Author: Dann Corbit
Date: 11:31:46 09/17/01
Go up one level in this thread
On September 15, 2001 at 02:01:48, Jeremiah Penery wrote: >On September 14, 2001 at 17:21:45, Dann Corbit wrote: >>On September 14, 2001 at 16:25:00, Rafael Andrist wrote: >>>On September 14, 2001 at 14:33:42, Dann Corbit wrote: >>>>Those look pretty strange to me, but any move that leads to a certain win must >>>>be considered a "best move." >>> >>>I think those moves should be considered as "good moves", but only one move is >>>the "best move" (except if there are different pathes with the same mate >>>distance). >> >>I disagree completely. >> >>If the game theoretic value of both moves are identical, then the value of both >>moves is identical. >> >>A move that wins in 2 is not better than a move that wins in 45. Both moves >>win. >> >>The 2 move win might be prettier, but it isn't any better. > >You're thinking like a computer here, Dann. This suite was written for >_humans_, and was never intended for a computer. Maybe that was the original intention, but computers play it as much or more than humans do. >Some positions are obviously >winning, with almost any move played. But the point is for the _human_ to find >the "best" move - the move that wins most easily, the win that is "prettiest", >or the move that easily reduces a lost game into a drawn one - even when there >are other moves that do the same thing. Sometimes, an alternative is prettier. Sometimes, the original choice is just plain wrong. There were at least two cases like that in the original WAC that I know of. >For a computer, it's fine to consider any move that gets +3 as a "winning move" >(in most cases). The computer sees that it's winning, and probably will go on >to win the game. > >It all depends on how you're using the WAC suite. But I think if you start >putting all moves that a computer might consider "winning" into the solution >set, for some positions you'll have practically any move. In that case, what >does it show that a computer can find one of them? That the computer will win the game if it plays that move. That's the goal of chess, IIRC. ;-) >The suite will become even >less valuable than it is currently. Debugging a test suite never causes it to lose its value. The value is only enhanced. >YMMV, of course. :) Indeed. ;-)
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.