>Hotz's initial tweets suggested his move was in response to an inquiry from NHTSA, stating that he "would much rather spend life building amazing tech than dealing with regulators and lawyers,"
well, this is not some stupid web-app that can't harm anyone even if it tried (except from maybe overheating your smartphone). This can kill people, i can't seriously buy dangerous stuff from somebody that says something like this. What's next? Producing food without following proper hygiene regulations? Disrupting the medicine-industry by testing your headache-pills on your customers (those regulations are only slowing us down!)?
You would have won so much Irony Points if only you had written "Producing food without following proper hygiene regulations? Disrupting the medicine industry with revolutionary tests without sufficient scientific validation?"
> Producing food without following proper hygiene regulations?
Sounds like Menu Next Door, who had to deal with regulation meant for restaurants when trying to be the AirBnB of food. Things like you had to have a warning about washing your hands in your own private kitchen.
Only this would be worse. Someone can choose to buy Soylent or use Theranos for themselves and no one else gets hurt. If you a person buys a crappy self driving car system it could kill me, a totally unrelated person.
I don't care how much of a libertarian you are, that's exactly the reason why we have regulations.
The law of the jungle is fine, but it won't preclude a shitty self-driving car from crashing through your living room and killing one of your kids. It's one thing to suffer harm to yourself, it is quite another to go through losing a child to something like that. I know, randomness, blah blah blah.
Technically, Soylent abides by regulations: they sell flaxseed mixed with peas and oat (I think: I’m in Europe and eat Huel, which has mostly those). None of those are not well-known food sources. There are no regulation about claiming that eating nothing but that is safe; they should (at least, Huel does) have the recommended daily doses of each major molecule on the bag.
No, it's just a wrong assumption. They're still selling the pre-made liquid, they just stopped their bars and powder products proactively, even though they couldn't reproduce the issue in testing. Do you even know anything about their production methods?
> Producing food without following proper hygiene regulations? Disrupting the medicine-industry by testing your headache-pills on your customers (those regulations are only slowing us down!)?
It would have been even worse than that because it could have potentially killed other people, too.
Even ignoring the safety aspect, people buying any remotely major product will have the expectation that it will ideally be supported for some time to come. e.g., buy a car and have the manufacturer around for long enough to help out with servicing, parts, etc.
(ACDelco is a house brand of GM, similar to Chrysler's Mopar or Ford's Motorcraft.)
The auto companies are an interesting example, because all of the major manufacturers have subsidiary organizations which exist just for parts manufacture, and all of them handle parts manufacturing and distribution for far longer than the original manufacturing run. While GM might only make a particular model car for a few years, ACDelco might continue to make consumable parts for it for decades, based on demand.
> stupid web-app that can't harm anyone even if it tried (except from maybe overheating your smartphone)
It can leak very sensitive personal data. That won't hurt you physically but will surely have consequences. The problem is perhaps the lack of regulations in the web-app domain. Any amateur can call himself a programmer and many do.
> The problem is perhaps the lack of regulations in the web-app domain.
There are and perhaps should be more regulations about safeguarding personal information, which for websites would mean salting passwords etc., but I am glad there are no regulations regarding who can and cannot program as that would seriously hurt innovation.
Maybe Linus wouldn't be able to program Linux? or Zuckerberg Facebook?
The current pace of large data leaks is alarming and unsustainable. Any other industry that would fail so badly would be completely taken over by regulations. The silicon valley is only keeping its independence because until now politicians didn't really grasp the scope of the problem. But now that data leaks involve their own email, I suspect Washington is going to start paying attention.
Yes, regulations do hurt innovation and also competition, it creates large barriers to entry (check how many new banking licenses are issued every years!). But the current state of infosec is just unacceptable. It's like if planes were falling from the sky every day.
I've said it before and I'll say it again: you can't apply Silicon Valley disruption to every industry because most software has a failure-rate that would be considered appalling in any real engineering discipline.
