Re: When Programs Benefit

From: Wei Dai (weidai@eskimo.com)
Date: Fri Jun 14 2002 - 15:55:01 MDT


On Fri, Jun 14, 2002 at 02:14:21PM -0700, Hal Finney wrote:
> Unfortunately this doesn't seem like a situation where you can try both
> and see what you like better. Suppose I am in a sim and someone offers
> to run me twice tomorrow, on two computers in parallel. So I try it out,
> and afterwards I am asked if I liked it better. But I don't see how I
> could perceive a difference from the extra run. So this line of analysis
> does not seem to shed much light on whether I should value reruns,
> since I can't perceive them, or at least can't remember perceiving them.

What you need to do is have a third copy observe your two copies run in
parallel, and have the third copy decide whether it's better. If the third
copy can't decide either, then I propose that you simply don't care very
much. To take the music analogy again, if you can't decide between
Beethoven and Mozart after hearing them both, then you're not likely to be
willing to spend much more resources on one over the other.



This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:14:47 MST