Author: Joshua Lee
Date: 13:48:16 08/14/01
Go up one level in this thread
On August 14, 2001 at 15:01:17, Dann Corbit wrote: >On August 14, 2001 at 14:19:56, Marc van Hal wrote: >[snip] >>Computers are in some cases far from GM strenght >>How can you otherwise explain the usage of epd files. >> >>I am telling this because I am curently overlooking the positions from the book >>offers (sacrefices) from Vucovic >>It takes quite some while before the programs find the compensation and then it >>is to late. >>At the moment computer programs do find these sacrefices they realy are from GM >>strenght >>Most of the time it is so that these sacrefices do increase the development and >>most of all which is important for programs the valeu of the advantage. >>Which over comes many draws. >> >>Then again when this is acomplished programs realy become unbeatable for humans >>Aslong as you overcome the problem of pre knowledge. > >I think that brilliant sacrifices and brilliant positional moves by GMs are >(possibly) nearly equalled by brilliant deep tactics found by computers. > >The method of play is a bit different. But the proof of the pudding is in the >eating. > >Give LCT II to your favorite computer on tournament time controls, and you will >see the weaknesses clearly. On the other hand, GMs also miss brilliant >strategic moves and sacrifices. > >I think that probably "advanced chess" is the strongest form of chess, because >the computers will spot the deep tactics and the humans will see the positional >moves and sacrificial moves. > >At any rate, I think computers play strong and interesting chess -- similar to >GMs in strength although with a contrasting style. > >Eventually, computers will see deep sacrifices and positional moves. Probably >sooner than anyone thinks (except for those that think they can already see >them). >;-) The Computers will find the solution if given over 15minutes Out of 6000+ positions i've found 300 and climbing that aren't solved by fritz in 15minutes(there will be 1,000 maybe when i'm done. I still have to run COVAX, MES, ECE , 82 positions from LK and 1,252 not solved in 3 minutes So i am willing to bet that computers in reality are not GM strength unless they are on more than 5Ghz , People can say all they want about Norms and the like but if a IM or GM made this move and it is the "Correct" move but the computer doesn't play it well that should tell you something . Despite many problems having the incorrect move given (MES and ECE especially have bad analysis. I narrowed from 17 or 1800 positions each to 1164 for mes and 1365 for ECE which the Tablebases Take care of. I would like to run the epd Z_Learn but the move to find is after the one given so basically the 2,105 are from 1 move before and i don't exactly want to go through on each one fast foward to the next move and sit there untill the move is found. If i get around to it i might go through and replace the positions like this and then run it but it is extreemly time consuming. Lastly the NY1924 EPD is huge and i think it is the bigest "Important" testset out there but not all positions are that great as it is supposed to be every position from the tournament including lost ones. However Sometimes computers are better than TAL.
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.