Author: Dirk Frickenschmidt
Date: 17:42:06 09/18/98
Go up one level in this thread
On September 18, 1998 at 05:13:58, Peter Herttrich wrote: Hi Peter, >On September 17, 1998 at 17:42:58, Dirk Frickenschmidt wrote: > >>On September 17, 1998 at 16:51:18, Thorsten Czub wrote: >> >>>"interesting" is IMO that you feel the need to quote this here, even to >>>translate this rubbish interview stuff here. >>> >>>there is NO excuse for the nunn-test. >> >>First funny remark. >>For what in heaven should the Nunn-test be excused at all??? >> >>Could you please hint to any even remotely debatable *content* for claiming >>this? >> >>Any chess judgement on the Nunn positions? >> >>Any anything? >> >>Or do you just think the fact that Frederic Friedel knows John Nunn since many >>years is any kind of argument by itself??? >>At least in conncection with your personal suspicion??? >> >>>the nunn test was, so says css itself, done by nunn for css. >>>he has done it specially for his friend frederic. >> >>If I understood it roight Frederic Friedel then wanted a test with a >>representative set of opening positions for testgames, without book and cooking >>effects. >> >>I want that too. >>The Nunn test was John Nunn's proposal for doing this. >>There may be other proposals. >>The Nunn test may be imbalanced. If anyone proves it to be so. >> >>So what??? >>Does your suspicion imply *anything* substancial??? >> >>>frederic is owner of chessBase. >> >>Since when this????? >> >>Did I miss him buying it? When did he do so? >> >>You probably can explain this claim to us? >> >>Or should you perhaps check your sources before quickly selling your darkest >>nightmares as reality? ;-) >> >>>it is the same when the owner of malboro asks a special cancer-doctor to write >>>an article about smoking not beeing dangerous. >> >>Frederic Friedel did not ask John Nunn to buy Chessbase for him, >>and then write that he (John Nunn) doesn't think it is dangerous to use it :-) >> >>Though some Mr. Czub may still think it is extremely dangerous... :-)) >> >>>you now quote anywhere that smoking is not dangerous because john nunn has said >>>this. >> >>Has he? I didn't even know he was smoking... ;-) >> >>Anything in one cooking pot? >>Reality, questionable comparisons and funny metaphors, >>all suddenly turning back into some new kind of expected reality? >> >>No matter if it fits, as long as you can cook it in one pot? >> >>>and frederic paid for this... brilliant. >> >>Did he? >>You probably have a copy of the bill? >> >>Or are you just misjudging reality from allegation phantasies again? >> >>I'm thankful for any kind of clarification of all the more than questionable >>points in one short post. >> >> >>Regards >>from Dirk You wrote: >Hello you fighters :-) > >I would like to put some oil onto the waves. > >I also followed the articles in css and came to the >conclusion, that the dependence Nunn/Fritz/cheating/Friedl/aso >is pure speculation. > >G. Mueller said a right thing: The Test only tests himself. > >I also invested some time with this test and build with my >rock-base a very huge book for crafty, which is normally >not for playing. >Then I followed the first three tests move for move with >this book. I had to notice with growing movenumbers a >very rare statistic for the moves. So I saw, that the >possibility, to reach this Nunn-position is statistically >nearly zero! No Chessprogramm with a good book would reach >such a position and had a chance to show it's ability, to >handle it. The Nunn-test has nothing to do with practically >real played chess. So I call it worthless. I don't think the value of such a test depends on how common the last moves leading to each position are. I always considered as decisive, how well a certain position will be able to mirror the playing skill of programs in a certain way. I was and am somewhat sceptical like you about the Nunn test being represantative for widespread opening theory, adding that most programs will not leave their books that early in most of their games. So it is not representative in a double way. But after having a closer look and seeing some testgames played with it I found out that it nicely checks important areas of middlegame playing skill with various pawn structures, requiring quite a variety of strategic concepts for doing well from the programs. So it will simply remain the best test of it's kind until someone piblishes a better one based on similar principles. >I make a difference between theoretical chess (openings and >positions) and practical chess (openings and positions). >Nothing against experiments ore new ideas in the middlegame. > >I hope, my point comes over. My english int't very good >to explain this. I hope I understood and answered what you were hinting at. >I have another idea, to check, how chess-programs play without >influence of openings. >How abt: Let the programs play the first 8 moves from the >book and then switch off the books. Perhaps we can say 10 >moves. This is a thing, what is to try. Yes, it would be an idea to cut off too long theory lines leading to a poor middlegame or even endgame. But it is difficult to do with a fixed length. Some variations lead faster to recognizable middlegame structures than others. It's something else to have 8 to ten moves in the Tromp, or in the Najdorf Sicilian or King's Indian, just to name some samples. >Another possibility is, to set up known themes (french, sizilian, >or wahtever) till move 8 ore 10 as a position and let play >the silicons without book. Ed once proposed Fisher-random, if I remember right. It's all debatable. But I would still prefer 20 (instead of 10) well chosen openings to around move 10-15 each to reach a developed middlegame positions with some more variety of problems and structures than in the Nunn test. And even the Nunn test has some advances, first of all: it is there, it's a sensible set of choices from a GM, and it does contain only 10 positions, which gives you a little chance to find enough time to even use it :-) >Only my two cents. > >Peter Same with me, Kind regards from Dirk
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.