nshepperd comments on Bayesians vs. Barbarians - Less Wrong

51 Post author: Eliezer_Yudkowsky 14 April 2009 11:45PM

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

Comments (270)

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

Comment author: Psy-Kosh 15 April 2009 07:04:06PM 0 points [-]

Well, that's more "what laws should there be/what sort of enforcement ought there to be?" I was more asking with regards to "what underlying rule is the Rational Way"? :)

ie, some level of being willing to do what you're told even if it's not optimal is a consequence of the need to coordinate groups and stuff. ie, the various newcomb arguments and so on.. I'm just trying to figure out where that breaks down, how stupid the orders have to seem before that implodes, if ever.

The morality one was easier, since the "obey even though you think you know better" thing is based on one boxing and having the goal of winning that battle/whatever. If losing would actually be preferable to obeying, then the single iteration PD/newcomb problem type stuff doesn't seem to come up as strongly.

Any idea what an explicit rule a rationalist should follow with regards to this might look like? (not necesarally "how do we enforce it" though. Separate question)

Even an upper limit criteria would be okay. ie, something of the form "I don't know the exact dividing line, but I think I can argue that at least if it gets to this point, then disobeying is rational"

(which is what I did for the morality one, with the "better to lose than obey" criteria.)

Comment author: nshepperd 22 November 2010 11:42:49PM 1 point [-]

How about this upper limit: when the outcome of (everyone) following orders would be worse than everyone doing their own thing, disobey.