Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Eliezer_Yudkowsky comments on Beware of Other-Optimizing - Less Wrong

79 Post author: Eliezer_Yudkowsky 10 April 2009 01:58AM

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

Comments (117)

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

Comment author: Eliezer_Yudkowsky 10 April 2009 09:58:02AM *  9 points [-]

Yeah, um... that might work, it might not. If you haven't been in CronoDAS's exact situation, be careful about dispensing this kind of advice, and really, be careful even then.

Comment author: Annoyance 10 April 2009 03:28:29PM 1 point [-]

No two situations are ever exactly alike. The only person ever to be in CronoDAS's exact situation is CronoDAS.

Comment author: cousin_it 10 April 2009 10:30:20AM *  0 points [-]

Why? I have no power of authority on this site. My advice is just incoming information with no persuasive weight. There's no bias stopping CronoDAS from evaluating it on its merits before acting.

Comment author: Eliezer_Yudkowsky 10 April 2009 02:38:18PM 15 points [-]

People who think they have no authority may be surprised by how much damage they can do just by talking in an authoritative tone. See believing everything you're told and cached thoughts. A simple "YMMV" might be enough to prevent that.

Comment author: arthurlewis 10 April 2009 01:28:02PM 6 points [-]

"There's no problem with you" can have a lot of persuasive weight as a response to a comment about what may or may not be a problem with CronoDAS. All things being equal, choosing the option that makes you look better is a fairly common bias.

Also, your status as a member of the lesswrong community and your tone, implying you've understood his particular situation, both lend you a slight boost in authority above random-person-on-the-Internet. I don't know whether this boost is trivial or not, but I think Eliezer is proposing a general rule which, although it will overshoot its mark, will guard against biasing the advisee, even in cases where you'd think it wouldn't be an issue. I believe there was an OB article on these sorts of rules, but I can't figure out what to search for.