And building a control system that can follow a perfectly black asphalt road with bright yellow and white lines between the lanes isn't even 1% as difficult as building a system that can actually be relied upon to safely convey its passengers under less ideal circumstances.
Reminds me of that old joke from the '80s(?) about what if MS built cars? You would get the blue screen of death while driving on the highway, and have to stop your car and restart it, you would get cryptic error messages when you were running out of gas, etc....
Yeah, this article is pretty shallow and has pretty significant levels of fanboy-ism. Like
"The little meaningful criticism of Tesla Autopilot (with its state-of-the-art “Autosteer") is rooted in what critics deem insufficient transition warnings"
... the fact that it killed someone notwithstanding.
TFA also seems convinced that there is a large difference between the autonomous capabilities of the Model S vs. other luxury cars, when it's pretty well documented that the main difference lies in more or less conservative safety choices and marketing.
This paragraph from the article is so stupid it reads like a satire on Silicon Valley:
> Does anyone really believe Hotz would give up with $3M+ in the bank and VC firm Andreessen Horowitz behind him? Hotz could have responded to NHTSA. He just didn't want to. With a lean operation, a growing pool of crowdsourced data and seemingly unlimited swagger, Hotz just bought himself press, time and additional mythos.
> China has the same level of regulation as other industrialized countries.
I have to assume you've never driven or been driven in a major Chinese city, because while this may be technically true in practice it is very much not. Cars on the road in China are often lacking in terms of safety.
It is something of a moot point, because I don't see how any self driving car system could perform in Beijing traffic conditions. Basic concepts like lanes frequently go out the window.
They weren't given any government requirements, with the the possible exception of the FHMVSS rule that you can't remove the rear-view mirror of the car. The NHTSA Special Order asked only for documentation of extremely basic test plan stuff (road, speed, weather, and traffic limits for the device, along with user documentation).
That was their request for information, yes. But I have trouble believing that the NHTSA would allow you to market some half-baked crap that takes over control of your vehicle.
Maybe that's legal, though. Not great if it is. Anything short of full, reliable automation seems incredibly dangerous.
People have a lot of theories about what this letter meant, but the people on HN who have actually worked with NHTSA orders say the order means just what it says it means, and that NHTSA is pretty reasonable to work with.
Certainly the questions in the order --- I posted them on the first thread about the comma one --- seemed pretty basic and reasonable.
The cover letter was very "do not blow off the NHTSA". One commenter who had experience with NHTSA provided examples of orders they sent other companies with similar language.
Typically when you get to the point of getting one of those it's because you've been blatantly ignoring regulations or regulators up until now and they feel like they need to get your attention.
Well if they were planning on going to market without being able to pass an audit then they deserve every bit of it. I've worked in am FDA regulated field for a little over a decade and getting in touch with regulatory bodies up front is par for the course. Sounds to me like they don't know what they're doing and are in over their heads.
On the surface it looks like a small speed bump compared to what some startups face in, say, medical devices.
But it depends on where Hotz got his capital. We know it's not a big number. The compliance requirements might have cost more in consultants who are experienced in replying to such information requests, than his paid in capital to date, IF Hotz could find good consumtants in time.
It seems like he had alternatives: To delay roll-out and look for more capital, for example. But everything other than shutting down the project would cost a significant, and possibly, depending on Hotz's access to investors, an insurmountable amount of money within the time he had.
Shutting down might get regulators off his back long enough to take the project offshore.
Or, you know, the gigantic field of possibilities between those two things: it works a little bit under some conditions.
Fortunately for us and unfortunately for Hotz's ego, regulators look down upon things that work a little bit under some conditions when they also happen to control two ton metal cages capable of traveling at 120mph.
> it's [the] NHTSA that remains behind the curve in evaluating the myriad different technologies
Uhhh... that was exactly what they were requesting to do with their letter.
The one fatality with Autopilot happened because the system worked too well, the driver came to trust it when he shouldn't have. The shittier the Comma one is, the more in-the-loop the human has to be, and risk compensation becomes less of a factor.
