Author: Robert Hyatt
Date: 09:54:52 06/02/98
Go up one level in this thread
On June 02, 1998 at 11:22:58, blass uri wrote: > >On June 02, 1998 at 10:04:31, Robert Hyatt wrote: > >>On June 02, 1998 at 09:05:17, blass uri wrote: >> >>> >>>On June 02, 1998 at 05:03:32, Inmann Werner wrote: >>> >>>>Alpha - Beta is silly? >>> >>>I am sure about it >>>grandmasters are better than computers and do not use the >>>Alpha- Beta >>>It is hard to find a good algoritham >>>and the target should be to find less silly algoritham >>>> >>>>Yesterday, I tried the LCTII Test Suite, position LCTFIN01. I was >>>>amazed, that INMICHESS (my program) needed depth 9 to solve the >>>>position. (running pawn) >>>> >>>> >>>>The position: Kc2/Pf2/kc4/Pg4/Pf5/Ph6/pb7/pf7/pg7/ph6 >>>>Best Moves: 1.f6 2.gxf6 3.f4 4.b5 5.g5 6.fxg5 7.fxg5 8.hxg5 9.g6 >>>>....->Queen >>> >>>I did not understand the position >>>white and black cannot have the same pawn at h6 >>> >>>Uri >> >> >>actually, I've been playing chess for 40 years now, and *I* use alpha/ >>beta when I play. IE once I analyze my first candidate, and then start >>on the second candidate I often quit quite quickly when I see something >>bad, without really analyzing to see "how" bad. So it is not that >>"foreign" a concept to chessplayers, they just don't realize they are >>doing the same thing... > >good players do also other things > >1)they know they should analyze a move >that they think it leads to an unclear position >(for example if they think to sacrifice material for initiative) > true.. but have you ever started analyzing such a move, notice that "hey, there's a smothered mate on the end of this" and go on to the next move? Or did you try to find out if your opponent had an even faster mate? >2) if they do not see a forced line >(they have no time to check all the possibilities) >they can use a statistics to decide >if they see in a position that every line they analyze >leads to their win when they play against themselves >they decide to go to the position. I don't believe human players do this. I personally analyze concrete variations when appropriate, and rely on intuition when it is not possible due to time constraints. Given the chance and the time, I'll analyze to reach a position I feel I can win. Given less time I'll resort to my intuition to recognize positions I should be able to win... and I will be wrong on occasion by doing so. That's one of the things that separates the GM from the IM and so forth. The GM's "intuition" is better. > >3)good players see there are moves that should not be analyzed >when in the alpha-beta we analyze every legal move to prove it is not >the >best. How do you *know* that a good player can "see that there are moves that should not be analyzed"?? How do you *know* that some quick parallel subliminal mental process doesn't look at *every* move and simply discard some based on pattern analysis? I've never seen any explanation of *how* humans play chess, although several (like DeGroot for one) tried to quantify this... So it is not quite kosher to fault the way programs "don't do it like humans" when we really have no idea how humans "do it" either... > >I think that it is very complicated to define exactly what they do >so the target should not be to tell the computer to do what good players >do but to do something better than computers do now to get closer >to what they do. I'd go farther. I'd state that it is *impossible* to tell a computer to do something like a human does it, when we have absolutely no understanding about how a human does "it"... > >Uri
This page took 0.01 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.