Author: Sandro Necchi
Date: 12:55:47 09/12/04
Go up one level in this thread
On September 12, 2004 at 15:28:32, Sune Fischer wrote: >On September 12, 2004 at 13:58:58, Sandro Necchi wrote: > >>>I don't understand why this is such a big deal, a chess package consists of >>>several things like GUI/eyecandy, database/books, engine, server account etc. >> >>Yes, a chess package in reality is the program: >> >>the engine is checked by the team working on the book and the lines are selected >>to suit that program. >>Then test games are made. >>At that point there are talks between the programmer team and the opening book >>team and they study COMMON things to make further improvements; I mean how to >>improve what is not played well enough related to the selected lines. >>This means also that some lines/defences are avoided as not suiting the program >>style. >> >>At the end the program will be a package of opening book, chess engine and >>endgame database. >> >>To replace one of them may cause a weakening of the program. >> >>Like there are no players that can play very well every opening/defence, the >>same is true with chess programs. >> >>So, everybody can make any test they want, but the tests are pratically >>meaningless because of above. > >Hi Sandro Hi, > >I disagree that it would be meaningless. > >You're right of course that removing a book that has been worked on for years >and is perhaps optimized for an engine, will no doubt weaken the book + engine >package. > >However that is also the whole point of it all, to see how much strength comes >from the special book and how much from the engine. I understand this, but you should check this on the positions the program likes, not on the ones it is not planned to play. You see, sometime in order to make the program play better those openings it is necessary to make changes which weaken the program on the positions it does not play and testing it in those you would not have a correct picture of the program strenght. > >I think we're just looking at it from two different perspectives, your angle is >that the engine and book must be used as a team going into tournaments. > >My angle is that most of the time they are used for analysis by serious chess >players, and it is with this in mind they should be tested. I understand your angle, but the best way is to buy more than one program in order to use the one which is best in that particular line. How to find which one is good for that line? Checking if that opening is played at tournament level or not (in the book). > >One way to make this test is to create a broad variaty of opening, middle and >endgame positions and then let the engines play both sides of the board against >eachother. Again doing this you will see that some program will decrease the strenght or increase depening how suiting to them are those positions. > >I think someone suggested having an analysis Elo and a tournament Elo. >Doesn't sound like a bad idea at all actually :) > >>To make it more clear: >> >>If you take the Ferrari engine in a Mercedes car, with tyres taken from BMW, you >>do not get a very good car unless these are modified to be combined all >>together. >>The same is true for a chess program (which is a package as you said). > >Imagine if the Ferrari engine had twice the power of the standard Mercedes >engine and the BMW tyres had twice the grip of the normal Mercedes tyres, then >it might not be a bad combination after all :) > >>>I expect a program do the best it can, even in objectively lost situations. >>>There is honour in fighting for a draw as well :) >> >>The program does only what it is designed for and will fight until the >>end...unluckily they do not get VERY VERY UPSET when they are forced to play a >>line they do not like at all by the owner, but maybe in future we will have such >>option available full of bad words too... >> >>Wouldn't that be nice?:-) > >Rise of the machines? Let me think about that.. ;) Of course they will...it is just a matter of time...:-) > >-S. 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.