Author: Uri Blass
Date: 11:08:32 11/27/02
Go up one level in this thread
On November 27, 2002 at 11:04:47, Ricardo Gibert wrote: >On November 27, 2002 at 08:15:46, Uri Blass wrote: > >>On November 27, 2002 at 07:57:09, Bruce Cleaver wrote: >> >>>I believe there have been world-class backgammon progs that used a 1-ply search; >>>i.e. all legal moves are explored to one ply and the best is played. I am not >>>sure what this tells us. The branching factor for backgammon is ~400, including >>>dice rolls and doubling cube. >> >>I think that this tells us that humans never knew how to learn backgammon >>seriously and programmers who thought about the problem could teach programs >>things that humans do not know. > >I may have created a misimpression. The top BG (=BackGammon) comps and the top >BG humans are very close in strength. I *personally* give the edge to the BG >comps. Because of the element of chance intrinsic in the game, there is a lot of >wiggle room in accessing who or which is better. You can argue all day. > >The BG comps have learned a lot from humans by using neural nets. It is also >true that humans have learned a lot from comps. Comps have altered the way >humans play for sure. They've been highly influential. > >The other thing to bear in mind is that BG is played very quickly compared to >chess. It is more like blitz even at the highest levels. > >> >>I think that it is also possible that in some games like backgammon part of the >>good players of today became good players thanks to the fact that they are good >>programmers and >>they started to learn the game by generating a program to calculate the right >>evaluation function and only later they learned the evaluation from their >>program to get an advantage against other humans. >> >>Uri > >As in chess. it is hard to find the individual in BG who both excells in the >game both as a player and as a programmer. In BG, you *will* find many human >players who are stronger thanks to BG programs written by some else. In the near >future, I believe the same will be true in chess. I do not know It is possible that there is a world champion that is also a good programmer but nobody knows about that, because he told nobody about it because his target is to be the best player and not to sell his program. If he sells his program other people are going to learn from it, so he is going to lose the world championship. I do not think that there is a contradiction between the ability to be the world champion in chess and the ability to write the best program. Good ideas are important for both tasks. Uri
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.