ICCA Journal, Volume 20:  Number 3  (September 1997)




TABLE OF CONTENTS
Editorial:                                                                                          
    Early Retirement (H.J. van den Herik) ...................................................... 145
Contributions:                                                                                      
    Learning Piece Values Using Temporal Differences (D.F. Beal and M.C. Smith) ................ 147
    On the k-Best Mode in Computer Chess: Measuring the Similarity of Move Proposals                
        (I. Althoefer) ......................................................................... 152
Notes:                                                                                              
    How DarkThought Plays Chess (E.A. Heinz) ................................................... 166
    Genius-3 Cooked Endgame Studies (P. Wiereyn) ............................................... 177
Reviews:                                                                                            
    One Jump Ahead (D. Hartmann) ............................................................... 180
    Advances in Computer Chess 8 (D. Hartmann) ................................................. 183
Literature Received:                                                                                
    Man versus Machine: Kasparov versus Deep Blue (D. Goodman and R. Keene) .................... 186
    Kasparov versus Deeper Blue (D. King) ...................................................... 187
Reports:                                                                                            
    The Othello Match of the Year: Takeshi Murakami versus Logistello (M. Buro) ................ 189
    Report on the 12th AEGON Man-Machine Tournament (C. de Gorter and Y. Nagel) ................ 194
    Diep, Deep Trouble (P. Kouwenhoven) ........................................................ 200
    Board Games in Academia (A. de Voogt) ...................................................... 203
    Workshop Summary: Kasparov versus Big Blue (R. Morris) ..................................... 204
    A Report on the Fredkin Prize for Computer Chess (T.A. Marsland) ........................... 206
    Calendar of Computer-Games Events 1997 ..................................................... 207
    The ICCA Journal on Internet: A Follow-Up (P. Beck and A.E.M. van den Bosch) ............... 208
    Information on the 15th World Microcomputer-Chess Championship (T.A. Marsland) ............. 209
    The Swedish Rating List (T. Karlsson and G. Grottling) ..................................... 211




ABSTRACTS OF SCIENTIFIC ARTICLES


Learning Piece Values Using Temporal Differences
Don F. Beal and Martin C. Smith

[20(3):147-151]   This paper describes experiments where we attempt to learn the relative values of chess pieces by the use of temporal difference learning applied to minimax searches. We show that we are able to learn suitable piece values, and that these values perform at least as well as piece values widely quoted in elementary chess books.


On the k-Best Mode in Computer Chess: Measuring the Similarity of Move Proposals
Ingo Althöfer

[20(3):152-165]   Some commercial chess programs allow to compute not only the best move but also the k best moves for each position. Most frequently used are k = 2 and k = 3. Unfortunately, sometimes the moves are too similar in such a k-best list, or otherwise stated they do not discriminate among plans. A scheme examining the similarity and eliminating move proposals that are too similar is suggested. A simple approach dealing with this problem is presented.


How DarkThought Plays Chess
Ernst A. Heinz

[20(3):166-176]    DARKTHOUGHT is a bitboard-based chess program developed at the University of Karlsruhe that has successfully participated in all world championships since 1995. On a 500MHz DEC Alpha-21164a with 128MB RAM, DARKTHOUGHT routinely reaches speeds of 200K nps (nodes per second) in the middlegame while peaking at over 650K nps in the endgame.

The article describes the design and inner structure of DARKTHOUGHT. To this end, it presents detailed accounts of the chess engine while elaborating on some of its innovations: rotated bitboards, a fully programmable leaf-node evaluation function, and versatile search parameterization.


Genius-3 Cooked Endgame Studies
Paul Wiereyn

[20(3):177-179]   (Text still missing ...)




EDITORIAL


Early Retirement
H. Jaap van den Herik

[20(3):145-146]   Human geniuses are unpredictable. In the early 1960s Robert Fischer complained that Mikhail Botvinnik only played occasionally in tournaments, whereas a World Champion was supposed to show his supremacy as often as possible to the community. As soon as Fischer himself became World Champion he quit the scene, only to return some twenty years later to play the same opponent again. Maybe there was a good reason for Fischer to stop playing, maybe not. Many top sportsmen receive the proper advice of stopping with their activities as soon as they have reached their peak. A few do stop, but most of them continue nevertheless. They enjoy their sport, they feel at ease in the community, or they would like to raise their performances, showing that the last results were by no means their best.

Many fans believed that the ultimate IBM challenge was to achieve near perfection in chess, viz. playing a match against Kasparov for the title of World Champion. So far, only in the world of Checkers we have seen such a match: CHINOOK versus Tinsley. Indeed, that match had its own complications, and its success was tainted by Dr. Tinsley's forfeit.

The world of chess and computer chess have a common interest: both would like to see a third match between Kasparov and DEEP BLUE. However, IBM has decided otherwise: they followed the proper advice, and stopped. They have won a good reputation and can now sleep at ease. Although it is not our choice, we have to respect their decision and are still grateful for what they have offered to computer chess.

Meanwhile DEEP BLUE's victory over Kasparov has invoked many thoughts, opinions and proposals for future research. In three workshops (Princeton New Jersey, Providence Rhode Island, and Nagoya Japan) the question arose which game should take the prime research position over from chess? There were pros and cons for Go, Shogi and Chinese Chess. But still, Chess itself was also in the running. My opinion is that fundamental computer-chess research has not finished with the 1997 Kasparov- DEEP BLUE match. DEEP BLUE's early retirement makes the point more acute: it will take three to four years before another program reaches DEEP BLUE's current level. Moreover, we still do not understand DEEP BLUE's strategies behind its moves. This holds true, even for perfect moves selected from databases. Hence, there remains much scope for chess-oriented research.

Yes, the results achieved so far have enormous impact. Researchers in other areas will investigate how to apply some of the techniques, how to explain them, how to use them, and how to elaborate on them. For instance, DEEP BLUE's match result brought computer chess back into the minds of philosophers: on October 8, 1997 Daniel C. Dennett will give a lecture entitled: Can Machines Think? DEEP BLUE and Beyond.

Of course, Kasparov grudgingly received the bad news of DEEP BLUE's early retirement. ``It looks like a researcher who markets his finding, takes the profit, and then withdraws the product.'' Whatever the case, IBM has promised to have DEEP BLUE JR. around, playing a match now and then. Indeed a demonstration game has been scheduled for October 20, 1997 in Rotterdam at the beginning of the deciding match for the Dutch Championship Timman-Nikolic. Murray Campbell will act as operator.

Clearly, IBM is doing its utmost to preserve its reputation. In another reaction, to Kasparov's statement, C.J. Tan promised to publish the scientific results achieved by the DEEP BLUE team. Our readers will be glad to hear this, especially since the ICCA Journal welcomes any submission from their side. Whatever the future way brings us, your Editor believes that such early retirement should imply an intriguing return.



Created by Ernst A. Heinz and Heiner Marxen, Tue Aug 8 18:33:33 EDT 2000