Whenever I start thinking about very unlikely hypothetical situations -- often self-glorifying to an extemely embarrassing extent -- I try to think about abstract/qualitative decision theory problems instead: things that I'm very unlikely to make any useful progress on, but are about as likely as the crazy hypothetical situation I was dreaming up before, and with much much much higher expected utility. I like thinking about such problems, too; the only reason I normally don't is because it's become so habitual throughout my life to be incredibly narcissistic. When I consciously think about e.g. game theory for superintelligence, I find the time spent much more rewarding. This has gotten to the point where I might be at a friend's house chilling by the pool and, instead of feeling bored or whatever, I'll close my eyes and remember interesting problems I've encountered and try to solve them. I'm not sure, but I think that if more really smart people picked up this habit then there would be an appreciable growth in scientific output.

More directly to the point, I think that very small hacks like the above would allow one to approach MTP towards their true goals than a very thorough analysis of exploration/exploitation in one's daily life.

I liked this post.

(No spellcheck used for this comment, I sincerely apologize for any errors.)