Author: Robert Hyatt
Date: 19:02:14 08/29/98
Go up one level in this thread
On August 29, 1998 at 21:44:04, Howard Exner wrote: >On August 29, 1998 at 18:35:13, Shay Bushinsky wrote: > >>Yudasin vs Junior 5.0 >>================= >> >>We have completed three interesting and exciting days of a match against >>international grandmaster Leonid Yudasin. The event took place in the luxurious >>Hyatt Regency hotel, Ein Bokek, the dead sea, Israel. >> >>The conclusion of the match was as follows: >> >>1. Standard time control part (120/40 and 60/rest) 1.5 - 0.5 in favor of Yudasin >>2. Rapid time control part (30/all) 2 - 2 draw >>3. Unofficial blitz match (5/all) 4.5-1.5 in favor of Junior 5.0 >> >>We have only but appreciation to grandmaster Yudasin for taking the >>challenge and playing great chess against Junior. Yudasin, is not >>only an all round first class positional expert and tactician, but is >>a very spiritual person as well. He had taken the challenge beyond >>just another chess match to really try and investigate the differences >>between the human thought process in chess and the way computers >>play the game. > >Yudasin sharing his thoughts on the games is an added bonus. It sounds as if >he was eager to communicate with you about this. >> >>Along the match, Yudasin seemed to have changed his mind in his >>effort to classify Junior as a chess playing entity. >>Indeed he reckoned that Junior 5.0 is of above 2500 playing strength. >>Though after the standard time control part, Yudasin felt that Junior >>is an improved tactical "brother" of his predecessor. It was after game 3 >>of the rapid match, Yudasin had recognized Junior's positional strengths >>as well. >> >>In his summation of the match, Yudasin felt that the result was justified. >>He believed that he is still slightly superior in the slower time controls. >>Yudasin tried to explain the difference in what he termed human's "second order >>of chess understanding" - the ability to appreciate when positional features >>are good and when the same features are bad - giving the example of Junior's >>central pawns in game 1 which were a disadvantage due to their relative >>weaknesses. > >This will remain a hurdle for programmers - deciding on what features make >a position good or bad. The example he cites from game #1 is something I'd >like to see a strong GM like him expand on. Perhaps even in a book once more >40/2 games are out there to annotate. > This is actually an interesting positional point. Bruce and I had this discussion a couple of years ago, when in a similar position, crafty had one side ahead, bruce had the other doing better. I don't remember who was playing, perhaps a deep blue match or something, but the position was similar... I pointed out that the "duo" was very weak, while he was scoring it as being very healthy since the two pawns were side by side. This came from quite a few GM conversations a year prior to that, even. IE in the first Junior game, Crafty evaluates those two pawns as simply two weak pawns, because each is undefended by a pawn, can't be defended by a pawn, and can't advance because of black pawns attacking the squares in front of them, which means an advance turns into an isolated pawn... There are times when it is wrong... but in general, such ideas are important. And they come best from either experience (I did this same analysis in Cray Blitz because of one particular sicilian where white had no c-pawn, and had played d5/b5 leaving black's c7 pawn very weak. But we didn't catch that and the pawn was eventually lost... Lombardy and I had a few long conversations about such ideas and I refined what I was doing as a result... But there aren't many that are willing to talk for free and explain and refine such a positional idea... Lombardy was (although I haven't talked with him in a year or so) as Roman is (I talk to him regularly, via phone and chatting online)... We need more of these guys... I also find IM's more willing to talk, and they are also quite valuable as resources... >>Yudasin also thought that if computers would be given a "relative sense" e.g. >>would be >>able to adopt their game to the opponent's or to "bluff" sometimes, they would >>become >>extremely strong. >> of all the things I have done, this probably describes it best. It can work. And it can lead to weaknesses as well. I get more GM feedback on this than any- thing else (remember the famous Roman comment, "enough with the aggressiveness, already, Bob!") I think this is more important at blitz, because there a bluff can hardly be overlooked. But at standard, I worry that "doing something, even if it is not necessarily really good" can backfire. time will tell. But it certainly backfires against computers more often than not.. >>All of the games were transmitted live over the ICC. The winning internet >>comment was from one of the observers who disagreed with us when we claimed >>that the dead sea is the lowest point on earth. He insisted that it is rather >>Washington D.C! >> >>From our point of view, as programmers, we felt that the match was very >>valuable. It exposed some obvious problems such as: Qb8, Bh8 and >>Bb7 in STC game 2 etc. but we felt that still the games were of high quality. >>Personally I liked game 3 of the rapid part where Yudasin tried a "sit and >>wait policy" and was positionally punished by Junior. I also >>can't stop admiring how Yudasin handled rapid game 4 where I'm >>sure that not many humans would of escaped Junior alive. >> >>A technical detail: >>Junior 5.0, soon to be released by Chessbase gmbh, ran on a Pentium >>333Mhz utilizing 80MB of RAM. Thanks to everybody who helped us >>have this event. Special thanks to: >> >>Matthias Wuellenweber of Chessbase, Rinat Zukerman of Hyatt, >>Guy Zadik of Packared Bell and Andy McFarland of ICC >> >>and of course to grandmaster Leonid Yudasin and to Amir Ban my partner. >> >>We hope you enjoyed the event. > >Yes, I appreciate the efforts that go into making such events a reality. >Thanks. Especially the standard 40/2 controls are interesting to follow. >There are so many games out there to play over that I for one just can't >find the time to play over the fast, non standard games. But the comp vs GM >or IM at 40/2 are rare and so spark greater interest in me. > > For your convenience the match >>games are enclosed. The next event on Junior's agenda will take >>place Tuesday, September 8th. IGM Dov Zifroni will play a two game >>match against Junior.
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.