Author: Robert Hyatt
Date: 17:15:31 06/22/99
Go up one level in this thread
On June 22, 1999 at 15:49:11, walter irvin wrote: >On June 22, 1999 at 13:54:19, Robert Hyatt wrote: > >>On June 22, 1999 at 11:47:16, Pete Galati wrote: >> >>>On June 22, 1999 at 11:15:24, Robert Hyatt wrote: >>> >>>>On June 22, 1999 at 10:43:53, Pete Galati wrote: >>>> >>>>>On June 22, 1999 at 01:16:28, James Swafford wrote: >>>>> >>>>>> >>>>>>On June 21, 1999 at 20:40:05, walter irvin wrote: >>>>>> >>>>>>> my program is not ready yet in the one game it drew the mach iv it was winning >>>>>>>but the search produced move that allowed counter play.i have to fix that >>>>>>>problem to me thats unexceptable play . >>>>>>> the reason its not an easy problem to fix is the search is totally different >>>>>>>than conventional chess program .its based on pattern rec .i only know it would >>>>>>>never play that type move again . but need further testing . >>>>>> >>>>>> >>>>>>Now I'm intrigued. Pattern recognition? In such a short period >>>>>>of time?? With results like that??? >>>>>> >>>>>>:-) >>>>> >>>>>I'm skeptical. But he's welcome to prove me wrong. >>>>> >>>>>Pete >>>> >>>>I've been falling all over myself this morning... Can't seem to walk because >>>>"something" keeps "pulling my leg"... >>>> >>>>:) >>> >>>Yeah, I didn't figure you had missed the mark so bad with Crafty, too bad you're >>>going to have to start all over again! Amazing. You might want to hang on to >>>your code for a while anyhow. >>> >>>Pete >> >> >>This happens about once every 1-2 years... IE about 2 years ago on r.g.c.c >>we had a 'new program' announced that was playing good chess. Many of us >>asked for games. "nada". We kept insisting... Finally two games were posted. >>Both matched _exactly_ the moves played by the current version of Crafty. No >>more was heard from this 'breakthrough'... >> >>:) >beating up on the mach IV is not much of a break through , also this program has >nothing much to do with crafty, crafty is good , right now more sound than mine >, but look at the amount of time that has been spent on crafty .crafty on my mII >300 mhz bench mark 85,000 nps, mine does not even search this way ,my 2 main >factors are hard drive space 20 gig, pattern rec .i want to make sure i win so >further refinements are in order . i think my concept is a break through but i'm >not there ,getting closer .to me there would be scores like 22/1 against rebel >ect without these bad moves showing up ever so often. people should be glad >there are people like me who will try new way of doing things maybe i have >success maybe not ,time will tell. > i know right now the program plays too unstable sometimes gm move sometimes 900 >move although i have weeded out alot of problems .plus no gui yet . > if all i was going to do was write another program that searched the way >crafty,fritz,rebel ect. i would not do it they are doing the best that can be >done with that method . this method so far some great and not so great results , >but this way is different unrefined so far , but im not going to give up because >people say it is impossible . All I meant to say/imply is that I am much more impressed by actual results than I am about 'promised results.' If you deliver, great. But the world of computer chess is littered with the corpses of "new breakthrough algorithms that make alpha/beta minimax obsolete".
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.