Author: Robert Hyatt
Date: 14:12:58 01/24/99
Go up one level in this thread
On January 24, 1999 at 13:02:33, Frank Quisinsky wrote: >Hello Bob, > >I understand what you mean, however you were not present in Leiden so you have >got information different from mine. > >Regarding the SMP protocol of yours Hans Secelle apparently made a mistake. He >did not know that he might not apply the protocol. In order to get equal chances >Bionic still employs the SMP technique. From what I know, Hans has offered his >apologies for this. Hans has nevertheless in the past proven that he can program >without Crafty's source code. If you knew Hans personally you would feel >different about him, I am positive. He is a witty and open-minded person. > First, let's separate the 'people' from the 'issue' here. I don't know the Bionic people. I don't have a negative thing to say to or about them, not knowing them. I'm not going nuts about what happened. What originally got my attention was email from 5 different participants in the Dutch tournament, somewhere during the first weekend. They were pretty upset. All play crafty regularly on ICC and all get beat regularly. Partly because crafty is perhaps older and more mature and more robust than their programs. Partly because crafty uses a 'big hammer' when it plays, a multiprocessor machine that really gives it a significant performance advantage over even the fastest PC you can find (ie 450mhz is the best pentium so far, I'm running 4 X 400 PII/xeons which are about as fast on a processor by processor comparison. or about 4x faster overall). Lets go back to the 1970's... I was going to the ACM events, and everybody was doing different things... some good, some bad. We all compared notes, worked for another year, and came back primed with new ideas. Imagine if Slate and Atkin had made chess 4.x public domain, and when I showed up at an ACM event and faced a field that was 1/2 chess 4.x clones? I'd have had little chance, and may have given up after getting clobbered a few years as who would be willing to support such work? That has a very real chance of happening now. Dual PII's are cheap. And they are 2x faster than a single cpu. And if you don't think 2x gives crafty a big advantage, you should try it... And now we stand the chance of seeing all new programs get thrashed over and over by the same (or nearly the same) program using different names. And that puts me in the position of having done something bad for computer chess, when I had intended the opposite. I have given a lot of thought to no longer distributing this stuff. But each time I talk myself out of changing... because I believe that this is the right thing for _me_ to do... I have enough experience doing this stuff that it makes sense to pass it along to newcomers, and supplying source code is a good way of accomplishing that. Although it obviously has its bad points too... >I wouldn't like to meddle in this too much because as a mere user I simply do >not have that right. I wrote a report about the Dutch Open for the >Computerschach und Spiele magazine and the contacts I had in Leiden with many >programmers gave me the appropriate background info (I was there during each of >the rounds). > >The whole matter is, I should say, silly and has been conducted stupidly! > >Hans was honest and directly said that he used parts of Crafty's source code. >This could be crazy or not, but he has not profited anything from YOUR work! > that's hard to say. He beat quite a few amateur programs in that event. _they_ would probably feel differently of course. And in the bionic project, it takes 'results' to keep someone interested in maintaining the project, paying expenses, and so forth. I know. I've been there doing that for years with Cray Blitz... >However, I must still add that I read with great interest what Ed Schroeder >wrote about this issue and I think that he is right. > >The source code of Crafty (so helpful for other programmers)has led to more >Freeware programs lately and it is often unjustly assumed that Crafty is behind >it. There are different free sources and codes on the Net. For example, the code >of Phalanx, which, on the basis of our results, has proven to be a first-class >chess engine and is at least as good as Crafty. > >For the development of computer chess this is great, because other programmers, >Hans for example, have nice ideas and can incorporate them into an already >invented "automobile" (The German concept is paraphrased -- why should anyone >try to reinvent the automobile?). In this way computer chess is advancing and >the end-user profits. > >On the other hand, how many Crafty-clones play openly on the Chess Servers? dozens... but all say 'crafty version xxx'... there is _no_ doubt what they are running... > >Regarding WinBoard it is just great that programmers can see for themselves how >this perfect integration based on Crafty functions. Or, Crafty can point at how >to implement good hash table support. So, more and more programs are being >created. That is not only thanks to Crafty, but Crafty has contributed to the >development of computer chess and for this as a user I would like to thank you. >I am pleased about new programs such as Bionic and would very much love to use >all the programs all at once. > >Nevertheless, Phalanx, to name one, just like some other programs have nothing >whatsoever to do with Crafty. > there are many non-crafty programs around, many becoming public. Phalanx is but one. The crazy bishop, knightcap and others are following the idea, which is to share everything, as opposed to what we have seen from the commercial programmers for the past 10 years... >I understand both you and Schroeder very well and was grateful too for the >interesting explanations offered by Don and Bruce. > >Please, believe me, Bob. The results of Crafty 15.18 to 16.3 do not match the >results of Bionic Impakt 4.01. The evaluations and moves are different. I can only say that for the first weekend, the games Vincent sent me didn't just match, they _really_ matched... I tried three games... > >If I am to prove this to you, could you please send me the positions you wish >and I will check them honestly with the Crafty versions 15.18 to 16.3. Just tell >me in detail, please, which time controls and settings to use. > >My Email address is: Quisinsky@in trier.de > >One more thing: from what I know, Hans apologised for his mistake and should not >be treated so roughly here even if it is your source code and even if a lot of >your work is built in this code. We all know this and appreciate your >interesting postings in this forum. But, please bear in mind that Hans has too >has put in a lot of effort and implemented his ideas in the code. I have no intention of treating Hans 'roughly'... if it came across as such, I certainly apologize for it. I am treating the 'concept' with a lot of contempt however, because while one 'programmer' benefitted from this, several others got a '0' in their result column for the same thing. I did _not_ intend to hurt those guys in any shape for form, but indirectly I certainly did... and that is the part of this I do not like. At least the Paderborn WCCC won't have this problem, but it will come back no doubt... > >In any case you can contact Hans Secelle or exchange your experiences and >perhaps Hans could contribute with a tip so that Crafty would continue to >improve. this is another 'sore point.' Ie it seems to be 'ok' to take 50K lines of code and work on it, but no feedback? That has been the first such case I know of, as many others are doing tweaking and send me results when they find something worth trying. That's exactly how Linux has become so popular, people use it, and then contribute their ideas and effort to the project. the 'bionic' project hasn't done that. Ie it is like a one-way communication with a commercial chess project, they listen, but they don't talk. That is _not_ the spirit in which crafty was initially released, and it was _not_ what I intended... > >I have hopefully done with this explanation and am back to our WinBoard >tournament. In round 7 Crafty won its match against Comet 2.5-1.5 (Kai Skibbe is >still running the remaining 4 games against Comet...) scoring thus a victory >against its fearsome opponent -- as Comet has so far garnered more than 50% in >its games against Crafty! > >Greetings from a fan of Amateur programs! > >P.S. I sat here the whole night trying to get everything right and my English >was getting better, but is still rather lousy, so I had Djordje help me >translate this text. > >Best wishes >Frank grammer I can work around, of course... :)
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.