I really appreciated how you actually worked through the paper and talked to the authors about it. It was like daylight compared to the fog that David Shapiro brought.
It is also great - and terrifying, of course - to see predictions made by the AI safety community from a long while ago being confirmed.
I like Twitter Beefs; I liked this as well. I think that you can do both if you have the time/energy. But the paper analysis would obviously need to be saved for important papers, imo.
(Oh: I also got a kick of seeing Thurstone being used as I have used the same pairwise comparison system for ranking maths assignments - it made me feel a tiny bit connected to the paper and to the smart people behind it)
I really appreciated how you actually worked through the paper and talked to the authors about it. It was like daylight compared to the fog that David Shapiro brought.
It is also great - and terrifying, of course - to see predictions made by the AI safety community from a long while ago being confirmed.
I like Twitter Beefs; I liked this as well. I think that you can do both if you have the time/energy. But the paper analysis would obviously need to be saved for important papers, imo.
(Oh: I also got a kick of seeing Thurstone being used as I have used the same pairwise comparison system for ranking maths assignments - it made me feel a tiny bit connected to the paper and to the smart people behind it)