AFAICT nobody would actually read all 127 pages of the report, and most potential reasons for writing the report to me seem better served by faster feedback loops and significantly smaller research artifacts.
Often I write big boring posts so I can refer to my results in shorter, more readable posts later on. That way if anyone cares and questions my result they can see the full argument, without impairing readability on the focal post.
My model is that a text like this often is in substantial parts an artifact of the author’s personal understanding. But also, my model of Open Phil employees totally read 100-page documents all the time.
I notice that Joe Carlsmith dropped a 127 page paper on the question of deceptive alignment. I am confused; who is the intended audience of this paper?
AFAICT nobody would actually read all 127 pages of the report, and most potential reasons for writing the report to me seem better served by faster feedback loops and significantly smaller research artifacts.
What am I missing?
Often I write big boring posts so I can refer to my results in shorter, more readable posts later on. That way if anyone cares and questions my result they can see the full argument, without impairing readability on the focal post.
My model is that a text like this often is in substantial parts an artifact of the author’s personal understanding. But also, my model of Open Phil employees totally read 100-page documents all the time.