Author: allan johnson
Date: 04:12:58 06/26/99
Go up one level in this thread
On June 26, 1999 at 02:10:29, Dave Dory wrote: >It's VERY ENLIGHTENING to watch a Master or GM who is familiar with playing >against computers, just take out the hinge pins from those tiny little silicon >gates! > >As Dr. Hyatt amongst others have noted, most computer chess players really ARE >not up to the standards of the human GM's. Even Deep Blue, who had ZERO >published games for G.K. to study over, was really quite fortunate to win their >abbreviated "match". Props to D.B., and it is (was?) a truly extraordinary piece >of hardware and software, but I think everyone also knows it was a lucky little >chap, also. > >The reason you don't see positional sacs in computer vs. computer games is >simply because programmers who have tried to introduce this knowledge into their >program have found, rather dreadfully, that the program will now sacrifice it's >material quite readily, until it has no hope of winning!! > >The horizon effect has been pushed back, by both hardware and software advances, >but it has not been pushed beyond the horizon of a really good chess player - >one who's familiar with the program, and has done his/her homework. > >Efforts to SOMEHOW increase the knowledge of the program, apart from the info in >the game tree or end game data bases are (IMO) very challenging. We shouldn't be >surprised since we still have to TELL the program how to win many end games, >move by move! > >[ My P.O.V ] > >What I would love to see is an emphasis on programs that use NO opening book, >and NO end game data. NOW we can play the program, and see what the program >really knows, NOT HOW SMART THE OPENING AND/OR END GAME DATA BASES ARE!! > >If the program can't play a good opening, on it's own, it should be made to do >so. If the program can't play an ending correctly, it should NOT BE ABLE TO JUST >LOOK UP PROPER PLAY MOVES IN A MAMMOTH DATA FILE. It should be programmed to >solve endings, on it's own, over the board. > >** NOT JUST LOOK IT UP!! ** > >I know from experience, that most computer chess enthusiasts don't agree, but it >strikes me as CLEARLY unsportsmanlike to allow the human no books or printed >data (let alone a computer) of any kind, while at the same time, allowing the >computer to possibly dig out a great move from some internet resource halfway >around the world, or look through millions of move sequences laid out on it's >own hard disk. > >[ Just my P.O.V ] >( program reports soap box safely stored away for later use! ;-) ) > > >Dave I think you've articulated the problems with computer chess programmes very well Dave. allan
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.