Author: Walter Koroljow
Date: 05:57:47 12/27/99
Go up one level in this thread
On December 26, 1999 at 21:53:27, Albert Silver wrote: >On December 26, 1999 at 11:30:33, Walter Koroljow wrote: > >>After 18 games, I am losing 1.11 points per game. I have studied the game some, >>so I am not a beginner. It's too early to tell yet, but it seems very strong >>(or I am very weak!). >> >>This is the "tutor" version and it gives numerical evaluations and suggested >>moves when it thinks I could have played better. It also gives evaluations of >>positions and, in particular, all doubles. It thinks its doubles are all well >>considered... > >That's interesting. I wonder how the doubles calls were chosen? Was it the >result of its own neural development, or does it incorporate the research of >players like Kit Woolsey? >I myself have only played the Jellyfish 3.5 demo, so I don't have access to this >information. Does it say in the manual or help notes? > > Albert Silver > >> >>This is a very nice toy. Now if I only had the time to study chess, backgammon, >>etc., etc. >> >>Cheers, >> >>Walter There is no manual, just a few sheets of paper. The on-line help is good, but does not talk about doubling theory. When doubling, or upon request, the program provides the following independent numerical information: 1. Probability of win, of gammon/backgammon, and of backgammon. I assume the "/" is a logical "or". 2. "Volatility", which is not defined. I haven't read Kit Woolsey's analysis, but I would certainly like to someday. Years ago I used win probability and a measure of volatility to generate doubling strategy (just to satisfy my curiosity). To me the hard part is defining and getting the data. E.g., is volatility a constant for the rest of the game? How much does its constancy matter? Have people figured all these things out already? Cheers, Walter
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.