Author: stuart taylor
Date: 05:42:34 09/10/00
Go up one level in this thread
On September 10, 2000 at 07:12:52, walter irvin wrote: >On September 09, 2000 at 23:22:38, Andrew Dados wrote: > >>On September 09, 2000 at 17:12:08, walter irvin wrote: >> >>>can any programmer here fore see anything that might make a major diff in >>>computer elo???? im not talking about getting 4 elo point if we double the hash >>>size or some non sense . >>> >>>if not then maybe we have went as far as we can go in the current direction???? >>>maybe a new direction would help ???????????????????????????????? its fine to >>>make copies of crafty and tweek it here and there and say oh im a big programmer >>>.but whats needed is for some new ideas , some one not afraid to fail or try >>>something people say wont work . >>> >>>just a sugestion >> >>What exactly do you suggest? I somehow miss the point of your message. >> >>-Andrew- > > >i suggest we keep doing exactly what we are doing getting the exact same results >, that sound good to >you????????????????????????????????????????????????????????????????????????????????????????? >of course you miss the point because it takes inventiveness and the ability to >come up with NEW thoughts and ideas .most people even the ones THOUGHT to be so >smart are not really much more than parrots just barking out what they hear >.just off the top of my head here are some directions to look at . > >1.learning =current learning is subpar and worth very little ie pattern rec >where it alters the eval based on position ect .this is a huge untapped resorce >for future chess programs .everyone wants to be lazy and wait for processor >speeds to save their weak programs . >2.anti-GM =more imagination is needed here whats bugs people most is the >unknown so a good anti GM setting would be to have several settings and the >human would have no idea which ones he was playing against . >3.openings-openings should always be geared toward either the playing style of >the program or at the very least when in anti GM mode geared towards positions >that make people squirm . >4.should know how much time the other player has on their clock and when time >gets very low move faster to keep other player off balance to make them lose on >time it works for me vs humans. >5.computer styles =since no one knows how to program a computer to play closed >positions then every effort in the world should go towards opening books that >are open in nature ,plus if left on their own programs should choose open lines >of play . > >i have many more ideas but you get the picture. Did Hiarcs skip a year because 2 years ago he made an upgrade good enough for 2 years? (basically, hiarcs 7.00 was the last upgrade. 7.01, was a little patch, and 7.32 was redoing it in 32 bit, and was not necesarily stronger, according to ssdf.)I didn't feel that the programs of this past year were convincing enough as upgrades. A bit convincing, but not enough. My old Hiarcs 7.00 was not really outplayed. S.Taylor
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.