Author: Steven Edwards
Date: 09:08:34 04/11/05
Go up one level in this thread
On April 11, 2005 at 11:18:54, Dann Corbit wrote: >On April 11, 2005 at 11:05:47, Steven Edwards wrote: >>On April 11, 2005 at 10:34:10, Dann Corbit wrote: >>>I am curious if any parts of your engine (e.g. the C++ part) will become >>>available for experimentation. >> >>At the moment I am undecided as to how much of the toolkit source should be >>released. I will likely delay my decision on this until the programming effort >>is completed and the Big Paper has been written. My main reason for this is >>that I simply do not have the capability to answer even a small fraction of the >>support request e-mail that would be sure to come, just as has come in the past >>for other chess source I've released. > >Where will the paper be published? In print as well as on the net. It will probably be book length and maybe I can make a little cash from it. Alternatively, I could submit it as the dissertation for my 14 year suspended effort on my ABD PhD from Boston University; they gave Lewis Stiller one for his Connection Machine tablebase work and he never even distributed the data -- this was at the same time I was distributing TB data along with a generator program. Then again, I didn't use an impressive megabuck CM; all the TB work was done on my 1986 Macintosh Plus. >>On the other hand, a big motivator for the Symbolic project is to show how a >>cognitive search is a viable alternative to a traditional A/B search, and this >>is going to take a lot of documentation including at least some source and >>plenty of search narratives. I'll be uploading some of this to my website along >>with a FAQ; this will have to suffice until the Big Paper is published. > >I would be curious to see a graph of the project's Elo, as a function of added >functionality. Yes, this would be interesting. One caution is that some of the program's ICC opponents, rated many hundred Elo points lower, amazingly play 2600 Elo chess including perfect endgame play at times. (When caught, this gives them a permanent slot on the program's noplay list.) ICC fakery like this distorts the rating somewhat and is unfortunate as there is no easy way to avoid the problem. The Big Paper will surely have an appendix with test suite results over time including comparisons with results from other programs.
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.