Nick_Tarleton comments on Practical Advice Backed By Deep Theories - Less Wrong

42 Post author: Eliezer_Yudkowsky 25 April 2009 06:52PM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (112)

You are viewing a single comment's thread. Show more comments above.

Comment author: Nick_Tarleton 26 April 2009 11:41:48PM 0 points [-]

It doesn't particularly apply in this case, since the argument is that randomness is the best available option for now, because intelligence doesn't work yet for this case.

But pjeby isn't even saying that – even reading completely random books, which AFAICT he doesn't advocate, invokes a powerful optimization process (writers and publishers).

Comment author: Vladimir_Nesov 26 April 2009 11:44:49PM 0 points [-]

You always do the random thing relative to the options you are given. That doesn't change the problem, as far as I can see, just applies it to a different situation

Comment author: Nick_Tarleton 26 April 2009 11:47:34PM 0 points [-]

Point taken; still, different from my very literal interpretation of letting a random number generator decide what you need.

Comment author: Vladimir_Nesov 26 April 2009 11:58:39PM *  0 points [-]

You can't literally make only random actions. You can't make random muscle movements. You may use random long-term goals, which can be analogized with being a fanatic, or middle-term goals, analogy with a crazy person, or random short-term goals, analogous to being clinically mad. In any case, whatever I could mean by random action, it's necessarily already quite abstract, selected from few intelligent options.