Mark Kelly has responded:

The intention and awareness of the errors is a big part of this. I expect an intern to seek help in things they’re not sure of and to flag them for extra review if they’re still not sure.

I’ll second the Joplin recommendation. I often expect FOSS solutions to take a bit more maintenance, but I’ve been using Joplin for about 5 years and I’ve put in 0 extra effort after the initial setup. I’ve used it on Windows, iOS, and MacOS

Years ago on Stack Overflow someone edited one of my answers to “correct” my seed value from 47 to 42.

I was on there for a couple months. Then my client unexpectedly stopped working and I realized I didn’t miss it at all…

This looks fantastic—wish I had something like it 10 years ago!

The second edition turned into *Regression and Other Stories*, but that’s mainly a second edition of the first half of ARM—the introductory part. I think they maybe think of Bayesian Data Analysis as covering much of what they want to cover from the second half.

Okay but Critical Bowl would be a great name for a Gastronauts episode with Crit Role guest judges

My understanding is that, theoretically, BIC is one of the least well-founded criteria. But just like Danielle said AIC is unreasonably effective, when I’ve had a lot of variables and have been worried AIC includes too much, the switch over to BIC is simple and also seems surprisingly effective.

This is really beautiful.

Maybe the Cruel Intentions experience alienated and radicalized Clarissa Mao