want to know if the red light run caught on video was while the car was in self driving mode. Of course there was a human in there, and Uber blames them, but what role did the self driving play? Was it a case where the self driving software messed up and the human didn't catch it in time?
If so this is a pretty big deal. Wouldn't you expect the very first people they put in these things to be competent people that were doing their best? This would indicate to me that it isn't realistic to expect the humans to jump in that quickly.
There's a dude stepping out into the crosswalk, fer Chrissake.
> I want to know if the red light run caught on video was while the car was in self driving mode. Of course there was a human in there, and Uber blames them, but what role did the self driving play?
I'm guessing if they had a permit, they would've had to disclose this to the DMV.
>Wouldn't you expect the very first people they put in these things to be competent people that were doing their best?
I recently mentioned on another thread that I think the primary purpose of that driver is not safety, but to absorb culpability in the event that the tech fails.
I just don't think it's reasonable to expect any human to reliably perform the task of correcting the vehicle. Imagine sitting at the wheel in a passive mode for an hour as the car performs flawlessly. Then, in one intersection, it suddenly malfunctions. Now, you are supposed to a) recognize that it's failing and b) suddenly take corrective action--all in, perhaps, a split second?
When you're controlling the vehicle, you know your intent. When you are "monitoring" you must assess the vehicle's intent. Even if you do so perfectly there is, by definition, some delay. That delay could be costly.
Yeah, completely agree on how it is unrealistic to expect them to correct in such a situation.
I would think, though, that they'd still be smart enough to put someone competent in there. Because if there is an accident, it is still going to look bad on them and the tech. (and of course this supports what you are saying, that it simply isn't reasonable to expect them to do that)
Agreed. There is still a chance the human will see something that the vision-system misses -- like the red light far off to the side. The human may instinctively hit the brakes to avoid hitting someone/something.
In the long run regulation (read artificial market manipulation) hurts everyone. Law/regulation (as well intentioned as it may be) always moves slower than people's wants and needs.
What about cases where the free market is known to poorly allocate resources? I'm talking about negative externalities and public goods.
As a concrete example, take regulation regarding pollution that is known to be harmful to humans and therefore have a large cost. The free market will not generally take this cost into account.
If the public majority want to favor the environment over cheap rides that's fine. I just want the public to decide unanimously with their dollars. If pollution becomes a large enough problem to sway the public at large to reallocate their dollars I firmly believe they will react in each individual's best interest.
Fortunately, with the internet the knowledge about all of these subjects are known. I'm not saying people won't make mistakes even with 100% knowledge of a given subject. I'm just saying it should be the people themselves to make the mistakes, not a select few.
113 comments
[ 2.0 ms ] story [ 123 ms ] threadSo it's okay to violate regulations because your engineers want to see their work in their city?
Official statement: https://newsroom.uber.com/statement-from-anthony-levandowski...
Other than the pointless flaunting of privilege in this ... I think keeping engineers very close to the project is usually a sign that not all is well with the project.
I read that as "We can't let this project out of sight of the engineers ... they need to be able to make changes on a day to day basis, from direct observation."
That these engineers would be moving in breakneck speed (ha!), to fix whatever they see in person rather than the usual separation of concerns in testing mission critical software.
Through those glasses, it is roughly an admission that it is in "patch everyday" phase of development.
If so, I've worked on several hardware projects (IC design) and IoT projects as a software developer and I couldn't disagree more strongly. That the engineers are close to the product so they can make quick iterations without massive amounts of waiting and telephone game confusion is a sign that they are doing a good job and don't have an idiot culture of engineering bureaucracy. IMO.
If it gets a bunch of people killed, yes.
And here we get to the heart of the difference between what is called "software engineering" and what is, in every other field, called "engineering".
There's a reason why Canadian engineers get the iron ring and legend says the first ones were made from the ruins of a bridge that collapsed (due to poor design). There's a reason why people describe aviation as a field where every regulation is written in blood.
