alt Hacker News>I think it's fair to put the burden of proof here on Tesla.
That just sounds like a cope. The OP's claim is that the article rests on shaky evidence, and you haven't really refuted that. Instead, you just retreated from the bailey of "Tesla's Robotaxi data confirms crash rate 3x worse ..." to the motte of "the burden of proof here on Tesla".
https://en.wikipedia.org/wiki/Motte-and-bailey_fallacy
More broadly I think the internet is going to be a better place if comments/articles with bad reasoning are rebuked from both sides, rather than getting a pass from one side because it's directionally correct, eg. "the evidence WMDs in Iraq is flimsy but that doesn't matter because Hussein was still a bad dictator".
Replies
It's not a Motte and Bailey fallacy at all; it's a statement of a belief about what should be expected if something is to be allowed as a matter of public health and safety implications.
They're saying that Tesla should be held to a very high standard of transparency if they are to be trusted. I can't speak to OP, but I'd argue this should apply to any company with aspirations toward autonomous driving vehicles.
The title might be misleading if you don't read the article, but the article itself at some level is about how Tesla is not being as transparent as other companies. The "shaky evidence" is due to Tesla's own lack of transparency, which is the point of stating that the burden of proof should be on Tesla. The article is about how, even with lack of transparency, the data doesn't look good, raising the question of what else they might not be disclosing.
From the article: "Perhaps more troubling than the crash rate is Tesla’s complete lack of transparency about what happened... If Tesla wants to be taken seriously as a robotaxi operator, it needs to do two things: dramatically improve its safety record, and start being honest about what’s happening..."
I'd argue the central thesis of the article isn't one of statistical estimation; it's a statement about evidentiary burden.
You don't have to agree with the position that Tesla should be held a high transparency standard. But the article is taking the position that you should, and that if you do agree with that position, that you might say that even by Tesla's unacceptable standards they are failing. They're essentially (if implicitly) challenging Tesla to provide more data to refute the conclusions, saying "prove us wrong", knowing that if they do, then at least Tesla would be improving transparency.
I don’t think it’s a motte and Bailey fallacy because the motte is not well established. Tesla clearly does not believe that the burden of proof is on them, and by extension regulators, legislators.
So, there are two theories:
a) Teslas are unsafe. The sparse data they're legally obligated to provide shows this clearly.
b) Elon Musk is sitting on a treasure trove of safety data showing that FSD finally works safely + with superhuman crash avoidance, but is deciding not to share it.
You're honestly going with (b)? We're talking about the braggart that purchased Twitter so he could post there with impunity. To put it politely, it would be out of character for him to underpromise + overdeliver.
You're not replying to the author of the article.
The point is this: the article writer did what research they could do given the available public data. It's true that their title would be much more accurate if it said something like "Tesla's Robotaxi data suggests crash rate may be up to 3x worse than human drivers". It's then 100% up to Tesla to come up with cleaner data to help dispel this.
But so far, if all the data we have points in this direction, even if the certainty is low, it's fair to point this out.