Author: Dave Gomboc
Date: 14:18:07 04/16/99
Go up one level in this thread
On April 16, 1999 at 17:03:45, KarinsDad wrote: >On April 16, 1999 at 15:28:42, Bruce Moreland wrote: > >> >>It takes little effort and imagination to say that there might be patterns in >>these endings that would allow them to be solved algorithmically with something >>that is practical to write. >> >>It is not that people have not "gotten beyond their normal thought processes." >>People have been considering this for at least 20 years. ACC 2, written in >>1980, may as well have been titled, "Pissing Around With 4-man Endings", almost >>half of the book is on this topic. >> >>Beal did it with KP vs K, via 3 pages of Fortran. It is clearly the easiest one >>to do, other than KQ vs K and KR vs K. >> >>So here is someone that conceived of this idea and executed it 20 years ago. > >Bruce, > >I was not trying to say that nobody has thought about this before (obviously >that is not the case). However, I have seen no discussion on this in the 5 >months that I have been on this forum, so I threw it out to get people to think >about it. And of course it is difficult. If it was easy, somebody would have >done it by now. However, computer technology today is considerably stronger than >it was 20 years ago, so maybe some math whiz here knows of some pattern >recognition software that he can run against a tablebase that may enable him to >crack this nut for some cases. Who knows? I do not dismiss things out of hand >just because they are difficult. In fact, it is the difficult problems in life >which are worth attempting to solve. Today, writing an expert level chess >program is a piece of cake, it just takes time. 20 years ago, it would have been >extremely difficult. > >KarinsDad :) > >> >>Maybe it is possible to do it with more complicated endings, but I think that it >>would be very difficult. That is what would take significant effort and >>imagination. People tried it with KR vs KN but I think they all ran into walls. >> >>I've taken issue about this kind of thing in the past. Some things, the idea is >>everything and the implementation isn't that big a deal. Other things, the idea >>is not particularly difficult, but the implementation is a bitch. Deriving >>algorithms from endgame databases is in this latter category. >> >>bruce So KarinsDad (or whatever your real name is :), are you suggesting that such a rule base be built using the tablebase info, or are you suggesting it be constructed without access to a tablebase? If the former, what would be the point? We already have perfect information that fits on hardware worth a couple of hundred bucks. If the latter, how? Dave
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.