Uber is moving into a field where no amount of "don't bring politics into this" can hide the fact that a "bug" in their product will potentially directly kill humans. Requiring people to A) actually be fucking careful and B) actually prove it is how this world works, and I don't think Uber's going to manage to skirt it with their usual regulatory arbitrage technique. The sooner they grow up and start acting like real engineers, the better.
I should have place my commas differently to emphasize that I have worked on many IC design projects as an EE then later on IoT projects as a software engineer.
No one replying to my comment has at all explained why during early iterations when the product is quickly evolving this separation is a good thing. Frankly it seems to me like (a) people are taking pot-shots are Uber and (b) none of you have actually worked on a hardware project.
There is a time when a strict separation between designers and testers is critical. That time is referred to as Qual, or qualification and it is for validation that a part or system operates correctly over all corners of the devices operating parameters. This is not the phase that Uber is in.
Again, I don't dispute at all that Uber are being arrogant and reckless jackasses. I'm just saying it has nothing to do with the fact that their engineers desire to work near to the systems they are designing since that is actually a sign of a smart and high functioning engineering team, not the opposite as people have suggested.
Do you know what auto makers do in this kind of situation? They build a test track and a workshop at the test track and probably a larger facility adjacent to it, and situate the engineers there. They don't say "well, our engineers live in one of the densest urban areas in the country, let's just test our new car there to be close to them".
Uber should get a test track out in the desert or something and relocate some engineers to it. And maybe without the distractions of the SF office they'd manage to get some basic stuff tested, like "recognizes red lights" or "understands how to turn from the rightmost lane".
Maybe they should relocate everyone! Total tangential point!
That said, if Uber does want an actual autonomous more eventually, not getting that permit, and thumbing your nose at the authority who gives them out, is probably not a winning strategy...
...
Okay, just looked into a PDF[1] linked form the other article, and according to that, autonomous clearly includes what's going on here, and with Tesla's autopilot. I think Uber may have made a big mistake.
1: https://www.dmv.ca.gov/portal/dmv/detail/vr/autonomous/testi...
2: https://www.dmv.ca.gov/portal/wcm/connect/d48f347b-8815-458e...
Tesla is there, Uber is not.
Yeap. I have no idea why they choose not to apply for a permit. Money or resources cannot be the issue for sure.
For city driving, they have engineering vehicles that are properly registered and report disengagements and accidents like everyone else.
This is what Uber is doing. Frankly, Uber is reckless and someone is going to die just so that their 2nd tier engineers can feel goey inside. They are running through red lights, cutting off cyclists with their half-baked products.
[1] http://www.wsj.com/articles/tesla-restricts-autopilot-on-res...
Interested in how enforcement by the DMV would work beyond working with Uber at the corporate level.
Is the possibility that the car could be driving autonomous enough to pull them over, and how to actually prove that was? (Presumably having self-driving hardware fitted isn't illegal, just using it might be?)
If the police want to sidestep that argument, I'm sure the police could find something like "failure be in control of the vehicle" to cite the drivers with, if they see that they're not touching the wheel.
What a bunch of law breaking douches. Yes, let's call them what they are. No: no one is above the law.
Edit: downvoters: please explain why someone should be above the law. thanks.
> What Uber is doing is perfectly reasonable... especially given that they don't believe they're breaking the law.
That is an interesting point of view. Perhaps we should call it the psychopaths' charter?
Bill Hewlett's oscillator or Moore/Noyce's processor or the Macintosh have unsurprisingly little need for regulation compared to an autonomous car on public roads.
[1] https://www.dmv.ca.gov/portal/dmv/detail/vr/autonomous/testi...
[2] https://newsroom.uber.com/san-francisco-your-self-driving-ub...
EDIT: https://news.ycombinator.com/item?id=13181514
https://news.ycombinator.com/newsguidelines.html
https://news.ycombinator.com/newswelcome.html
We detached this subthread from https://news.ycombinator.com/item?id=13198125 and marked it off-topic.
