Author: Jeremiah Penery
Date: 20:51:30 09/08/01
Go up one level in this thread
On September 08, 2001 at 09:18:53, Mark Young wrote: >I don't know how we can tell for sure that Deep Blue's moves are the best >moves. For that matter, how can you tell which of Deep Blue's moves are not the best moves? Maybe if you give Fritz/Junior/whatever a million hours to search something, it will finally choose the same move as Deep Blue. What if Fritz and Deep Blue both choose the same move, but it's still a bad move? If they pick different moves, how can you be sure which move is the "better" one, without some deep human analysis to back up the machine analysis? And even then, GM analysis is refuted often enough to make this not even 100% sure. I guess what I'm really trying to say is that this is a neat experiment, but I don't think it can prove anything. There are lots of possible reasons why programs pick a move. Just because Fritz finds a move faster than Deep Blue doesn't make Fritz better; just because Fritz finds the same move as DB doesn't make it the best move; just because Fritz fails to find a Deep Blue move doesn't make DB's move a bad one; and just because Fritz doesn't find Deep Blue's move doesn't mean Deep Blue is a lot better. In some sense, comparing Fritz to Deep Blue is like comparing Apples to Oranges. Sure, they're both fruits, but they have very different qualities. Some people like the apples more, and some people like oranges. Without some quantitative data, you will never be able to prove which is better.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.