Author: Marc van Hal
Date: 10:28:00 11/12/02
Go up one level in this thread
On November 11, 2002 at 19:48:09, Eelco de Groot wrote: >On November 11, 2002 at 16:10:13, chandler yergin wrote: > >>I congratulate the authors for this one volume Opening Encyclopedia; a massive >>project & well done! They did use ChessBase & Analysis Engines to evaluate. >>In light of the many Positions recently posted; where Fritz & Hiarcs could >>not solve some simple positions in a reasonable timeframe; I wonder what >>timeframe might have been used by the authors before concluding the analysis was >>correct. I must admit some disenchantment with these Top Programs; and recognize >>that in a Tournament situation under the USCF Time Controls, the Programs, >>Tactical monsters they may be, often cannot properly evaluate some simple >>positions. Under the restrictive time controls for these Human/Machine matches, >>Opening Books, Horizon effects etc, I suspect some of the Humans are just >>laughing all the way to the Bank. Am I over-simplifying..? What am I missing >>here? Just a Novice here asking what maybe a dumb question, but would appreciate >>any thoughts/comments. >>Thank You! > >Hello Chandler, > >I think NCO is an outstanding one volume opening book. It is already three years >old, which if you are a tournament player that wants to know all the latest >novelties probably makes you yearn for an update. Alternatively you can buy a >chessprogram with a really good opening book like Jeroen Noomen's work, and use >it to check your favorite lines to see if they haven't been sunk recently. But >NCO is very good to get an overview, to explore openings that you don't know >very well. The authors used Chessbase and custom software extensively in >formatting their book and assembling the information. A lot of the actual >analysis is still honest handywork though, mainly for the reasons that you >mention. With such a huge amount of data to evaluate it is next to impossible to >assemble that without errors creeping in. That I think is where the computer >analysis was most useful, by letting it calculate along the lines they were >considering, it was possible to at least eliminate the gross tactical blunders. > >I quote John Nunn from the introduction: 'All the authors used Chessbase and a >large database to assemble the information for each table. This was then pruned >down to roughly the right size, while an analysis engine looked over the >author's shoulder, ready to spot any nonsense that might otherwise creep into >the book. The author then evaluated each line.' So I don't think many positions >were analyzed very long by computer, there simply wasn't the time for that kind >of investigation. The authors could have made use of computer analysis like from >the CAP project, but everything still would have to be evaluated by a human >expert to get to really the standard of evaluation that was achieved here by >John Nunn, and his coauthors. I doubt they got much salary per invested hour for >their work, even if the book isn't cheap to come by... > >Regards, Eelco The time used to solve the problems is not of importance. Like my analysis have proven ove and over again. You simply have to look for more positions to get the the right evaluation. You also should not only look what the program says but also look your self what the right move might be acording to you. Especialy looking for a solid plan. Looking at the problems the engine might face could help here. The difference between a program and a human then is that the human detects the problem in an earlier stage. So finding the solution is more easely. Most important is never take anything the program says for granted! Keep on looking yourself how stupid the move you find might be in the programs point of vieuw. I once had a position on the board wich was evaluated -1.72 by the programs (They still do) 1 ply further the evaluation was +3.00 about the diference of full rook. Regards Marc van Hal
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.