I think the downvotes are coming because people don't realize you're doing the exercise at the start of the post, and rather think that you're making these claims after having read the rest of the post. I don't think you should lose karma for that, so I'm upvoting; but you may want to state at the top that's what you're doing.

It's a very unusual disclaimer that speaks well of the post.

The default journalistic practice at many outlets is to do an asymmetric search once the journalist or editor decides which way the wind is blowing, but of course nobody says this in the finished piece.

Ben is explicitly telling the reader that he did not spend another hundred hours looking for positive information about Nonlinear, so that we understand that absence of exculpatory evidence in the post should not be treated as strong evidence of absence.

Which is why I said that the probabilities are similar, rather than claiming the left side exceeds the right side.

I'm surprised (unless I've missed it) that nobody has explicitly pointed out the most obvious reason to take the responses of the form "Kat/Emerson/Drew have been really good to me personally" as very weak evidence at best.

The allegations imply that in the present situation, Kat/Emerson/Drew would immediately tell anyone in their orbit to come and post positive testimonials of them under promises of reward or threat of retaliation (precisely as the quoted Glassdoor review says).

P(generic positive testimonials | accusation true) ≈ P(generic positive testimonials | accusation false).

The only thing that would be strong evidence against the claims here would be direct counterevidence to the claims in the post. Everything else so far is a smokescreen.

Plenty of "weird and atypical" things aren't red flags; this one, however, is a well-known predictor of abusive environments.

I believe that a commitment to transparently reward whistleblowers, in cases where you conclude they are running a risk of retaliation, is a very good policy when it comes to incentivizing true whistleblowing.

Ben, I want to say thank you for putting in a tremendous amount of work, and also for being willing to risk attempts at retaliation when that's a pretty clear threat. 

You're in a reasonable position to take this on, having earned the social standing to make character smears unlikely to stick, and having the institutional support to fight a spurious libel claim. And you're also someone I trust to do a thorough and fair job.

I wish there were someone whose opportunity cost were lower who could handle retaliation-threat reporting, but it's pretty likely that anyone with those attributes will have other important opportunities.

Re: QB sneaks, the obvious tradeoff that's not factored into the naive model is the risk of an injury to the QB. Running the QB can be worth it if the expected points added are high enough (goal line or pivotal 4th and 1, or if your QB can often pick up 10+ yards on a draw), but I doubt you want to roll those dice on a midfield 2nd and 4 sneak in the first quarter of a regular-season game.

(God, American football is a beautiful game. I miss the days when I could enjoy it because I didn't know about the CTE.)

I worked in the AI/ML org at Apple for a few recent years. They are not a live player to even the extent that Google Brain was a live player before it was cannibalized.

When Apple says "AI", they really mean "a bunch of specialized ML algorithms from warring fiefdoms, huddling together in a trenchcoat", and I don't see Tim Cook's proclamation as anything but cheap talk.

Re: mileage, if the car is a Tesla, it's not an accident: they set their numbers to the maximally optimistic ones at every stage (without regard for e.g. temperature) and have a team dedicated to diverting people who call in about the reliable inaccuracy.