Do you read that report to suggest that Chapin is getting less done now? He mentions in it having three jobs, and my impression from the kinds of things he posts on Twitter has been that he seems productive, though of course one that can be fake (but if so he seems at least productive at faking productivity).
It is hard to tell. Some of Chapin’s jobs like the coaching stuff are pretty much impossible to judge externally: we couldn’t tell if they even exist short of hiring him personally. I can only say that I feel like I’ve seen his Substack writings discussed less post-PNSE (but this is also obviously confounded by, among other things, Twitter attacking Substack over a similar time period and what feels like a general Internet-wide collapse of linking/sharing); and that Nick Cammarata says he’s gotten far more productive but his DL interpretability work outputs look the same over time to me and I see no changepoint.
He may have decided to revise it all. He left a long reply to my followup question about whether he had read the PNSE paper before he wrote this post (since his first reply was ambiguous, and someone could reasonably wonder if the PNSE paper had framed his expectations and so this anecdote is not as parallel & independent confirmation of the PNSE syndrome as it looked), but then by the time I clicked on the link in the email version, his reply (but not the post) had been deleted.
Do you read that report to suggest that Chapin is getting less done now? He mentions in it having three jobs, and my impression from the kinds of things he posts on Twitter has been that he seems productive, though of course one that can be fake (but if so he seems at least productive at faking productivity).
It is hard to tell. Some of Chapin’s jobs like the coaching stuff are pretty much impossible to judge externally: we couldn’t tell if they even exist short of hiring him personally. I can only say that I feel like I’ve seen his Substack writings discussed less post-PNSE (but this is also obviously confounded by, among other things, Twitter attacking Substack over a similar time period and what feels like a general Internet-wide collapse of linking/sharing); and that Nick Cammarata says he’s gotten far more productive but his DL interpretability work outputs look the same over time to me and I see no changepoint.
Huh, the post is down now, and also not available on the internet archive or archive.is.
He may have decided to revise it all. He left a long reply to my followup question about whether he had read the PNSE paper before he wrote this post (since his first reply was ambiguous, and someone could reasonably wonder if the PNSE paper had framed his expectations and so this anecdote is not as parallel & independent confirmation of the PNSE syndrome as it looked), but then by the time I clicked on the link in the email version, his reply (but not the post) had been deleted.
There is an archived version here.
No, just a general point. I do not know Sasha Chapin.