Author: Robert Ericsson
Date: 08:33:08 04/23/99
Go up one level in this thread
On April 23, 1999 at 10:15:07, Harald Faber wrote: >On April 23, 1999 at 09:51:17, Robert Ericsson wrote: > >>I've just read an article on Rebel's home page about SSDF, comp-comp and >>comp-human results. >> >>There seem to be two statements made out of this article: >>1) comp-comp tournaments (for instance SSDF games) can be tweaked with >> different/special opening books. At least that's what the comp-human >> people argues. >> SSDF's list is not 'showing the truth' >> >>2) comp-human tournaments are better as they will give a true FIDE-ELO >> where the programs have performed (good or bad) against humans in OTB >> (over the board) tournaments. >> More 'accurate' ELO list >> >>But hey, wait a minute!? Why is statement 2 so much better then 1? > >Easy: Because comp-comp in most cases is playing without a (reasonable) plan. Yes, but performance rating of the comp. is very much biased by the fact that OTB players blunder or overlook tactical sequences a lot (or even loose on time). > >>In comp-human tournaments the computer will always benefit from the fact >>that humans always make minor tactical misjudgements/blunders during a >>game and the computer will not. > >That is correct, on the other hand humans do have a better positional >understanding, especially when there is no tactics around. Certainly true but tactics are always lurking around ;-) >Programs have become strong opponents especially OTB. But my experience is that >for analysis you can't leave them analyzing on their own. It still needs some >brain for the right ideas. And that's certainly true! That's why I use 'watch' command in Fritz and not cc analysis feature when looking through my own played games :-D > >>Why not a computer vs cc player (correspondence chess player) tournament? >> >>The humans can check for blunders with a computer themselves, leaving >>us with a game where the computer's skill of planning will be tested >>against the cc player's skill of planning. > >Allow the cc player to check his ideas with a program and he will crush the >opponent program. I'm not so sure. Maybe if the cc player have an ICCF-ELO >2500 but not otherwise. Hmm... clarification: if comp uses 40 moves in 2 hours and cc player can think much longer it's true. With equal time I don't think so. > >>Maybe 'advanced chess' tournaments with time control more then 40 moves in >>2 hours can be worth considering? Maybe 40 moves in 4 hours? > >This is not enough time for cc player to look up all his material... This was more a suggestion to an improved OTB Tournament. Seems that I ended up talking about two possibilities :-)
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.