Author: Thorsten Czub
Date: 11:09:11 05/28/98
Go up one level in this thread
On May 28, 1998 at 11:26:08, Tony Hedlund wrote: >>Thats exactly what ChessBase needs. >I don't now what Chessbase needs. Yes . It looks like you don't know. And you canot imagine. I wonder why. It is obvious. >Is it not what it's all about. Had Fritz5 got 100 points less you would >have ignored it. You know what I would have done ?! Again - i would not have ignored stupid behaviour. I have never overseen this. Not in the years before. And not yet. And I will not oversee your strange methods in future. >>I am discussing the question why you test a program against opponents , >>nobody is interested in. > >Who decide what's interesting, you? Obviously it is not up to me to decide which programs you test. You don't see this too ? Strange. You really seem to have limited imaginations. But coments of the ssdf-board said in the past, that WEAK programs get WEAK platforms. So you KNOW in forward which programs are strong, before you test it. I always thought the sense of the matches is to find out about this matter. Your own board told me: it is different. >>Why don't you test the new Comet-Version instead ? >>Why don't you test Virtual2 instead ? >>Why don't you test new Diogenes instead ? >>Why don't you test crafty instead ? >We haven't got them. If you know the programmers, please ask them if >they can send some copies to us. Maybe you could try what we all do, buying chess programs. This is a form of SUPPORT to computerchess. Downloading comet from gambit site should be possible for you, unless you have no modem. Virtual2 e.g. is not expensive. Diogenes is also free. Crafty too. Your "POWER" to optimize the list seems very limited. >>There are many programs you could test that do not appear on the list. >>Instead you do "senseful" testing against Nimzo3.5. >Yes I agree. I also think it make sense to test against Nimzo 3.5. I have a version of Mchess 1.32 somewhere. Shall I send it to you, that you can test it ?! It's for free. And I would send it to you :-))) >>Depending on what you want to show. If you want to show that your ELO >>ranking is ok, than you can continue like this. > >I can give you a recent exemple. I played the match Fritz5 P200MMX 44MB >- >Comet32 P90 17.5-3.5. Aha, you say, there Fritz5 got some extra points. >But instead it lost a few. Beacuse according to Arpad Elo's system the >match should have ended 18-2. I guess most of the CCC people understand >this point of the >Elo system. I belivie you are not among them. The question is not if you do statistical senseful things. You can add beans as much as you want. And your formulas may be accurate. But when you test with wrong programs, and unimportant predecessor programs, you measure something different than elo-strength. I know you will not understand this point. It is very difficult to understand that doing mechanistical RIGHT things causes WRONG results. Statistics do never show the reality. They are a function of the methods. We know which methods you have. >>You work like a self-fullfilling prejudice. You know in forward which >>programs are not strong enough >I had the impression that it was you who had that gift. Study the official comments and the output of your board concerning testing hiarcs, and you will remember that THEY said that they only test strong programs on fast hardware. After hiarcs won the championship they did not counted it as strong. So - who has the right gift. >>that they need state-of the art harware, >>and when you oput them on nostalgic platforms and let them play against >>the state of the art hardware and they fail to win, than you say: oh - >>we SAID that this would happen. >> >>Why do you test anyway, if you know in forward that some programs don't >>need a fast hardware >We don't now that. Then write better comments. >>meanwhile other programs (Fritz) get it by guaranty/agreement. >In general all new programs get the fastest hardware. But it is all >about rescorses. Our's limited. Ah yes. resources. You always say this. You said this 1993 and before. Resources. Limited. It seems in sweden many things are limited. Not only "software" but hardware. It is possible to test genius on fast hardware. But not to test hiarcs on the same hardware. And you call this limited resources. I call this limited will. >Let me ask you a question. If you dislike us so much, why bother? You >spend hours and hours to bickering us. I f you don't like us, just >ignore us. >Take it or leave it. I do not dislike you. And I will not ignore you. You prejudice again. >Tony
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.