There is a crisp, substantive and comparatively readable comment exchange at LessWrong on a post in this sequence between Paul Christiano and Eliezer Yudkowsky.
These comments might interest those who liked the topic of this series, but, like myself, found it long and difficult to read.
Also, uh, this is not consistent with EA forum norms and not virtuous, but it might be satisfying to those who put a lot of effort into reading the posts, because out of (maybe related) frustration, Paul Christiano turns into some spicy hummus:
Your comment makes it sound like Paul is only arguing in that style in the LW comments, but there’s also plenty of it in this EA forum post, FWIW.
Paul’s frustration is that Eliezer kept repeating the claim “Paul’s view should have been surprised by AlphaGo” and Paul is like “No it wasn’t, particularly, but if you think you know where my view should be surprised (and wrong), let’s see it” – but then Eliezer doesn’t have easily identifiable predictions. Paul’s saying that Eliezer should either stop treating it as though there’s now a track record in his favor (with AlphaGo), or produce bets on similar things coming up.
There is a crisp, substantive and comparatively readable comment exchange at LessWrong on a post in this sequence between Paul Christiano and Eliezer Yudkowsky.
These comments might interest those who liked the topic of this series, but, like myself, found it long and difficult to read.
Also, uh, this is not consistent with EA forum norms and not virtuous, but it might be satisfying to those who put a lot of effort into reading the posts, because out of (maybe related) frustration, Paul Christiano turns into some spicy hummus:
E.g.
Also:
Your comment makes it sound like Paul is only arguing in that style in the LW comments, but there’s also plenty of it in this EA forum post, FWIW.
Paul’s frustration is that Eliezer kept repeating the claim “Paul’s view should have been surprised by AlphaGo” and Paul is like “No it wasn’t, particularly, but if you think you know where my view should be surprised (and wrong), let’s see it” – but then Eliezer doesn’t have easily identifiable predictions. Paul’s saying that Eliezer should either stop treating it as though there’s now a track record in his favor (with AlphaGo), or produce bets on similar things coming up.