Author: Robert Hyatt
Date: 12:08:47 03/04/98
Go up one level in this thread
On March 04, 1998 at 11:04:30, Ed Schröder wrote: >>Posted by Robert Hyatt on March 04, 1998 at 08:34:58: > >>>Perhaps I don't like to be called stupid? >>>Of course you never did... >>>Pfffff.... > >>>- Ed Schroder - > >>Then perhaps "stubborn" instead? > >Much better but wrong again :) > > >>I recall several responses to people asking you "why no book >>learning?" And I notice that nothing has changed about learning >>in the last few years. > >Then you have noticed wrong. >Your information about Rebel is out dated. > > >>And I notice that it took me about one night to write Crafty's >>learning code. And perhaps another 3-4 nights spread over a year >>to refine it... > >Sure, because you are not stupid :) > Let me explain a revolutionary concept to you: writing a stupid piece of code, does *not* mean that the person that wrote it is stupid... examples: (from my code) 1. I didn't have book learning for 1.5 years of Crafty, plus 20+ years in Blitz/Cray Blitz. I simply chose to not tackle it. Not doing so was stupid, in hindsight, but I wasn't then (nor now) what anyone would call a "stupid person." 2. In the NPS game last year, I had some silly mistakes in the evaluation that favored that passed center pawn over king safety. Again, stupid programming. Bottom line: You chose to not do book learning until (apparently) Rebel 9 (Yes I know about the memory-resident tricks in rebel 8 but that's a SSDF trick, not something that helps normal book play from what you have said in the past.) That was a stupid mistake I'm sure you'd agree. But there's no implication that *you* are stupid. I've probably made far more stupid mistakes than you have in programming issues, no doubt, because I have written several million lines of code over the past 30 years. But again, I would not call myself stupid... > >>Tis far easier to take evasive action by fixing a serious problem than >>it >>is to (a) come up with a passive solution that requires manual/automatic >>elimination of doubles or (b) grumbling about the way the testing is >>done. > >Ahem, we will see when Crafty appears in the SSDF list. it may lose every game. But it won't lose the same game twice to the same opponent on the same machine, *ever*. > > >>As far as the NPS thing, we both know how that was likely going to turn >>out had 10 games been played... That hasn't changed at all... > >I was referring to the word stupid. If you call me stupid then it is >allowed to pester in return. Be sportive... as I said, I did not call you stupid, nor did I mean to imply such. We can both be smart as Albert Einstein and still make stupid programming decisions from time to time? > > >>As far as being called "stupid" I don't believe that was directed at >>you. > >As predicted... again, see above... > > >>Several programs fit that description, from Rebel > >Again? > >Not many people will call Rebel a stupid program. I'd call *any* program stupid that will let me beat it, and then let me play the same moves again tomorrow and beat it again. Check your definition of "civilized man" -- who learns from his mistakes. I've had a stupid program in that regard for nearly 30 years. No More. I may have a stupid program in other regards, still... of course... But every program has some really stupid weaknesses. Check out your king safety against kingside attacks to see what I mean... Doesn't mean you don't have offsetting strengths, of course. First game I played vs Rebel 9 on a server showed this problem. Rebel seemed to not notice that Crafty had two bishops on the queenside with open lines to the kingside, and played g4-g5 and ended up attacking itself. On balance, Crafty has done similar or even worse things, of course. But a "smart" program doesn't "self-immolate"... usually... I haven't seen but maybe one "smart" program to date.... > >Hope you are in the minority dear colleague... :) > >- Ed - > > >>to fritz 4 and before >>to Genius (earlier versions, don't know about 5), CM versions (don't >>know >>about the most recent) and so forth. And based on (a) the known problem >>with books and (b) the minimal effort needed to get learning >>operational; >>then (c) there was no good reason for everyone not to be "learning" at >>least a couple of years ago.
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.