Author: Rolf Tueschen
Date: 03:43:29 11/01/03
Go up one level in this thread
On November 01, 2003 at 04:13:54, Mike S. wrote: > >Btw. on this occasion, do you have a comment for these results provided by a >german computerchess fan: > >http://www.miko42.de/DasDuell/duellindex.html > >excerpt: > >Index Fritz 8.0.23 - Shredder 7.04 F8 - S7.04 % Fritz % Sh. games >----------------------------------------------------------------------- >01 eigenes Buch - eigenes Buch 131,0 - 145,0 47,5 52,5 276 >12 eigenes Buch - ohne Buch 143,5 - 151,5 48,6 51,4 295 > ====== >(engl. own book vs. own book, or own book vs. NO book) > >Without book only 1,1% less! Certainly somewhat surprising. :-) > >So, what can the difference be between King with CM9000.OBK (which the opponents >most probably couldn't/didn't prepare for!), or with that old general.ctg? More >than in the comparsion Shredder without (!!) book to Shredder with your big new >S7 professional book (+1,1% only, against a similar strong opponent)? I don't >think so. A book compared to another book (both reasonable quality) can't make a >bigger difference than NO book compared to a Necchi book :-) And if you claim it >may make a bigger difference, than you'd have to explain why your book added >only 1,1% in the above match comparison. > >I'd say forget it! The difference, if any, will be statistically insignificant >with the high number of games SSDF plays. This is wrong - for several reasons. 1) What does it mean if you claim a "high number of games" in SSDF? Don't you know that the games of a single match are very small numbered? 2) Then you present results of one Michael42 from Germany who has [allegedly] proven that the importance of a book is insignificant. That is false. Here are the aspects why the test results of Michael42 are uninteresting at least. 2a) I quote "und es wird sicher welche geben die aus diesem Grund meine Rangfolge ablehnen oder als weniger "gültig" betrachten. Das kann jeder halten wie er möchte, ich werfe einfach meine Ergebnisse zusammen, rühre einmal rum und lass EloStat die Resultate ausrechnen. Ich erhalte dann eine Liste wie sie im "menschlichen" Schach Gang und Gäbe ist wo auch nicht jeder Spieler die gleiche Anzahl von Partien gegen jeden Konkurrenten gespielt hat. Man verzeihe mir meine unwissenschaftliches Herangehen an dieses heilige Problem ;-)" In English a rough translation: some people will object against my ranking, that doesn't bother me, I put all my results together and let run EloStat. That is how it's done in human chess. Sorry for my unscientifical proceedings in facing the holy problem ;-). 2b) I quote by heart: I de-activate the learn-files and make them write-protected. Already played games with the same scores are deleted. Now you ask here how people take the astonishing results of Michael42. Answer in short: - Michael42 himself says that science doesn't interest him - Michael42 plays without activated learn-files - Michael42 deletes certain repeated (formerly already played) games with the following result: two top engines like SHREDDER and FRITZ are almost equal, SHREDDER is even insignificantly better; the existence of a book is insignificant! In other words: If you do NOT control your experimental setting, if you dont make exact documentation of your experimental design (your decisions for instance), if you then disable learn-file techniques, a major factor of the strength of the actual top programs for tournament play and matches, and if you then also make the statistically incredibly stupid decision to delete a whole lot of won or lost games just because a machine already had followed that same game score, THEN you can prove that the existence of a professional book makes no statistically significant meaning. Goodness gracious me! I put it in even shorter words: If you forget about all science, de-activate the most important feature of modern chess machines, and if you violate the basics of statistics, THEN you get a statistically insignificant result for book existence. What a big surprise!! Really? Of course NOT! Since the book was designed for the complete package and NOT such a nonsense crippled entity. The book is created in close relation to the specifications of the engine and the learning file technology. And by nature it is statistically proven as making sense - for that specific engine. Now perhaps the whole spooky nonsense becomes clear. If you cripple the whole setting then suddenly you find that the book makes no sense at all. Well done. I quote Michael42: If you are beyond 40, you should stop such a nonsense activities and should only concentrate on a smooth way of living. Some such could also be a good advice for your questions, Mike! I think you can't trouble Sandro with such naive questions. The presentation of all these quoted numbers of results is a forbidden action, because it implies that the numbers had been achieved in a controlled experiment. But exactly this has been excluded as the author himself has stated. Nothing wrong with the private hobby of Michael42, but it's of no use for a serious debate about opening books. You should have known that as the CSS-forum spin doctor. ;) Hope this helps. Rolf (spin doctors "doctor") > >Regards, >Michael Scheidl
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.