Author: Robin Smith
Date: 22:19:28 05/16/05
Go up one level in this thread
On May 16, 2005 at 23:54:43, Komputer Korner wrote: >Since my 5 years away from computer chess, I have noticed that Backsolving has >spread from program to program like an unstoppable weed. Chiefly promoted by >Mike Leahy, it now threatens to permeate computer chess the way that >chiropractic has permeated medicine. If the lie is big enough, it is >unstoppable. However unlike medicine, there are no unknowns in the theory of >computer chess. The only unknowns are moves that haven't been played yet but the >actual unplayed moves are irrelevant to our discussion of the merits of >backsolving. Perhaps the single biggest argument against backsolving is you need >an actual evalustion at the last node. Without an evaluation you can't >backsolve. However why would anybody depend on the last moves of a game to >decide what the evaluation will be at the first nodes? Remember, backsolving is >used in conjunction with studying openings. There is a long way between the >end of a game and the opening. So then others would argue that okay we don't >need to backsolve the last moves, we will only backsolve all the nodes from the >15-20th moves all the way back to the 1st moves. Isn't this studying openings >from the wrong way? I thought studying openings was spending my time analyzing >the first moves of chess games, not the last moves. But the backsolvers argue >that if you know the result then you know the beginning. However you really >do't know the result. You only know an infitessimal number of results compared >to the total. Therefore you are wasting bytes and time taken to actually >backsolve. don't forget that backsolvers always want every node annotated. that >is the whole purpose of backsolving. however moves are added to the book one >line or move at a time. You could dump other books and into your book and batch >backsolve but most of the time you are adding one move to your book at time. if >you have to backsolve each time, then you will be doing alot of backsolving >over the years. If you don't backsolve everyday but instead only backsolve every >week then you have defeated the purpose of backsolving because backsolving means >always having every node annotated. If you don't have every node annotated then >you are likeme and my opening book, just the critical nodes at the begiining of >the book are annotated with all the deep lines pruned off because you will never >get to those moves anyway because of the vast middlegame number of move choices. > My opening book is just that, an opening book, not a complete games book >desguised as an opening bok. But the true backsolvers will say that they prune >the moves as well. If that is the case,then they don't need to backsolve because >they can manually add in the annotations wherever there is a fork in the road. >(more than 1 move choice at a node). So in the end what does backsolving >accomplish? NOTHING................. Of course backsolving, like any other tool, can be misused. You point out some potential pitfalls. However that does NOT mean that backsolving does not have some _very_ valid uses. That you don't seem to understand these uses does not mean they are not there. How about if you spend your time trying to understand when and why backsolving can be a very useful tool, as many others have already pointed out, rather then waste time trying to convince people that a screw driver is no good for pounding nails. -Robin
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.