A system that is seemingly half blind (repeatedly crashes into vehicles that only partially obstruct the lane), decapitates a guy, crashes another one into a fucking post
It didn't work too well.
It pretended to work too well and gotten people killed. There is a reason MobileEye walked away from Tesla
They are being rather cavalier with their attitude to safety, apparently. (Apparently the falcon wing doors can go nom nom on you if you are not careful too - sure the guys are laughing their asses off, but it's not a good design)
One fatality in 200 million miles of driving reflects a cavalier attitude to safety? It's not like the guy who died in that accident wasn't familiar with autopilot's reccommended usage, he read the manual, Tesla made it pretty clear you still gotta watch the road while it's engaged. I mean, if you're into misusing cars, any of them will allow you to accelerate into a brick wall if that's your inclination.
I'm not sure what Falcon wing doors has to do with any of this.
There is at least 1 other death in China where autopilot is cited as the cause, and there are many other accidents in China which are not widely reported in the United States (it's easy to find several on Google though, many of which are even posted on YouTube with video proof).
Statistically speaking, Tesla's autopilot is more dangerous than regular driving. They spun bad statistics to claim otherwise. They ignored the difference between highway driving (which is safer) and all driving, and ignored demographics (most Tesla drivers are older and affluent).
It is too early to really draw many conclusions about autopilot, because there is actually very little data. Comparing the number of miles driven by Tesla to the number of miles driven by all other cars in one year is a difference of several orders of magnitude. But they are under investigation, and as more time passes the more damning it seems for Tesla.
It's true that there's too little data to draw conclusions. A Toyota spokesperson claimed we'd need about 8 billion km of autonomous driving to draw statidtically relevant data. and of course, by the time a single company, or even all companies combined have gathered that many miles the technology will have progressed and the early data will be irrelevant.
Safety arguments are only meaningful in these sorts of abstract discusssions, though. Keep in mind that when we decided to place the automobile at the centre of American life in the early postwar era, fatalities per vehicle mile travelled were 10x what they are today. We didn't do it because it was safe, we did it because cars are cool, and some big companies stood to make a lot of money.
You think maybe part of why there haven't been more numerous malfunctions might be because of rigorous NHSTA testing of various parts of the automobile?
Also just because a person on birth control gets pregnant doesn't mean we start pushing pills with unknown success rates in order to get people to use condoms.
If you're imagining that autopilot is anything more than driver assist technology, then there already is numerous malfunctions, autopilot fucks up regularly. With or without NHTSAs involvement, Autopilot (or Comma one) wasn't designed to be safety critical autonomous driving technology. NHTSA has about as much business regulating it as it does regulating whether or not you can hang dice on your rear view mirror.
If it were a form of birth control, level 2 would be the pull-out method: no guarantees, so be vigilant.
Nope. I'm imagining that NHSTA keeps people safe, and I don't want a car anywhere near me that's controlled by something that the NHSTA hasn't had a chance to even evaluate.
It's foolish to believe that real humans, driving real cars, will treat these technologies as mere "driver-assists." Autopilot is used -- shockingly enough -- as an autopilot. And even if they /were/ used perfectly, these systems are exerting physical control over the car -- something that the National Highway Traffic Safety Administration has a distinct interest in managing. The NHTSA would be grossly negligent if they did not attempt to regulate these devices.
I have yet to see the simplest proof that comma ai work is just a cute trained NN over a camera, piped to basic controls, with some potential creative ideas (in a conf, hotz mentioned working on time based depth inference instead of the traditional stereo camera setup).
I'm biased for conservative companies, I have no idea what they're doing and how much resources. Maybe they're doing something at the same level. I believe they have more refined physics model to compute surroundings and car physics in a way to ensure sane and safe behaviour. Exception made of Tesla AP, which I always found borderline lying.
The medias and the crowd seeems to believe wild hacking and current year tech is giving you better products, but physics are physics. I'm still waiting for a proof of solid math, system and safety engineering there.
psedit; I forgot, I peaked at commaai github repo, the NN code is way above my head and what I could imagine, so I have to give hotz' team that.
