> You're making guesses about my standards that aren't warranted at all. They absolutely can be met. First off I'm specifically worried about large objects, a meter or larger.
I've personally watched my car go around trash cans and bikes in the road, and (obviously) stop for halted vehicles. So, you'll abandon this argument and concede the point? I suspect you won't, because my anecdote isn't enough for you, and those obstacles aren't sufficient proof, etc...
I think I'm more right than you want to admit. There's nothing I can say here to change your mind, and so coming here and demanding "evidence" isn't really an argument in good faith.
And an edit just to pick on this bit:
> Okay, if FSD can manage 1 billion miles of freeway driving with [emphasis in original] zero or one crashes into large objects, then I'll be convinced they're solving the problem.
That's simply a ridiculous argument. That level of safety is way, way, WAY beyond anything you get from existing transportation systems of any kind, period. A quick google shows that in that "one billion miles on the freeway", you'd expect not merely the "one accident" you're demanding, but in fact TWENTY FATALITIES (I couldn't find statistics for mere collisions, but needless to say it's going to be at least an order of magnitude or two higher).
So basically you're sitting here and glibly demanding that this product be 200x safer than its competitors before you'll consider it acceptable... and getting huffy when I call you unserious?
> So, you'll abandon this argument and concede the point? I suspect you won't, because my anecdote isn't enough for you, and those obstacles aren't sufficient proof, etc...
Are you trolling with this?
I'm asking if it can reliably do that. Of course I need more data than one person can collect. I am not being unreasonable to ask for more data than your personal experience. Most people with autopilot never saw this problem either.
And specifically I need to know about freeway speeds, because speed is an important factor.
There is plenty you could do to change my mind. If you link to something published by tesla or a government body showing autopilot and FSD accident rates by type then that would be more than enough.
> So basically you're sitting here and glibly demanding that this product be 200x safer than its competitors before you'll consider it acceptable
I'm not asking for a lack of fatalities. I'm asking for a lack of hitting stationary vehicle-scale objects on the freeway.
Do you think that particular kind of accident is responsible for a majority of fatalities, or something? My expectation is that it's a very rare kind of accident and I also feel like it's a good canary.
Also the freeway fatality rate is about 5.4 per billion miles, not 20.
It's so amazing how this discussion goes every single time.
> I'm not asking for a lack of fatalities. I'm asking for a lack of hitting stationary vehicle-scale objects on the freeway.
Exactly! You've constructed an impossible gateway to understanding; I either find a statistic to fit exactly your imagined failure mode or... I'm wrong, and you don't need to change your opinion.
I'm, sorry, I truly am, that I don't have a statistic to hand you showing the frequency with which Tesla vehicles with FSD beta hit 1m+ stationary obstacles on high speed roadways. I don't. And I won't, and likely never will.
So, again getting to the point upthread: you're safe. You can't lose this argument framed like that, and I concede that point. I'm just saying that that's not a very serious position to take if you're actually interested in genuine safety using metrics that other people care about.
> Also the freeway fatality rate is about 5.4 per billion miles, not 20.
Not the headline I saw immediately, but sure. That sounds plausible too. The fact that you want to claw back an error factor of 200 by 3.5x is also good evidence that you aren't taking the discussion seriously.
> The fact that you want to claw back an error factor of 200 by 3.5x
I don't. Again, I wasn't talking about total accident rate at all. I was talking about a much smaller number. The "200" is nonsense and that was just another reason it's wrong.
> So, again getting to the point upthread: you're safe. You can't lose this argument framed like that, and I concede that point. I'm just saying that that's not a very serious position to take if you're actually interested in genuine safety using metrics that other people care about.
Ugh. Look, I can wait for general safety statistics, but it will take longer. Those will exist, and they can convince me if they're within 2x of humans.
Maybe it's unfair for me to want specific statistics here, but it shouldn't be so hard to get them.
But it's just as unfair for you to act like a single person's anecdotes are enough. You can't just say it's "obviously" safe and treat that like a real argument. Of course the discussion is going to go the same way every time if that's the level of evidence you expect people to accept.
And I thought you were claiming that no evidence could convince me because I'm unreasonable. If your real claim is "nobody has bothered to collect much evidence, therefore there is no way to convince you without doing that job" then yes I agree and I don't think that's my problem.
I've personally watched my car go around trash cans and bikes in the road, and (obviously) stop for halted vehicles. So, you'll abandon this argument and concede the point? I suspect you won't, because my anecdote isn't enough for you, and those obstacles aren't sufficient proof, etc...
I think I'm more right than you want to admit. There's nothing I can say here to change your mind, and so coming here and demanding "evidence" isn't really an argument in good faith.
And an edit just to pick on this bit:
> Okay, if FSD can manage 1 billion miles of freeway driving with [emphasis in original] zero or one crashes into large objects, then I'll be convinced they're solving the problem.
That's simply a ridiculous argument. That level of safety is way, way, WAY beyond anything you get from existing transportation systems of any kind, period. A quick google shows that in that "one billion miles on the freeway", you'd expect not merely the "one accident" you're demanding, but in fact TWENTY FATALITIES (I couldn't find statistics for mere collisions, but needless to say it's going to be at least an order of magnitude or two higher).
So basically you're sitting here and glibly demanding that this product be 200x safer than its competitors before you'll consider it acceptable... and getting huffy when I call you unserious?