Author: Sandro Necchi
Date: 03:43:03 02/19/06
Go up one level in this thread
On February 19, 2006 at 06:17:16, Uri Blass wrote: >On February 19, 2006 at 05:45:47, Joseph Ciarrochi wrote: > >> >>> >>That's cool (using a special book), unless you are interested in the engine's >>ability to analyze openings and find novelties...... > >I agree > >I do not know about other users but for me results of tournament like CEGT or >CCRL are more important than result of ssdf. > >I use engines mainly for analysis and I do not care much about factors like >book. > >If the target is to win the ssdf then in theory it is possible to design some >killer book when you consider the fact that some ssdf opponents are static >targets and you can simply get wins directly out of book. Well, no one is doing this as this would not pay...you can do it against older programs but you do not know how it would score against other programs. I mean changing an opening line to score well against a specific opponent quite often would give a bad score against most of the other competitors. What the opening book makers do is to refine the theory as there are several mistakes still and also to select opening that the engine can understand and play strongly enough... There is a lot of difference! Also this is interesting as they would offer an "improved theory"...of course for the time being as this is developping all the time... > >As far as I know it is not done but my opinion is that programs can easily get >more than 100 elo improvement by a better book and you only need enough computer >time to test. No, this type of improvement is not possible if not for a very selected opening repertoir which cannot be offered for commercial purposes as too restricted... Do not forget that to find novelties that work against all programs is not easy and needs a huge amount of work... > >killer book may even help to give results of 100% against some opponents. Only if specifically devoted against a weaker older program...the new version may will play different and one may loose the game...if you do not believe this check the famous play-off between Genius and MChessPro 5.0 at the 1995 WCCC... > >If you want to get 100% against weaker opponents then you can do the following. > >You can simply play a match Shredder9(1200) against Fritz5.32(P200) in ssdf >conditions. >If you see that shredder9 lose or draw you fix the line that it lost or drew by >adding moves to the book in order to have a win against Fritz5.32(P200) and put >it in book and start a new match. > >I believe that after enough tries you will probably get 100% result maybe possible but a useless work... > >In theory I believe that this strategy can help to get 100% against every >opponent but against strong opponents you may not have enough time to try enough >tries. it depends how stronger is the opponent... if the opponent is 100 or more points stronger, a better or optimized book will not be able to do anything unless you know exactly what that program will play considering the book and the hardware used as well... > >Uri Sandro
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.