Actually all my money was on google, i expected them to drive into the scrapfield, after the car companys and Tesla pushed each other into a early market situation and upped Legislation, kicking themselves out of the market, with a carnaggedon.
But they backed out, because their managers want the chicken to exist before the egg. So, why not bet on the outsider?
I liked Google's approach a lot. It seemed justly dimensioned, came out of Thrun DARPA success, had years and millions of miles of testing. What's not to love. Also G has the lever to accelerate and change tech if their old approach is obsolete.
Sad they went silent since a year or so.
I said it in other place, including HN, I don't consider Hotz an outsider, but mostly glorified hacker (his attitude with institutions and media don't help me clear this impression); and I don't think hacker mentality bodes well with this task.
Isn't Google's approach dependent on expensive LIDARs and high resolution 3D maps [0], or have they changed that? I don't see that kind of system scale well.
> I was really hoping it was actually a C64 reference, namely: "where the 'drive' goes"... ,8,1
Actually, the drive number is the “8”; the “1” is a flag which means to load the program to the same address in memory it was saved from, instead of to the default BASIC program start address.
I would include an (unlikely) 3: Hotz really, really does not want to be in the same room as the Feds (he is a confessed hacker) or spend time dealing with them. Maybe more likely: he can sell a ton _now_ in China, had a partner car company, and expect that having those work there would make his success in US either negligible in comparison, or faster thanks the example.
With your dignity intact? The guy basically lost all his credibility with this action.
Walking away with dignity intact would be replying to NHTSA, "Dear Government, we've found out we are unable to provide the information you asked for because the product is nowhere near mature enough. Therefore we are officially abandoning all public tests and sales effort, and are going back to the drawing board until such time that we are able to get back to you with the answers you requested."
I think he meant "ego" instead of "dignity". This is how you walk away with your over-inflated ego intact. Behaving this way indicates that you have no dignity.
We don't need another crap Level 2 system. Lane keeping plus auto braking just isn't enough. (Crap auto braking, as with Tesla, is even worse. Four collisions with big, obvious stationary vehicles so far.)
The NHTSA, Google, and Volvo have it right. Systems where the driver has to back up the automation will not work in practice.
I think it's far more likely that the NHTSA was caught flat-footed by Tesla's autopilot system, and didn't react in time, with fatal results. Now they're seeing a scrappy startup trying to dive into the same space where multi-billion-dollar car companies ought to fear to tread, and they've done their best to nip it in the bud before every tom dick and harry with a webcam and a graphics card starts trying to send their car on solo missions.
Worth noting that Alex Roy and Hotz probably have/had a business/working/personal relationship:
"From August 24th-27th of 2016, Roy and teammates Warren Ahner & Streetwars founder Franz Aliquo broke the Electric & semi-Autonomous Cannonball Run records again, driving a Tesla Model S 90D 2,877 miles from the Portofino Inn to the Red Ball garage in 55 hours, breaking the prior record by 2 hours & 48 minutes. Tesla Autopilot was engaged 97.7% of the journey. GPS data and video evidence was captured using both a US Fleet Tracking device, and Comma.ai's Chffr video logging software."[0]
It would not be a leap to assume that Roy might have a vested interest in puffing up his acquaintance. Also, Alex Roy is a "journalist" insofar as he probably got paid to write that piece. But other than that...
Remember a few weeks ago when it was a headline that the "going rate" for talented engineers working on self-driving cars was $10 million per head?[1] Meanwhile, comma.ai the company only got $3.1 million from a big-name investor?[2] Yeah, don't quit your day job...
The only embarrassed entity here is George Hotz. I find it extremely hard to come up with a charitable narrative in all this but at least the system appears to be working: hacked together crap is not going to be sold to consumers.
It's one thing to break other systems, it is an entirely different thing to produce safety critical software systems interacting in real time with hardware under real world conditions and it is not a given that someone who is good at the one will also be good at the other.
