Author: Uri Blass
Date: 01:47:17 02/21/06
Go up one level in this thread
On February 21, 2006 at 01:23:08, Sandro Necchi wrote: >On February 20, 2006 at 18:17:28, Uri Blass wrote: > >>On February 20, 2006 at 15:45:26, Sandro Necchi wrote: >> >>>On February 20, 2006 at 02:28:36, Uri Blass wrote: >>> >>>>On February 20, 2006 at 01:21:44, Sandro Necchi wrote: >>>> >>>>>On February 19, 2006 at 18:43:40, George Tsavdaris wrote: >>>>> >>>>> >>>>>>>If you believe so than try to see how long it takes to a program to believe this >>>>>>>move is interesting to be played... >>>>>>> >>>>>> >>>>>> Once again you use inductive logic..... The fact that all programs can't find >>>>>>THIS specific position, can't contradict none of my above statements! You should >>>>>>find other arguments for that....:-) >>>>> >>>>>I do not want to contradict your statements, but only show you one example which >>>>>show how little is the depth reached by computers to find good novelties. I did >>>>>not chose a special one, but the first one I saw on the new Chess Informant. >>>>> >>>>>After nearly 30 years of testing on computer programs and having checked with >>>>>them probably some several thousands of games (200.000 - 300.000; I did not >>>>>count them) I am 80% sceptical. >>>>> >>>>>I mean that they can find some novelties in some specific positions, but most of >>>>>the time they are not good for the reasons I told you. >>>>> >>>>>If you don't want to believe me it is up to you. >>>>> >>>>>I am not interested to teach things to people, but possibly only tell them my >>>>>findings. >>>>> >>>>>Ciao >>>>>Sandro >>>> >>>>I think that part of the novelties are not correct and only has practical value >>>>because the opponent is not ready against them. >>> >>>Of course I am referring to strong novelties...those that can work at GM >>>level...for low rating players even childs can propose some... >>>> >>>>Top GM's can choose a move that leads to objectively slightly inferior position >>>>against correct defence when they are almost sure the opponent is not going to >>>>find the correct defence(and even if he find it they have good chances to draw >>>>the game because slightly inferior position does not mean losing the game). >>> >>>If a strong GM cannot find the correct moves in the game, than the noveltie is >>>good. >>>If you look the theory development you can find a lot of variations which were >>>good, but not anymore. >>>Some of them were proposed by the strongest players of the time... >>> >>>We are not able to find perfection in chess yet. >>> >>>Sandro >> >>The problem is that GM's are weaker than computers so it is possible that the >>novelty that work in OTB games against a top GM is not going to work against the >>best programs. > >I meant strong GMs...over 2700 and not GMs in general. > >They are strategically much stronger than current engines. I do not know if this is the case. I did not see adams superiority in strategy against hydra. Adams did not get superior positions only to lose for tactical errors. Hydra did not have a big book so if it is correct that adams is stronger in strategy then he could show it in his games when hydra was out of book. > >> >>It is clear that with the fast improvement in the last years in chess programs >>we get closer to perfection and I expect computer programs to find many good >>novelties in the opening if you only give them enough time. > >WE are far away from perfection...to be close to that computer should search at >least 64/126 in the early stage of the game in most openings...we are far from >that. My opinion is that good evaluation is enough for it and evaluation of chess programs become better. > >> >>For novelties of top GM's that they do not find my guess is that most of them >>are not the best move and they can work only if the opponent is not ready and is >>not a top program. > >Sometimes they do so, but they are also moving the theory and not the >computers...computers rely on the book makers which refine and test variations. I believe that it is the opposite. I believe that today most of the good novelties of top GM's are based on analysis with computers even if computers did not suggest them. I believe that humans rely on computers to find novelties and they decide to reject many interesting ideas that they think about because an analysis with computer finds that they are wrong. Uri
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.