JamesAndrix comments on Logical Rudeness - Less Wrong

65 Post author: Eliezer_Yudkowsky 29 January 2010 06:48AM

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

Comments (203)

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

Comment author: JamesAndrix 29 January 2010 04:01:31PM 0 points [-]

I'm not saying anything about the actual program or results, but that form of argument might be valid in this case. In a 'kill two birds with one stone' kind of way.

Looking at the efficiency of any one thing may not be the best strategy if you care about lots of things.

Comment author: ciphergoth 29 January 2010 04:14:54PM 1 point [-]

Sure, but (again in ignorance of the actual program) there should be at least one point on which you're prepared to defend its efficacy.

Comment author: Bo102010 30 January 2010 03:23:27AM 1 point [-]

Yeah, exactly. The problem with the outline above is that in step 2 they acknowledge that the program doesn't do A well. But then in step 4 they act like they never conceded that.

Comment author: JamesAndrix 29 January 2010 04:49:06PM 1 point [-]

Really, we should be trying to look at the total effects of any given expenditure. (including where we get the money from in the first place, if that's variable)

But to simplify:

If spending $100 in a certain way benefits 10 parties as much as giving them $20 would, each could argue that it would be more efficient (by a factor of 5!!) to just give them the $100. But if you care roughly equally for all the parties, it would really be only half as good.

Comment author: ciphergoth 29 January 2010 05:21:33PM *  3 points [-]

The only such defence worthy of our attention is one where the speaker is prepared to explicitly state a guess at the dollar value of each advantage and show that the sum is greater than the cost spent.

Comment author: Eliezer_Yudkowsky 29 January 2010 07:58:52PM 2 points [-]

Or where the sum advantage is obvious compared to the next best alternative, without formally computing expected value.

Comment author: JamesAndrix 29 January 2010 05:55:24PM 0 points [-]

Agreed