Experts see deep flaws in Tesla ‘Full-Self-Driving’ videos | Hacker News Reader

Experts see deep flaws in Tesla ‘Full-Self-Driving’ videos

washingtonpost.com

Are these the same experts that said that he'd fail at reusable rockets? This is all I heard from "experts" for a decade, and now SpaceX is how the United States gets to space.

The Washington Post has no credibility with me.

Different experts

Is Musk credible to you given his rather optimistic statements regarding FSD?

How about make up your own mind, there are thousands of videos online across all the different versions of the FSD Beta. It's very apparent that it doesn't work _yet_. Musk's claims is that based on the number of disengagements going down over time, if you plot that curve then it should be at L4 end of this year. I doubt it myself, there are just too many edge cases to fix, but give it 2 or 3 years and it might be more realistic. If their approach will work eventually or not, the research community seems pretty split on that. I think this year will be indicative though on the approach. If it hasn't improved pretty massively by years end, then it might stay L2 for a long time, or L4 geofenced to easier areas.

So I should make up my mind about what the person I asked thinks about something instead of simply asking them?

This list IIRC was posted by someone on HN which details some of the claims Musk has made in the past which suggest he isn't the most honest of people.

Abridged history of Elon's self-driving predictions:

2015: "We’re going to end up with complete autonomy, and I think we will have complete autonomy in approximately two years."

2016: "I really consider autonomous driving a solved problem, I think we are less than two years away from complete autonomy, safer than humans, but regulations should take at least another year."

2018: "I think probably by end of next year (2019) self-driving will encompass essentially all modes of driving and be at least 100% to 200% safer than a person."

2019: "We need to be at 99.9999..% We need to be extremely reliable. When do we think it is safe for FSD, probably towards the end of this year then its up to the regulators when they will decide to approve that."

2019: "I feel very confident predicting autonomous robotaxis for Tesla next year."

2020: "I am extremely confident of achieving full autonomy and releasing it to the Tesla customer base next year. But I think at least some jurisdictions are going to allow full self-driving next year."

2021: "Tesla Full Self-Driving will work at a safety level well above that of the average driver this year, of that I am confident. Can’t speak for regulators though."

The thing that stands out the most to me here is the constant drumbeat blaming the "regulators", as if the only thing keeping their buggy, shitty murdercars from going fully autonomous is those backwards gummermint bureaucrats.

I'd be willing to bet Elon huffs enough of his own farts that he too genuinely believes this, but I can definitely see the "real-life Tony Stark!!!111" fanboy cohort lapping that right up. The Venn diagram between Elon Musk fans and borderline-ancap laissez-faire extremists is practically a circle.

Yes, we all know this, but boy crying wolf doesn't mean the wolf will never come. I say disregard what Elon says and look for yourself. Unlike every other AV player, they're doing this in public and broad daylight (literally), so you don't have to take his word for anything. I think his statement in Dec is way more specific than in the past though, since he's mentioning the trend-line of disengagements over time, which is data they have internally that is difficult to gather from Youtube videos. Whether that trend will hold, or if it's even measuring the right thing is left to be seen, but it's a very different claim this time than previous years.

That said, again, I wish people would stop using his claims from years past as some kind of evidence that they will never make it work. What kind of logic is that?

> So I should make up my mind about what the person I asked thinks about something instead of simply asking them?

It's technology, you can make up your own mind about it, independent of what you think of Musk. If this is how you reason about the world, it's a very shallow way. "I think Apple will fail because Steve Jobs is an asshole that stole money from Wozniak", "Windows will fail because Bill Gates is a bully blah blah".

>It's technology

No it was a living human being that I asked the question to, not a piece of technology.

> Unlike every other AV player, they're doing this in public and broad daylight (literally), so you don't have to take his word for anything.

They're really not. It's happening in public. But even regulators are having to fight to pry every scrap of information about how it's happening out of Tesla's hands, and it is a constant fight to get even the tiniest fragments of data from them.

Musk claims disengagements are decreasing. I'm sure he'll be sure to have Tesla release that data, then, given Tesla's "corporate commitment to transparency". Any day now. Right after Trump releases his tax returns, no doubt.

When I see this list I think of the times that _I’ve_ overpromised and been unable to deliver. A deep pain in my chest every time I think of it and looking for ways out. At least I’ve been able to swallow my pride and explain that I can’t do what I said I could on the timeline I said it.

Musk meanwhile is the subject of a cult of personality and head of a public corporation… he doesn’t have that option. Does he just not feel it?

Musk has literally been claiming that self driving will work great by year end for like 5 years

I don't believe these experts are speculating on future performance. Rather they are pointing out that the thing isn't working now. Also, yes, different experts.

Well the experts are correct about point to point travel with starship. It's a pipedream and will never happen.

Literally no expert said that SpaceX would fail at reusable rockets.

The science behind rockets is literally decades old at this point. The U.S. designed reusable rockets back in the 1980s and never built them due to budget cuts.

Experts said they weren't sure if the cost of developing reusable rockets would be financially sustainable, given that SpaceX started while the U.S. still had a functional space shuttle program.

And...absent the NASA R&D contracts, SpaceX would not currently be sustainable on the basis of commercial launch activity.

I hear Heinz are working on reusable tin cans for their beans. Revolutionary.

Tesla has been selling FSD to paying customers since 2015. I think it is fair to look at the technology 7-years-later and see whether or not it does what it claimed to do.

All this article is... is a bunch of youtube videos and demonstration videos of common FSD errors. Given the video evidence, it seems reasonable to assume that FSD isn't working yet, despite claims made 7 years ago by Tesla and its marketing groups.

I am genuinely curious what percentage of time these videos represent relative to the total time all Tesla vehicles spend in FSD mode.

The goal should not be to avoid all errors but to minimize them without regression. Perfect is the enemy of good, so expecting perfection is irrational.

> Perfect is the enemy of good

Lying is also an enemy, and we're well into the "Tesla has been lying" about their $5000, $10,000, and $12,000 software.

Reply on news.ycombinator.com