Author: Peter Skinner
Date: 00:14:17 01/29/06
Go up one level in this thread
On January 29, 2006 at 02:50:36, chandler yergin wrote: >On January 28, 2006 at 17:43:37, Ross Boyd wrote: > >>On January 28, 2006 at 16:27:48, chandler yergin wrote: >> >>>Mark Twain once said.. "There are Lies, Damn Lies and .. Statistics" >>>It's hard to disprove Statistics, they are what they are. >>>Four hundred years of Chess Theory & Millions of games played have proven >>>that White has a significant advantage with the first move. >>>Black cannot overcome this, in fact Black cannot even equalize. >>>Whites Overall Performance is 55% compared to Blacks 45% >>>The error factor is insignificant. >> >>>Whichever Human or Computer plays the Openings by ECO Classification >>>that are better for White will Win. >>>Whichever Human or Computer plays the Openings by ECO Classification >>>that are better for Black will win. >> >>Don't you mean they will improve their win/loss percentage? >Yes Thanks.. >> >>>It's really that simple. >>> >>>So what does Testing Engines against Engines Prove? >> >>Not sure what exactly you're saying here. >> >>Testing engines vs engines is just a way of gauging relative strength. Rigorous >>testing against humans is prohibitive time-wise and there are other variables... >>humans have 'off' days which can skew the results... a computer sparring partner >>can be relied on to perform consistently. >Exactly, thanks. >> >>If you mean in the context of Opening books, its better to test using a set of >>standard balanced positions so that it is raw strength we are measuring rather >>than how good an opening book is. >Yes, Opening books are very shallow really.. >> >>Tuning the engine's opening book is just another step in an engine's >>development. >Yes, especially in Engine vs Engine games. >> >>Ross > > Whites Overall Performance is 55% compared to Blacks 45% > These seem to be the Magic numbers. > Millions of games played confirm this. > What would you conclude from this? > > >Anyone? >Thanks, >Chan Ross is simply stating what I tried to explain to you. When testing an engine it is best to use a set position rather than an opening book as you can tune books for opponents. When using single positions or even the Nunn positions you can see where the search/eval is lacking and needs to be tweaked or corrected. Since the opening book is only one facet to optimizing a chess program, the statistics of white/black wins is irrelevant in the context we are speaking about. I am not doubting your numbers at all, but rather the way you are trying to use them to disprove a well known fact. Engines _can_ be tuned outside the opening library, as _any_ position can happen at _any_ time. The search/eval then become more important than what was done from the opening book. Knowledge is one way a program can be improved, search speed it another, praying to a Buddha can be one as well. The opening library is _not_ the be all and end all of chess. Peter
This page took 0.01 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.