Author: Miguel A. Ballicora
Date: 10:19:53 05/07/02
Go up one level in this thread
On May 07, 2002 at 12:34:57, Uri Blass wrote: >On May 07, 2002 at 11:29:40, Miguel A. Ballicora wrote: > >>On May 07, 2002 at 02:29:26, Uri Blass wrote: >> >>>On May 06, 2002 at 22:31:28, Christophe Theron wrote: >>> >>>>On May 06, 2002 at 19:45:22, Uri Blass wrote: >>>> >>>>>On May 06, 2002 at 18:06:47, Christophe Theron wrote: >>>>> >>>>>>On May 06, 2002 at 15:34:01, Amir Ban wrote: >>>>>> >>>>>>>On May 05, 2002 at 19:58:09, Christophe Theron wrote: >>>>>>> >>>>>>>> >>>>>>>>"Knowledge" in the sense of positional evaluation (that's what most people think >>>>>>>>about when they talk about knowledge) makes for 10% of the strength of a chess >>>>>>>>program. >>>>>>>> >>>>>>>>Chess is 90% about tactics (which is a concept close to "search"). >>>>>>>> >>>>>>> >>>>>>>Before strongly disagreeing (as I guess I will), what does this mean ? >>>>>>> >>>>>>>If I freeze my search engine and work only to improve the evaluation, how much >>>>>>>do you expect the total strength to improve ? Is it limited ? >>>>>> >>>>>> >>>>>>I expect the strength of your engine to improve, but not much in regard to the >>>>>>energy invested. Because you are going to focus your efforts on an area that >>>>>>does not have the biggest potential in strength. >>>>>> >>>>>>On the other hand people will love it more and more because it will have a much >>>>>>better playing style. >>>>>> >>>>>>People can forgive gross tactical blunders, but not slight positional mistakes. >>>>>>Go figure... >>>>> >>>>>I think that people are different. >>>>> >>>>>There are people who will prefer the engine that is better in tactics and there >>>>>are a lot of people who are going to prefer the engine that wins without caring >>>>>for the reasons. >>>> >>>> >>>> >>>>That's different when your program plays against a gransmaster in a public place >>>>(or on the Internet). >>> >>> >>>comp-comp games are also interesting for many people. >>> >>>I think that programmers usually care more about comp-comp games and not about >>>comp-human games. >>> >>>I know that some simple ideas about time management that can be productive >>>against humans are not used by most of the programs. >>> >>>One of them is simply to play faster when the opponent is in time trouble. >>> >>> >>>> >>>> >>>> >>>> >>>> >>>> >>>>>>Here I'm talking about current top engines of today, naturally. >>>>>> >>>>>>Building a chess engine with a broken evaluation to demonstrate that a better >>>>>>evaluation could improve it tremendously is not in the spirit of my idea. >>>>>> >>>>>> >>>>>> >>>>>>>I understand that you are saying that it will change the style but overall >>>>>>>strength will not be much changed. >>>>>> >>>>>> >>>>>>I do not know exactly how far we will be able to go with the 10% I attribute to >>>>>>positional evaluation. >>>>>> >>>>>>I'm not saying it counts for nothing and that overall strength will not benefit >>>>>>from research in this area. >>>>>> >>>>>>I believe that the positional evaluation is the part of a chess program >>>>>>responsible for only 10% of the strength, and that the rest is done by the >>>>>>search. >>>>> >>>>>I do not understand like Amir what is the exact meaning of 10%. >>>>> >>>>>I believe that most of the amatuers can earn more rating from improvement in the >>>>>search rules and not from improvement in the evaluation but I also think that >>>>>the ratio is usually not 9:1 and I guess something like 2.5:1(I know that you >>>>>did not say that the ratio is 9:1 but it is a possible way to understand the >>>>>claim that search is responsible for 90% of the strength when evaluation is >>>>>responsible for 10%) >>>> >>>> >>>>:) >>>> >>>>You do not understand the meaning of my 10%, but you suggest that it is another >>>>number? >>>> >>>>So you must understand what I am talking about, somehow... >>> >>>I suggested a possible meaning for the 10% and said that by this meaning it is >>>another number. >>> >>>I did not say that it is your meaning and I see that it is not your meaning. >>>> >>>> >>>> >>>> >>>> >>>>>I guess that it is possible to improve most of the amatuers that are 400-600 elo >>>>>weaker than Junior by average number of 100 elo by doing a lot of work only on >>>>>the evaluation when you can improve them only by average number of 250 elo by >>>>>doing the same amount of work on the search rules without changing the >>>>>evaluation. >>>>> >>>>>The total improvement from working on both things may be bigger than the sum of >>>>>100 and 250 because after improving the evaluation the best search rules may be >>>>>different. >>>> >>>> >>>>I don't think the proportion is measured in elo points. >>>> >>>>My unit for the 90%/10% estimation is subjective. It's something like the amount >>>>of reward for a given programming effort. >>>> >>>>A successful effort in search get a reward 9 times bigger than a succesful >>>>effort in positional evaluation. >>>> >>>>Not to say that work on positional evaluation can be ignored. >>>> >>>>I notice that some chess players tend also to agree that chess is essentially a >>>>matter of search (tactics). >>> >>>I agree and I also think that search is the most important thing to work about. >>> >>>I think also that evaluation and search are connected and another thing to work >>>about is learning from the search to change the evaluation. >>> >>>Humans do not know that fortress positions is a draw by static knowledge but >>>learn from their search to change their evaluation. >> >>Humans understand a fortress position by logic, certainly not by search. >> >>Regards, >>Miguel > >I talk about positions when humans has no previous knowledge that the position >is a fortress. In those positions logic could be supplemented by a little of retrograde analysis if the fortress is complex, but "search" play almost no part in the "pseudo-static" evaluation of such positions by a human. Search might be needed to know how to reach the fortress, but the understanding of the fortress comes from pure logic and "static" considerations. A minimum of search might be required if a complex fortress could lead to a more simple fortress. Generally that would like a 1-2 plies search. The one on the Smirin match was a fortress that a human can understand with almost pure logic, i.e. describing it with words rather than variations. Regards, Miguel > >The logic that humans use include some selective search. >They do not know that it is a draw only by looking at the position without >analyzing and finding that they cannot make progress. > >They also cannot see the draw only by search without learning to change their >evaluation function because it is too deep to see repetition or 100 plies only >by search. > > >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.