Author: Bruce Moreland
Date: 17:05:53 11/28/00
Go up one level in this thread
On November 28, 2000 at 15:58:31, Mogens Larsen wrote: >On November 28, 2000 at 13:44:15, Robert Hyatt wrote: > >>I don't have any "suspicious speculation" in this case. Only a strong feeling >>that "right move, wrong reason" is not convincing me of very much. IE on ICC >>the other night we had a long discussion about a move Crafty had played. The >>GM (Don't remember who it was not, but not Roman) said "Rd4 was beautiful... >>I am very impressed that the program saw it as it led to a crushing position >>for it." I looked at the log and told the group discussing this "The score was >>very bad for it at that point... it actually thought that move was best, but >>that it was losing the game... until it failed high on the 'pondering search' >>after playing it." We all agreed that it was just "lucky" there that it found >>a move that turned into a win, even though it had no idea when it played it how >>it would turn out... >> >>That was my point. Yes it played the right move. No it didn't understand >>why. It played it fully expecting to lose. It could have played any of >>_several_ moves and _still_ expected to lose. It just _happened_ to pick the >>right one. In this case, for a reason (weak opponent pawn) that had _nothing_ >>to do with how the game actually progressed and was won by that move. > >The problem with this argumentation is that "right reason" only holds if you're >able to distinguish between moves by a rising score at a shallow depth. So it's >basically useless in most cases. Crafty played Rd4 because it evaluated the move >as being best. Labelling the reasons as either wrong or right for the chosing >the initial move is wrong. > >If the PV at the time of the move was close to what happened in the game then I >consider it to be chosen for the right reason, even though the score may have >been inaccurate due to insufficient depth. > >Either way, using human perception of "right" and "wrong" reasons based on score >alone is very simplistic and useless IMHO. > >Mogens. I call these kinds of solutions "sniffs". The program gets some idea that it has found a good move, but it doesn't know why. The world to it is a bunch of holes and mountains, but it doesn't always know how deep the holes are or how high the mountains are, since its vision is very poor. Sometimes a program will choose a move because it's the only one that doesn't lead into a hole of indeterminate depth. Maybe the move is a crater ring surrounding a really deep hole that can't be avoided. Maybe it's Mount Everest. Other times it may have selected the move because it thought all the moves were equally good or bad, or it saw a mountain that didn't really exist, and blundered onto one that did. When I'm counting solutions in big suites, I don't try to differentiate, I just want to grab big statistical handfulls. But I usually wouldn't talk about how smart my program is to solve it if it's a "white to play and win" position and my program scored it as -1.8. bruce
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.