Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Tesla calls self-driving technology 'failure,' not fraud

latimes.com

Post reply on HN
Print view View original

111–120 of 394 posts

Re: Tesla calls self-driving technology 'failure,' not fraud

#111

Post by meling »

0xbadc0de5 wrote: post #14

I've owned Tesla's since 2016. At no time did I base my purchases on what the cars might be able to do in the future, only what they can do right now. Nor did I see any compelling reason to donate $10k - $15k to their R&D effort. The cars were compelling enough on their own merits and remain so. The FSD fiasco really does a disservice to what is otherwise a revolutionary car company producing excellent automobiles.

Agree with this and I also haven’t paid for FSD. Except at some point Tesla offered FSD along with a computer upgrade for about $1500. I took the deal. But I didn’t pay for FSD when I switched to Model Y. Interestingly, I’ve been able to turn on the FSD even though I didn’t pay for it.

Re: Tesla calls self-driving technology 'failure,' not fraud

#112

Post by sgt101 »

slg wrote: post #101
gitfan86 wrote: post #96

Earlier quoted context omitted.

It is totally fine if you don't like Elon, but in the real world there isn't an EV that is cheap and available and has a good charging network. The ID4 is decent but that company was started by actual Hitler and Nazis, so if you are interested in not buying from companies that have a history of evil leadership that one isn't a good option. And gas cars are not good for CO2 emissions, so sometimes you have to accept t…

> The ID4 is decent but that company was started by actual Hitler and Nazis, so if you are interested in not buying from companies that have a history of evil leadership that one isn't a good option. This is a wild comment. The current leadership is slightly more important than the leadership 80 years ago. It is especially strange considering you could have made moral arguments against that company based off much mor…

Also I don't wear Hugo Boss because they made the uniforms for the SS.

And... Hugo Boss suits make me look like a wardrobe on legs... and are too expensive... but for sure the SS thing.

Re: Tesla calls self-driving technology 'failure,' not fraud

#113

Post by ghaff »

strawhatguy wrote: post #103
ncr100 wrote: post #67

Earlier quoted context omitted.

I feel our government should regulate this kind of speech more aggressively, to boost the safety of the public. The messaging by Tesla has been misleading, yes. There have been many other examples of couching date-slippage and over-promises with phrases like "no promises but Cybertruck|FSD|Model 3|Rockets|AI|Robotics|Neural-Link|End-Of-The-World should be here in two years." Announcing a capability, and importantly M…

No. I stopped at “the government should…” That never goes the way you think it will.

Advertising claims are in fact regulated. There just isn't a lot of enforcement but, at normal companies, legal teams are very conservative about making claims in advertising and marketing materials. If I claim our product is better than Brand X, you'd better believe legal is going to want to see the evidence and probably qualify the statement.

Re: Tesla calls self-driving technology 'failure,' not fraud

#114

Post by ethanbond »

superkuh wrote: post #32

I welcome the public's acknowledgement that self-driving vehicles are completely infeasible for even the medium term future. It's just unfortunate that the trigger for this change seems not to be anything about snow, or the inability to replicate human flocking on buried/unmarked roads during winter, etc, but instead more about the current political fallout from Musk's take-over of Twitter.

Tesla has been getting dragged for years for this stuff? It’d be way weirder if Man Does Bad Thing #2 should actually quiet Man Does Bad Thing #1.

He made himself the star, decided to be the villain, and reaps the reward of said villainy. Yes, everything is tainted by everything.

Re: Tesla calls self-driving technology 'failure,' not fraud

#115

Post by nostrademons »

I view self-driving cars (along with drones, ubiquitous encryption, and distributed manufacturing) as "post-WW3 technologies". The same way that atomic energy, computers, jet engines, and interstate highways were "post-WW2 technologies". These are things that could be done with current technology, but the current state of infrastructure & societal organization makes them uneconomical, and no single private firm has the market power needed to force through the social changes needed to drive adoption. However, they give a huge advantage to one side in wartime, where you are deliberately trying to kill people rather than avoid killing people. That gives the side who uses them the ability to win the war, which then gives them the political power to force through the societal & infrastructure changes needed for mass adoption. The war also gives the general populace widespread exposure to the new technologies, which makes them less scary to the winning side and drives political support for rearranging society to benefit from them (they still remain scary to the losing side, but they'll be dead).

