Author: Stephen A. Boak
Date: 13:44:18 12/29/05
Go up one level in this thread
On December 29, 2005 at 07:57:44, Uri Blass wrote: [..] Hi Uri, I've numbered your statements for easy reference. 1. >>I understand the reason that you give them but people cannot get conclusions about the difference in program strength in endgame based on the results. Conclusionary statement (negative), devoid of facts and reasoning. 2. >To be more correct I mean that the conclusion about the rybka-fruit difference based on the result is simply wrong. Conclusionary statement (negative), devoid of facts and reasoning. 3. >You cannot say based on result like 21-19 that rybka is almost the same strength as fruit in endgames. Conclusionary statement (negative), devoid of facts and reasoning. 4. >The practical result if we consider only matches of 2 games is 7-5 for fruit and 8 draws that suggest smaller difference than 21-19 Arguing convincingly against yourself (against your Points 1-3, above). Now you conclude that the results *do* suggest a small (perhaps tiny) difference in endgame strength of RYBKA & FRUIT. A. CANNOT DRAW A CONCLUSION In your Point 1, above, you indicate people cannot get endgame strength conclusions based on the results. You are a person. In Point 4, you draw a conclusion based on the results. Therefore you have completely contradicted yourself--without outside intervention. I don't know what to make of that. You are a mathematician. I believe that mathematics requires some internal logic. As does a philosophical discussion or debate. Here I don't see the logic in your reasoning. B. CONCLUSION IS WRONG Your Point 2, above, is made more strongly than your Point 1. You go beyond saying a conclusion cannot be drawn. You say the conclusion drawn by another is in fact wrong. In Point 3, you indicate what the conclusion is (by another) that you believe is wrong. In Point 4, however, you DRAW THE SAME CONCLUSION, i.e. RYBKA - FRUIT ENDGAME DIFFERENCE IS VERY LITTLE (contradicting your Points 2 & 3). Thus you agree with the very conclusion you have indicated is wrong!! Where is the logic in that? What did I miss? 5. >and I do not think that >fruit's strong point is simple endgames(it has a problem in fine70). You build and attack a self-constructed strawman. You tilt at windmills (a la Don Quixote). Why debate or add this misplaced point to bolster your paricular argument with Sune Larsson. Nobody said that Fruit's strong point is simple endgames. Sune Larsson did not. The Sune Larsson test did not use the Fine70 position. The Sune Larsson test was merely a relative test of FRUIT & RYBKA, in a particular set of selected endgame positions chosen for some logical reasons. Nothing more. The results are what they are. The essence of scientific testing is to attempt to gather evidence ... to support conclusions about the world (and strengths & weaknesses of chess programs!). Sune did just that. His conclusions are reasonable (have reason and logic behind them, are not too strongly made): 1) RYBKA & FRUIT are of similar endgame strength. 2) The results of this test suggest that FRUIT is a little stronger in endgames than RYBKA. If you do not like the particular test (Sune's), or how it was conducted, then conduct it yourself and gather your own data (results). Else devise your own improved (perhaps larger, with more positions) test and run it. Include Fine70 if you wish, but be sure not to bias the test against any paricular engine (FRUIT!). Perhaps you should include *all* the FINE positions. Better yet, all the known endgame positions! Even better, all the possible endgame positions!! Scientific testing in such fashion may broaden the types of results that are obtainable, and strengthen the conclusions that may be drawn. To set up any other strawman conclusions (not drawn by Sune) and to then knock them down (presuming you can do so!), does not reflect in the slightest on Sune's properly drawn conclusions. To use such devices shows weakness in your logical discourse and suggests the weakness of other logical arguments you are able to muster. Why add a weak (or invalid) point, to help support your own [supposed] strong point of belief? Simply show your strongest argument and support it as best you can. The validity of any scientific conclusion is certainly debatable--through the use of facts and logical reasoning. Validity often rests on the strength of the conclusion that is being drawn. Sune has drawn no unreasonable conclusions, based on his specific test. The strength of his conclusions are precisely suited to the particular test and the specific results thereof. On the other hand, your conclusions are made too strongly, and therefore must be rejected, as presented. 6. >I will not be surprised if fruit lose in this test against shredder or fritz >inspite of the fact that I believe that fruit is better endgame player than >them. > >Uri Please clarify, are you referring to the Fine70 single position or Sune Larsson multi-position test? Nevertheless, every program (no exception) has bugs in its software, mistakes in its search, inaccuracies in its eval, some particular positions which it does not handle correctly (or best, if there be such). One instance (Fine70) does not make a general case. It does not support a too-strong or too broad conclusion. ADDITIONAL OBSERVATION & CONCLUSION--It is okay to change topics in an email ... from discussing, for example, Sune Larsson's RYBKA-FRUIT engame test, results and/or conclusions--to introducing a new topic of interest, for example, how strong is FRUIT vs SHREDDER or FRITZ in a particular endgame test. But please make it clear when you are no longer critizing the test/results/conclusions of your original topic ... so you do not muddy the waters by your discourse. Best regards, Steve
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.