And people using them to hail a ride should really reconsider if they want to support a business like this.
I don't.
Uber has become too convenient to quit, it seems.
Argentina had an Uber-copy cat that played very nice with regulators: from day one they were working with officials and lawyers and were totally legal. They started to pick up demand and the gov. suddently passed a law specifically banning a way to ask cabs that only affected that company. They were crushed by the state one day to another, when they played as nice as they could have.
AFter that I realize that Uber's strategy is necessary. The guys that played nice got their assess kicked.
True. What about Bitcoin entrepreneurs? A cavalier attitudes towards laws isn't something inherently bad, or "immoral".
Uber is revenue unsustainable and driving for a monopoly status on delivery/taxi.
I don't think the ends justify the means here... especially if there's literally casualties involved.
I'd say reckless disregard for traffic laws is pretty universal - go read up comments on a story about texting while driving - 90% are screeds against those who text while driving, recklessly endangering others on the road.
For many kinds of changes and progress. Disobedience is one of those things that ends up measured by results.
> What is the end-game of Uber and does that really align with the best interests of their drivers and/or riders?
What do you mean? Uber literally makes all their money by providing a service that people willingly consume. They are really providing lots of value.
> Uber is revenue unsustainable and driving for a monopoly status on delivery/taxi.
Maybe, but does it matter? An Uber monopoly is a lot easier to fight than a state monopoly. To the very least, there are several companies competing in the space already. And if Uber goes bust, the only losers are the shareholders.
Uber is not making money. They are burning through investor cash in excess of $2B per year in the false hope of breaking the backs of the incumbents so Uber can achieve monopolistic rents far in excess of status quo.
Riders "willingly consume" Uber because of this investor subsidy which conceals the true cost of carriage. When the easy money dries up, it's game over for Uber unless they can go full autonomous sans driver.
Uber is running out of runway and they know it. That's why they're flaunting the DMV in the rushed hopes of getting this Hail Mary to market.
I'm wondering if Uber did the same thing at all. Because at the moment, it sure doesn't feel like it. Anecdotally I've heard many stories from friends in the area encounter autonomous Ubers committing traffic violations like running red lights.
The laws here are a bit different than they are in California:
• Do not drive in a bike lane. You may cross a bike lane when turning or when entering or leaving an alley, driveway, or private road.
• Do not move into or travel in a bike lane in preparation for a turn.
Makes me wonder how these cars are going to handle individual state laws like this.
If a few people get killed, Uber won't care, and neither will their defenders, any more than they care about hiring drivers who are medically unfit to drive, hiring rapists, or anything else Uber do.
Actually, traffic laws can even be specific to a city.
- In NY City: No right turn on red, unless otherwise posted.
- In the rest of NY State: Right turn permitted on red, unless otherwise posted.
...and now we see why. I'm not sure why journalists attempt to pin "intent." Enough with this "gotcha" bullshit.
Simply inform the public that one of the consequences of Uber's decision to circumvent the law is that they don't have to report accidents which then provides an advantage over its competitors.
simple.
Keep thinking that informing the public will actually change things. You need a different method.
If it were reported as you suggest, Levandowski would probably complain of being the victim of innuendo. Providing the responses are accurately reported, asking about motives helps to protect the innocent while possibly stripping the guilty of cover, at least if they end up contradicting themselves.
Cool! Look at that! See, the're safe! I want to try!
I really doubt anything criminal will be brought against Tesla, but it at least sets a precedent for an investigation. Considering Uber's complete refusal to follow regulation, it seems you could argue they're being negligent and if someone dies that might be interpreted as manslaughter (I am not a lawyer). At the very least, I'm sure it will be a PR nightmare and a huge court case that would put a chink in their armor.
http://www.businessinsider.com/tesla-gets-extension-in-nhtsa...
Why isn't Tesla finished after their extremely negligent death?
