Author: Roger Brown
Date: 15:00:21 01/29/04
Go up one level in this thread
On January 29, 2004 at 16:21:37, Sirotovich wrote: >My question is for all computer programmers, who read this news. > >Dear Chess Friends ! Hello > >Please open our eyes ! Does it make ANY sense to run engine vs. engine matches >or round tournaments on one machine ? From my point of view, it is a waste of >time and the results wouldn’t reflect the real strength of weakness of each >particular engine. Very well, this is an excellent reason for you not to engage in such an activity. > >I have bought my first Fritz 5 when it was just release many years ago. Since >that time I run many eng. vs. eng. matches and round tournaments with different >time control. Very often the results were too good to be trusted. Many days >and nights my PC worked non-stop and I just wanted to find which engine was the >best !? Ah, I do not understand you. You do know that there is a need to play a statistically significant number of games, don't you? Failing that, all that can be said is that Engine X did better in my basement competition... There is a pleasure in that as well I suppose. Certainly I find myself doing it on the odd occasion when I am not actually beta testing an engine... That having been said, your experiments can constitute a scientific process if they are reporducible. I may not want to play 100 one hour games on my machine but there are several persons who do that here. The one I bought just now for $100 or another 2-year old one, which >probably didn’t cost anything at that time. For example, the old man Hiarcs 6 >could win 4 round tournament ahead of its latest version to say nothing about >the other young and expencive rivals from Fritz, Shredder of Junior group !? I >foresee your questions about my PC. It is a good one with a lot of memory and >processor power. > >I read many results here when Fritz beat Shredder, Crafty beat Junior, Tiger >beat Hiarcs, or absolutely the other way around !? Nearly everyday someone >provides new results, which conflict with results sent by another peson. Can >we trust all these results, if they ware achieved on one machine ? When 2 >engines played ONLY 100 games, someone definitely wouldn’t be happy with this >result and he would propose to play 1000 games !? If another person managed to >organize 1000-game match, they would definitely propose him to play 10,000 >games. >I understand, it is a good fun to watch when 2 Grandmaster strength engines >perform a real spectacle for us, but do we need to take these results seriously,>if we don’t have another computer to connect them together ? Sad to say, even with connected PC's the challenge is how much is too little and how many is enough. I do not think that it is reasonable to ever suppose that one match of even 100 games could "prove" anything. The best that can be hoped for is that the process confers some pleasure. That in itself is sufficient reward I think, whatever the correctness of the results of such matches. Perhaps we should be honest that the matches give us pleasure and stop trying to find some TRUTH because even the SSDF list does not enable us to say which program is best beyond argument. All it says is given the margin of error and timecontrol, this is the expected rating. let's just lighten up and enjoy life. It is a wee bit too short to argue about stuff like this. Especially since I am about to publish the result of a two game match where I whipped Engine XZ! :-) Bear in mind that some of us can only afford the one machine. Do not tell me that there is to be a law about the single machine folk? Oh, the humanity! Surely we do not need one more thing dividing us instead of uniting us? Sorry, I am crazy. > >Please help me or other friends to safe our time and energy or a battery life !? It is my computer and my battery life and I am going to be as stupid with both as I want to be! :-) Later. > >Best wishes > >Anatoli Sirota
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.