Author: Komputer Korner
Date: 15:29:10 05/17/05
"However unlike medicine, there are no unknowns in the theory of computer chess." When I said the above, I really meant to say that there there are no unknowns in the mathematical theory of trees and tree paths especially chess trees which is what an opening book is. The one exception is the travelling salesman problem which in the end may be impossible to develop any algorithm to solve. However the travelling salesman problem has no relevance to a chess tree. I repeat: I can see no value in backsolving whatsoever. Some posters have tried to connect backsolving with classification of positions. The two are not the same. The process of starting at the end of lines and backing up in order to classify lines is very different than what backsolving actually does. Backsolving as per the way that Michael Leahy (and others) does it in Bookup is a process to annotate positions. This has nothing to do with classification of positions. We can get into another thread on a discussion of classification of positions if you want but in this thread we are talking about backsolving. It simply copies the end node evaluation backwards to the beginning of the line and wherever there is more than 1 branch, you need an evaluation at all the end nodes for the backsolving to work properly. However as I tirelessly explain to the great "UNWASHED masses", backsolving doesn't help you in any way to speed up or strategically enhance your opening book study. Indeed it slows your opening study down because you waste time implementing the backsolving process. A key question involving any use of computers apart from their fun aspect is DOES THE AUTOMATED PROCESS GET YOU A BENEFIT FOR A NET GAIN OF TIME? In the case of backsolving there is no gain of any kind. Studying openings demands that you put together an opening repertoire from the beginning of the game. If you decide to do it the backsolving way, you will constantly be backsolving any new input into the tree with extra bytes needed for storage of the annotations against all the nodes. So that is the cost. What is the benefit? There is none because any time there is a choice of moves in your opening repertoire let us say EX: 12.Ng5 or 12.Be6, you are better off simply to put an engine or a great mind to the task of evaluating those 2 moves themselves; NOT some lonely esoteric result 40 moves later in each line. Backsolvers are in the fruitless task of adding whole games to their repertoire in the hope that the whole game result is an indicator of the value of the move node back at the 12.Ng5 12.Be6 fork in the road. Even if the backsolvers prune off the very last moves at the point where the advantage for one side approaches a pawn, they still have to make their opening books unnecessarily large and they still are not as well off as entering and analyzing more opening moves at the beginning. That is because the probability of actually someday playing the other opening move lines near the beginning is far larger than the probability of ending up in the 30 move line. Backsolvers have it all wrong. They won't solve chess and during the process they are left far behind others who are concentrating their efforts at the beginning of the game when doing their repertoires. Michael Leahy has sold you SNAKE OIL. I am not saying that Bookup isn't a good program. All I am saying is that the backsolving feature is worthless.
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.