The first automobiles killed many people and have continued to do so in ever greater numbers. I think everyone knows we're going to have some deaths moving to computer controlled cars but that the price is worth it.
So for now I'll stop using Uber... at least for 30 days.
They've been working on autonomous vehicles for awhile, obviously they knew about the permit requirement months if not years ahead of time. Are they trying to set some sort of precedent, or are there hidden fees or restrictions people aren't talking about that Uber is trying to avoid?
So I think if they are charging for the rides they aren't really just testing the system.
There's also a bunch of requirements about formally reporting incidents and so on.
I think people underestimate how cut-throat technology and business is, especially in the market Uber is in. Uber is doing what is takes to be successful. Ultimately you have to ask yourself, who is going to have the best and brightest minds, lawyers, and lobbyist.... Uber or the Government?
This is typical of Uber's "blame the driver and avoid liability" approach. If Uber signs up for DMV's testing program, Uber posts a $5M bond and accepts liability. With Uber's plan, the driver has to pay, if they're not an employee. Uber's manuvering here is probably so they can have contractors drive their self-driving cars without Uber taking on liability. Uber already tries to dump as much liability as possible on the driver.[1]
Tesla doesn't sell "autopilot" any more.[2] The original system was way under-sensored and responsible for at least four major crashes. (I've posted links to the videos before.) Newer Tesla cars come with better sensors, but without the software to use them. Tesla hopes to return with full self-driving, but so far, they're still testing and issuing press releases. Tesla's crashes are still in litigation in several countries, and an NTSB investigation is in progress.
[1] https://www.hg.org/article.asp?id=37714 [2] http://www.businessinsider.com/musk-tesla-remove-some-autopi...
- Agreeing that they are doing testing, would probably make it more difficult to move to "production" - when do these testing requirements end?
- The California regulations are quite light, but other cities might come up with more complicated requirements. Agreeing in California would set a precedent and make it harder to avoid complying in other cities.
- There's requirements for the test driver. These might be difficult for Uber if they want to use semi autonomous cars in normal business. There is for example requirements for specific training, driver has to be employee or contractor of the company etc.
- If they agreed to apply for testing permit, it might be that the permit is declined or later suspended.
- As others have stated, not having to disclose the accidents and other statistics might allow them to avoid bad publicity and keep stuff hidden from competitors.
- Looks like one part of the requirements is to enroll into "Employer pull notice program". I don't think that applies just for test vehicle drivers. Would this have some implications for company like Uber? https://www.dmv.ca.gov/portal/dmv/?1dmy&urile=wcm:path:/dmv_...
If so this is a pretty big deal. Wouldn't you expect the very first people they put in these things to be competent people that were doing their best? This would indicate to me that it isn't realistic to expect the humans to jump in that quickly.
There's a dude stepping out into the crosswalk, fer Chrissake.
I'm guessing if they had a permit, they would've had to disclose this to the DMV.
I recently mentioned on another thread that I think the primary purpose of that driver is not safety, but to absorb culpability in the event that the tech fails.
I just don't think it's reasonable to expect any human to reliably perform the task of correcting the vehicle. Imagine sitting at the wheel in a passive mode for an hour as the car performs flawlessly. Then, in one intersection, it suddenly malfunctions. Now, you are supposed to a) recognize that it's failing and b) suddenly take corrective action--all in, perhaps, a split second?
When you're controlling the vehicle, you know your intent. When you are "monitoring" you must assess the vehicle's intent. Even if you do so perfectly there is, by definition, some delay. That delay could be costly.
I would think, though, that they'd still be smart enough to put someone competent in there. Because if there is an accident, it is still going to look bad on them and the tech. (and of course this supports what you are saying, that it simply isn't reasonable to expect them to do that)
As a concrete example, take regulation regarding pollution that is known to be harmful to humans and therefore have a large cost. The free market will not generally take this cost into account.
In a world where all information was equally distributed, sure, maybe anarcho capitalism would work.