Author: José Carlos
Date: 07:07:08 05/05/02
Go up one level in this thread
On May 04, 2002 at 12:32:48, K. Burcham wrote: > > >1. program A is down a point with last book move, has stronger search and beats >program B. > >2. program A is up a point with last book move, has stronger search and beats >program B. > >3. program A is down a point with last book move, has weaker search and losses >to program B. > >4. program A is up a point with last book move, has weaker search and beats >program B. > > >I know it must be lots of work to have a good book for programs. >If program C shows up in last four tournaments with inferior book, >and losses most games due to book, then we cannot know strength of search for >program C in tournaments. >Knowing and studying opening theory, has always been a big part of chess. > >In comp-comp tournaments, I would prefer to compare search methods with equal >hardware, and starting from a "neutral position out of book". >This way we are comparing search methods, time control, and eval between each >program. You can do that with Nunn test and things like that, but remember you won't be comparing _strength_ then, but something like "analisys capabilities". To test strength you must use the whole program, and this includes the book. Note that there're other people here that strongly disagree with my opinion. This is just that: an opinion. >Also it would seem that anyone that thinks they have created a comp "Book", >for a program---I dont think you have. >You have consulted 1000's of GM game openings, to put your book together. >I am not saying that you havent worked your a__ off in what you think would make >this a better book. I am saying, tell me the date you finished your book, >and I will go back and search all your openings, and compare your book opening >dates to previous dates all these openings have been played by humans. >It would seem to me that the only opening lines you can take ownership of, would >be a variation that you can prove has never been played. > >And to follow up this last point---if a GM studies his a__ off and finds a new >opening line or variation that has never been played, uses this line, gains >advantage, and mates Kasparov,,,this line will be studied and played from then >on, will be in everyones book. if a GM that put in time to find a new variation, >and proves it will work, he cannot protect his own creation once it is exposed >in tournament play. it will take on his name, but that is it. > >how many actual lines have the program book authors actually created from >scratch, that are not in any opening database, or have not been played by a GM? Many. I sometimes (when I have spare time to do that) study Averno's loses and analyze the opening. If I find there's a bad move there, I start my computers (4 including the laptop), set my board and start analyzing with different programs and my own ideas. I always come up with something new, that I insert in my program's book. I guess many people do something similar. José C. >I like the idea, about putting in code that will allow another program to play >an inferior line, but wont allow your program to play the same inferior line. >this will protect the book that you have tuned for comp-comp play using someone >elses opening theory. > >kburcham
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.