NancyLebovitz comments on Harry Potter and the Methods of Rationality discussion thread, part 4 - Less Wrong

3 Post author: gjm 07 October 2010 09:12PM

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

Comments (649)

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

Comment author: shokwave 29 October 2010 06:03:09AM *  7 points [-]

The more rational someone is, the less you can trust them. The less rational someone is, the more you can trust them.

I think in this post when you say 'trust' you really mean 'predict'. A trivial counterexample: the more rational someone is, the more I can trust them to be free of errors in their reasoning. And it IS easier to predict a religious zealot staying religious, or predict that a bigot will remain bigoted, than it is to predict a rational agent attempting to maximize their utility (especially if you're an obstacle to their utility).

Is it reasonable to think that people have evolved to be less-than-optimally rational?

Well, yes, if there was some shortcut that gave the mostly-optimal answer, or gave the optimal answer most of the time, and gave it in a significantly faster time than optimal rationality. The common example is, I think, reacting to the presence of a lion. Abject, heart-pounding, run-for-your-life terror is not optimally rational (it generally precludes climbing a tree) but it gives a mostly-optimal answer in a much shorter time than attempting to reason out the optimal course of action.

Comment author: NancyLebovitz 29 October 2010 01:12:06PM 4 points [-]

This gets to one of the Hard Problems, both for FAI and a great deal of life. How can you tell who can be trusted to do a good job of taking your interests into account?