Author: Mark Young
Date: 19:18:11 08/02/99
Go up one level in this thread
On August 02, 1999 at 21:32:56, Victor Valenzia wrote: >On August 02, 1999 at 18:53:54, Mark Young wrote: > >>On August 02, 1999 at 18:36:07, Chris Taylor wrote: >> >>>I have started a match between Hiarcs 7.32 and Doctor? 3.0 >>>Doc is running on an amd 400 with 32 meg ram, and using General.ctg >>>Hiarcs is on an amd 350 with 64 meg of ram. Using its own book! >>>The time is 40 mins each for the whole game, resign set to early >>> >>>Just like the result Tania got Hiarcs is 5 up, and the Match, is still on? >>>If anyone has the slightest interest, tommorow, i will stop the match, and post >>>the results? >>> >>>Chris Taylor >>>England >> >>With all due respect to Bob Hyatt, having an opinion on something is one thing, >>and having data on something is another. It is Bob opinion that under the >>Chessbase interface engine vs engine testing with one computer will result in >>flawed data. Having done both and looking at the data generated by both, I have >>found both methods when using the chessbase engines give pretty much the same >>results with a acceptable +/- for the amount of games played. I can find nothing >>wrong in engine vs engine testing from the data I have generated. I think when >>your match is done, it will pretty much agree with Tania's one computer engine >>vs engine results. > >There is one huge difference in playing on two computers, that being the ponder >mode, or thinking on the opponent's time. This can be especially significant in >matches with long time controls. For instance, in my Hiarcs vs. Fritz matches >on two machines, the programs will often respond instantly (0 seconds) if they >have predicted the opponent's move. In engine vs. engine matches, they start >from scratch on every move. Correctly predicting the opponent's moves can save >huge amounts of time that can be allocated to more difficult positions. In my >opinion, this is more accurate. > >Imagine if you will, two human chess players with alzheimer's disease. After >one of them makes a move, he is forced to leave the board, then the other human >with alzheimer's comes to the board to make his move. This is repeated until >the conclusion of the game. In essence, each player is making a cold analysis >of a position each time he steps up to the board. Compare this with two >mentally healthy chess players sitting at the board, each one calculating >counter-attacks during his opponent's move. I think that this is a fair analogy >of engine vs. engine matches and matches played on two machines. The question is not, which one is more accurate; I would always take two computer results Vs 1 on principal alone. The question is can you get accurate data playing engine Vs engine on one computer. I have found the answer to be yes. All the talk about pondering, incorrect time usage etc. does not reflect hugely in the ratings between the two methods.
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.