Author: Frank Schneider
Date: 00:08:17 02/19/99
Go up one level in this thread
On February 18, 1999 at 16:11:47, Peter Kasinski wrote: >Last May, I decided to write a chess program. >A computer professional and a bona fide computer chess addict I always felt I >ought to give it a shot. Surely some chess skills (currently 2263 on ICC) would >help me, and who the hell are these guys who dominate the Swedish list anyway >:-). > >I took TSCP as a starting point >and after 3 months (after hours, such as life) I >started to think of a name for my program. My code still had sections exactly as >Tom left them (interface changed very little, for one), but I felt I had changed >enough. Search was rewritten, move generation and q-search changed, null-move, >hash tables, selective extensions, the works. > >At the time I received hints and answers form this forum. These helped me >immensely. I also received direct pointers to Bob?s code (one email came from G. >Mueller) which I thought was disappointing if only because of the assumption >that I couldn?t think of it myself. But in reality I felt intimidated and >discouraged by Crafty. Why? OK, it is very strong and one has to get accustomed to loosing against it, but why not? Crafty is much older than your program and it is normal that it is better. >It didn?t take long to realize that no satisfaction could >come as a result of implanting this stuff into my program. Again: why? Almost all free programs are based on the same algorithms and heuristics and using them is ok (but copying sourcecode is a different story). Inventing new ideas is of course more interesting than implementing well known things and that is one reason for the discussions about Bionic and Voyager. If you want to test new ideas you will not want to write the n'th openingbook, userinterface, winboardinterface and so on. It is very ineffective that many people (including me) write so many things by themselves instead of reusing them. So it makes sense to base your experiment on crafty, and in that case you have to make the changes publicy available. But if you want to compete with other programs I think there is a difference and reusing for example crafty is an unfair advantage. The question is: science or competition? > I ended up coding a >poor man?s hashing algorithm and hell it felt great. I took Pedestrian to ICC >and (operating it manually) played some games. I was wiped by the Amateur but >Will was nice and offered encouragement. My rating stayed around 2000 and given >a LONG list of bugs and no endgame knowledge I knew I could improve it. When it >called the Shirov mate-in-14 in under 60 seconds (PII-333) even my wife smiled. > >After the initial outburst of energy I took a break and didn?t do much in the >last few months. Now with the Voyager case I don?t know if I want to anymore. >Even before I thought of all these beginners who enter the list at >stratospheric heights > but somehow it didn?t turn me off as much as this recent affair. To me >the idea that anyone could come up with Crafty as their first chess project is >preposterous. I've worked on Gromit for some years (reusing nothing but a few lines of book from an old gnuchess) and whenever a very young program is much stronger than Gromit I feel kind of stupid. But knowing that Bionic and Voyager are based on Crafty I have no problem loosing against them. By the way: it took me a long time before Gromit was where TSCP is today. > How reassuring it was to listen to Amir (on Junior CD) saying that >when he was around 17 he had thought about writing a chess program but decided >that it might have been too difficult and wrote a mate-in-2 solver instead. > >I have tons of respect for Bob. But Crafty is not a realistic staring point for >beginners. Don't forget that your program is based on tscp, and I'm sure you saved a lot of time using it as a starting point. > Bob?s contribution to our community can hardly be overstated, but I >would rather read his answers than look at Crafty. On the other hand some people >will use it to gain an unfair advantage over the very beginner it is supposed to >help. > >PK. > >I would like to close with this suggestion for an opinion-poll question: > >What would be the rating of Crafty if Bob had access to the source code of >Hiarcs 7 and CM6000 for, let?s say, 45 minutes? Frank
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.