Hm, well the article certainly is flattering toward Hotz, the project, and the impression (having first-hand experience with several systems does strike me as important).
Still pretty staunchly of the opinion over here that he's got something worth still working on and this isn't the last of it. AKA contrary to a good number of comments which basically say he's either a fraud, huckster, or otherwise dumb for not being interested/willing/prepared to engage with the NHTSA at this stage.
The comma only works on a few (maybe just one) Honda car that has a built-in front facing radar that they're able to tap into. With that, comma is using roughly the same sensor package as the V1 Tesla autopilot.
Thanks. I didn't know Comma One also uses front facing radar. Do you happen to know what state of the art millimeter wave radars are capable of? I found some references from 2013 about Panasonic product with resolution of 20cm, but maybe there are better options now?
LIDAR is still not that good for driving. It's better than cameras but it's still pretty bad, because its vision can be obscured by weather conditions. Radar is the gold standard for vision like this.
EDIT: cameldrv pointed out in a reply to my original comment that Comma One uses its cameras and also taps into the vehicle's existing front facing radar.
I was thinking of something like Quanergy's solid state LiDAR. I know the planned $100 version is probably a few years out, but it seems promising. And there are other companies/institutions working on cheap LiDARs that work in bad weather.
Following link is of a talk Quanergy CEO gave at Stanford.
A certain type of click bait journalist has an interest in creating EEStor-type narratives. True believers in the Magic Product keep clicking links for months or years at a time.
What this article tries its damnedest to elide is the narrative-wrecking fact that Holtz, confronted with a standard request to demonstrate that his product meets regulatory standards, took his ball and went home. /Exactly/ why he threw a tantrum is anyone's guess -- but there doesn't seem to be any real good possibility. Either his product couldn't meet standards and this is an attempt to obfuscate that inconvenient fact, or he withdrew a working product because he was so affronted by the fact that the NHTSA dared challenge him -- hardly a defensible, or confidence-inspiring, behavior.
There were major problems with every demo I saw of the product. In one video, the camera lost tracking of the road due to lighting conditions not being favorable for contrast. In another, Hotz seemed far too amazed that the car properly took a gentle bend in the highway.
Confidence is not something I would assign to his product. Or any product that is a camera-only lane-follower.
That Holtz caves at the slightest thought of a regulator looking at his work easily shows how immature he is as an engineer.
In general I like the approach of using deep-learning instead of complex hardcoded rules (see the article above for more details), but even with deep-learning you're going to need a complex ruleset in order to deal with the edge-cases.
Getting a prototype of a self-driving car working is probably much less than 5% of the overall amount of work that's required. Getting the car respond correctly to the long tail of edge cases is the other 95% of the work.
Actually I think fewer lines of code is less risk. Added complexity creates more points of failure, more possible bugs, more unexpected edge cases. Complexity is not inherently desirable.
Well, maybe the reason is that these regulator requests costs tons of money to fulfill because no, you cannot you give simple laymans answers to the questions, you need a team of lawyers to do this work, because everything you say in these "simple" requests can and will be used against you.
And perhaps there are better uses of their time, and better regulatory environments where they can release their product.
Maybe in a couple months we are going to hear about the imminent release of the Comma Two in Not-America. I seriously doubt they've gone home. There are tons of places where they could starting selling right NOW.
>Well, maybe the reason is that these regulator requests costs tons of money to fulfill because no, you cannot you give simple laymans answers to the questions, you need a team of lawyers to do this work, because everything you say in these "simple" requests can and will be used against you
No, what you need is to have followed a sufficient quality process to begin with. Lawyers aren't involved heavily in this sort of thing; your quality group is (well, it may take more lawyering for them at this point.)
This guy obviously didn't have a clue when he set out to build this thing and now his ignorance is biting him in the ass. I've worked on FDA regulated devices my entire career. You don't get to just throw things that represent a substantial safety risk onto the market without oversight, and that's a good thing.
