It appears that Mr. Khan’s expressed preferences are very likely transitive, but it is difficult to see how this could be argued regarding Mr. Bagehot’s preferences. Unlike Mr Khan, Mr Bagehot makes his own desires dependent upon the expressed opinions of those around him.
Eliezer advocates the “rationality is about winning” position, as timtyler note sin his reply to you.
And this is actually a Humean point. The idea is that passion is about what you want i.e. want qualifies as winning and rationality is about getting what you want i.e. how to go about winning.
As for Mr Bagehot’s preference set, it’s true that transitivity is a necessary condition for rationality because an agent with intransitive preferences has no coherent utility function.
However, I don’t think that’s an issue here. Bagehot’s preferences are dependant on others, but that doesn’t make them intransitive. I fact there’s no way to test for intransitivity with fewer than three alternatives to choose from.
Is it permissible to assert that rationality is about winning? Hume might argue that “winning” is about ends, whereas “rationality” is about means.
You’re right. And the “rationalist win” slogan gets annoying for that reason—a good point but not technically correct. There’s something along the lines of ‘most likely’ or ‘maximise’ that is missing.
Is it permissible to assert that rationality is about winning? Hume might argue that “winning” is about ends, whereas “rationality” is about means.
-- D. Hume
However, it is sometimes argued that the word “rational” can be applied to ends, as well as means.
--M. Phipps in “Must Rational Preferences Be Transitive?”
It appears that Mr. Khan’s expressed preferences are very likely transitive, but it is difficult to see how this could be argued regarding Mr. Bagehot’s preferences. Unlike Mr Khan, Mr Bagehot makes his own desires dependent upon the expressed opinions of those around him.
Eliezer advocates the “rationality is about winning” position, as timtyler note sin his reply to you.
And this is actually a Humean point. The idea is that passion is about what you want i.e. want qualifies as winning and rationality is about getting what you want i.e. how to go about winning.
As for Mr Bagehot’s preference set, it’s true that transitivity is a necessary condition for rationality because an agent with intransitive preferences has no coherent utility function.
However, I don’t think that’s an issue here. Bagehot’s preferences are dependant on others, but that doesn’t make them intransitive. I fact there’s no way to test for intransitivity with fewer than three alternatives to choose from.
That’s a Yudkowsky theme:
http://lesswrong.com/lw/7i/rationality_is_systematized_winning/
http://wiki.lesswrong.com/wiki/Rationalists_should_win
You’re right. And the “rationalist win” slogan gets annoying for that reason—a good point but not technically correct. There’s something along the lines of ‘most likely’ or ‘maximise’ that is missing.
Wow. I’ve never comparison to Gengis Khan used as a way to make someone sound fickle. :)
Untrue. Bagehot desires pleasure, and pleasure is dependent on the opinions of those around him. This is consistent.