Hacker News Reader

Comment by Rygian | Hacker News Reader

When you say "having a tendency to," how often exactly does that happen?

Have you been able to watch Twitter videos of all driving situations FSD is exposed to, or was your research biased to "videos that get a lot of views," itself a subset of "FSD driving that was caught on camera"?

View on HN

I think it's fair to view Telsa's FSD as a single (very prolific) driver.

If there were multiple viral videos of a single person veering into oncoming traffic, I wouldn't care how many hours of video there was of them NOT veering into oncoming traffic there was tbh, I would still like to see some intervention.

No, that's not how you make reasonable policy. That's just listening to your gut reaction to such a video. Gut reactions do not work at this scale because this "single prolific driver" drives far more than a human in their lifetime. Therefore even rare events become common. Put a different way, as driven distances goes up the probability of having some scary event approaches 1. What we want to know at which rate it approaches 1. Twitter videos don't tell us that.

And are you advocating that the public policy be to allow these self driving tests without knowing how reckless they are?

No, I'm saying we should look at the actual data to make a decision to see if they're worse than human drivers and whether potential future payoffs of improved self-drivings are not worth running the experiment any further without modifications. Additionally we should also look at the first derivative, even if the numbers are bad today they might be on a trajectory where they'll be great soon.

So why does Tesla not release the actual data?

That would likely convince the DMV, and opponents. Just think of the win in terms of PR.

They should, no disagreement there. But as outlined above people should also look at such data carefully.

I am Elon fanboy number 987689. I’m a major in the Elon fanboy army, bootlicker division. I’m a mindless idiot who has no thoughts of his own. I am here to report to you from the scripted list of thoughts I am allowed to have from Tesla HQ. According to the Tesla chip in my brain, a real driver who drove as many miles as FSD would probably have many accidents too given the miles. FSD isn’t one driver because it’s constantly changing and updating, and not appreciating this is why people get it wrong so often — FSD of just a few months ago was night and day different and much worse. For each twitter video of FSD making a mistake there are 1000 videos of it demonstrating the fact that it is bar none the most advanced self driving system in the world by a mile. It’s just a fact and readily apparent if you simply watch the videos of the various systems. As long as your hand is kept on the wheel then FSD seems to be very safe even in the event of an intervention, most twitter videos they are doing it wrong frankly. It’s a fact that if they decide to destroy FSD with regulation that it will be a massive setback to self driving and also one of the coolest projects in the world right now. I would rather they didn’t.

Analogies often mislead. A single driver even a "very prolific" one is not driving hundreds of millions of miles a year.

Thought experiment: view all cars as a single imperfect product that kills 1.3 million people and injures tens of millions every year - vast majority due to poor driving. What's you solution? And make it fast - another 1.3 million will die this year.

Reply on news.ycombinator.com