> Maybe in a couple months we are going to hear about the imminent release of the Comma Two in Not-America. I seriously doubt they've gone home. There are tons of places where they could starting selling right NOW.
Then I'm very thankful for the NHTSA we have in America. I don't want a bunch of half-ass, poorly-engineered self-driving cars on our roads. I use those roads, and so do my family members.
While I'm looking forward to self-driving cars taking over, I don't want the Comma One in the field if they can't handle the most basic safety inquiries.
The guy that you're replying to seems to have no idea how these things work. I worked for an automotive manufacturer in India (hardly a paragon of virtue in terms of regulatory strictness), and we had rules about even seemingly-inane stuff like the HVAC being able to cool the car down in a certain amount of time, and maintain a given temperature overnight on a cold night etc.
You can't just walk in being an arrogant, know-it-all whiz kid and demand that the rules be bent and mere mortals be forced to step aside in deference to your superior intellect.
Nowhere, in the ENTIRE WORLD of 7 BILLION PEOPLE could you release something like this?
It doesn't matter if most people reject your idea, you only need a couple cities to say "approved for preliminary testing for a limited number/couple thousand beta testers" to go into Prod. And then when you go into prod you get real life data and can prove that your product really is as awesome as you say it is, THEN you go back to the countries that said no, but this time with more money and actual facts to back you up.
This isn't some super expensive product that is out of reach for the world. This is a thousand dollar conversion kit. If these things were available and I owned a Honda, I would have bought it to just mess around with it instead of buying the HTC Vive that I got last night.
I am sure there are lots of beta customers around the world who would be interested in getting a prelim beta version.
> Then I'm very thankful for the NHTSA we have in America. I don't want a bunch of half-ass, poorly-engineered self-driving cars on our roads. I use those roads, and so do my family members.
That sounds like you think that other countries would allow "a bunch of half-ass, poorly-engineered self-driving cars on our roads"?
As an EU citizen, I can assure you that this is widely off the mark.
I don't think he made any statements about the rest of the world, he just made a statement about his local situation without any implications aimed elsewhere.
Daniel Berlin, an attorney with regs experience, on another thread on HN, said: no, you can pretty much just respond to NHTSA with straightforward answers and whatever documentation you have, then arrange a conference with them to figure out what additional documentation you'd need to generate to move forward with your plans.
I don't believe it's the case that Hotz needed to incur six figure legal bills to respond to NHTSA's initial order. He probably could have responded to NHTSA personally, without counsel, and game-theoretically not damaged his potential outcomes at all. He had the option to find out, and threw it away.
The request is well within the realm of what anyone who wants to produce a product that could well endanger the lives of a lot of people needs to be able to deal with. It was the very minor ante required in order to play at a very high-stakes table.
The letter sent by NHTSA to Comma.ai[1]. I'm a layman but to me the penal provisions ($21k/day up to $105M for not responding by 11/10, prison sentence of 15 years for falsifying/withholding information) sound kinda harsh. Can someone confirm that these conditions are standard in requests of this type?
Honest to Odd, could have a normal, shed-hacker fullfill these requirements? Even if he had the money? Is there a way to shrink the requirement- like building cars that can drive not faster then 20 mph in full automated mode?
I think it got cancelled because it is a real startup now.
Let me explain! Put yourself in CEO's shoes: you have employees who love working for you but you need to get them a paycheck. From that perspective, it is easy to see that it is kind of crazy to enter into these lengthy government discussions that could possibly ruin your business and delay the entire industry! No thanks!!
If they're all really engineers who just love doin' the work, they will be a valuable contracting outfit doin' the exact same work for a steadier paycheck.
well, this is not some stupid web-app that can't harm anyone even if it tried (except from maybe overheating your smartphone). This can kill people, i can't seriously buy dangerous stuff from somebody that says something like this. What's next? Producing food without following proper hygiene regulations? Disrupting the medicine-industry by testing your headache-pills on your customers (those regulations are only slowing us down!)?