We're at peace now, which makes self-driving cars kind of a non-starter. But that's not guaranteed to hold forever, even in the near future, nor is it even true in many places of the globe (witness Russia/Ukraine). It also may not be the point - remember that self-driving cars initially came out of a DARPA grant.

A lot of the technical problems with self-driving cars are easier to solve with widespread infrastructure changes. For example, you solve construction by mandating that construction diversions use smart traffic cones with a small transmitter that coordinates with the other cones in the area to signal which lane has priority, and then broadcasts that to oncoming self-driving cars. You solve pedestrians by having a proper street/road split [1], where pedestrians are banned from roads and cars are banned from streets. Heck, you can make the whole problem a lot simpler just by banning non-self-driving cars from roads.

[1] https://www.strongtowns.org/journal/2017/10/30/the-stroad

Re: Tesla calls self-driving technology 'failure,' not fraud

#117

Post by sidcool »

The argument is being taken out of context. Not saying that FSD is flawless. But Tesla isn't calling it a failure. They are saying it's not yet achieved. And it's a perfectly sane legal argument. Again, this does not mean Tesla is right here, just that the title is a misrepresentation of their argument.

Re: Tesla calls self-driving technology 'failure,' not fraud

#118

Post by holmesworcester »

cloudking wrote: post #59

FSD beta is fairly good, certainly better than the standard Autopilot. It's definitely not perfect, but so far for basic driving from point A to B it seems to work well. I can only see the tech improving as it gathers more data and feedback. It's a ML training data problem at this point, no?

It's shockingly good and definitely worth 20% of the value of a car, for me. Two main reasons:

1. Your body doesn't feel tired after driving, because besides the effort of a hand resting on the bottom of the wheel, your body can be completely relaxed the vast majority of the time.

2. When FSD is on, you can completely stop paying attention to navigation-related decisions. You still have to pay attention to driving and intervene sometimes (most interventions are just a tap on the gas to give the car the confidence to go, because it's biased to caution) but every decision about when and where to turn is completely taken care of for you. This frees up a lot of mental overhead when you're in totally unfamiliar territory, including overhead that, without FSD, would pull your attention away from driving, making you less safe.

I didn't get the magic of #2 until I used it on an unfamiliar drive home in New England backroads with lots of turns and realized I was paying attention to the road and my conversation with my partner, and none at all to where we were going. Then we were home.

The one non-ideal thing about FSD Beta right now is that it's really easy to lose it for an arbitrarily long period of time.

What happens is it's fairly easy to have your hand on the wheel on a straight stretch of highway and still get the nag. And there's a small but non-zero false negative rate when you wiggle the wheel to satisfy the nag but your wiggle is insufficient. And because it becomes so habitual and because the car beeps at you for other reasons too, there's a fairly high rate (at least for me) of wiggling to satisfy the nag and failing to notice that you didn't wiggle enough.

The odds are small but after months of long drives it's pretty easy to accumulate the 5 "strikes" that trigger loss of FSD Beta without violating the intent of the nag, i.e. taking your eyes off the wheel or the road.

Re: Tesla calls self-driving technology 'failure,' not fraud

#119

Post by danaris »

BoorishBears wrote: post #107
mjhay wrote: post #88

Earlier quoted context omitted.

Real FSD, where it is not necessary to have a human ready to take the controls at any time, isn't something you can just get more training data for. It's just that current ML methods aren't up for the task. Basically, it turns out you need to have something that can think , and ML is nowhere near that.

That is what no one at all wants. Tesla seems to be taking the big ball of ML approach, and that relies on a model that can practically think, and that in turn means you're essentially waiting on AGI. Which is ridiculous. Instead you break down the problem that is driving. Throw pieces of ML at specific problems, like traffic light classifiers, but have a much more structured approach to the high level decisions that…

"No one at all" wants a car where you can just get in, input the address, then ignore the wheel and what's going on outside until you get there?

I think many, many people would beg to differ.

Re: Tesla calls self-driving technology 'failure,' not fraud

#120

Post by lawn »

fnordpiglet wrote: post #97

I’m been on the fsd beta for a long time now and I have to say it’s remarkably improved over the last year. If it continues to improve at this rate for another year it’s going to be pretty useful. I use it to drive around seattle (which is a difficult driving place with randomly followed infrastructure standards and randomly driving drivers and randomly parked cars) and it used to be a death defying gripping of the s…

That you're surprised that it managed drive door to door is telling. And not in the positive way you want to make it seem.
Post reply on HN