ICCA Journal, Volume 17:  Number 2  (June 1994)




TABLE OF CONTENTS
Editorial:                                                                                          
    Twinkle Chess (I.S. Herschberg and H.J. van den Herik) .....................................  49
Contributions:                                                                                      
    Self-Annotating Elementary Endgames (R. Seidel) ............................................  51
    Proof-Number Search and Transpositions (M. Schijf, L.V. Allis, and J.W.H.M. Uiterwijk) .....  63
Notes:                                                                                              
    Weak Zugzwang: Statistics on some Chess Endgames (I. Althoefer and B. Walter) ..............  75
    More, and More Perfect Prose (J. Nunn) .....................................................  78
Reviews:                                                                                            
    Nunn: Secrets of Pawnless Endings (I.S. Herschberg and H.J. van den Herik) .................  81
    Mysliwietz: Konstruktion und Optimierung von Bewertungsfunktionen beim Schach (I. Althoefer)  82
    Perrey: Einige stochastische Modelle fuer Zwei-Personen-Spiele (I. Althoefer) ..............  83
    Osterkamp and Schanz: Strategien fuer Systeme von zufaellig entscheidenden Beratern             
        und die Kompetenz von "Dreihirn" (I. Althoefer) ........................................  84
Reports:                                                                                            
    The 9th AEGON Man-Machine Tournament .......................................................  86
        Report on the Tournament (B.M. Verbaan) ................................................  86
        Results and Selected Games (C. de Gorter and B.M. Verbaan) .............................  89
    The 1st Dutch Rapid Computer-Chess Championship (J. Louwman) ...............................  96
    The INTEL World Chess Express Challenge (F. Friedel with F. Morsch) ........................  98
    The ICCA Journal Award (The Board of ICCA) ................................................. 105
        Peter Jansen: A Scientific Biography ................................................... 105
    The Best-Annotation Award (The Board of ICCA) .............................................. 106
    The Swedish Rating List (T. Karlsson and G. Grottling) ..................................... 109
    Calendar of Computer-Games Events 1994 ..................................................... 110
    The 4th International Paderborn Computer-Chess Championship ................................ 110
Correspondence:                                                                                     
    Straight on to Kasparov (D. Levy) .......................................................... 111




ABSTRACTS OF SCIENTIFIC ARTICLES


Self-Annotating Elementary Endgames
Rainer Seidel

[17(2):51-62]   A study on elementary endgames of the lone-king type is based on a theory that satisfies formal requirements as well as being intuitively meaningful. Two endgame domains, KRK and a significant component of KBBK, have been fully worked out and constitute operational programs. Owing to their adequate basis in theory, the programs are self-annotating to the extent that even a tutorial comment can be produced for any starting position in the domain and for any subsequent enemy response. The hierarchy of its chess knowledge is sketched syntactically and semantically; a generalized procedure for move generation is exhibited, as is the principal schematic of the KBBK program. It is suggested that the knowledge so gained may be extensible to the analyses of higher-order endgames.


Proof-Number Search and Transpositions
Martin Schijf, L. Victor Allis and Jos W.H.M. Uiterwijk

[17(2):63-74]   Proof-number search (pn-search) has been successfully applied to games such as connect-four, qubic and go-moku. Although pn-search is a game-tree search algorithm, minor modifications allow its application to the acyclic-graph representation of these games. To apply pn-search to cyclic-graph representations, e.g., of chess, several obstacles must be overcome, which have been the subject of our investigations. This article describes the application of pn-search to trees, acyclic graphs and cyclic graphs. Experiments on chess, using four different pn-search variants, indicate the suitability of pn-search for solving checkmate problems. The results show that pn-search adapted for cyclic-graph representations outperforms its tree and acyclic-graph counterparts.


Weak Zugzwang: Statistics on some Chess Endgames
Ingo Althöfer and Bernhard Walter

[17(2):75-77]   Recently, the concept of Zugzwang has gained some attention in computer-chess research (Roycroft, 1990; Jansen, 1992a, 1992b). In this note, we present some statistics on weak Zugzwang in the four chess endgames KQK, KRK, KBBK and KBNK. The traditional definition of Zugzwang is based on the won-drawn-lost value of a position on the board. Zugzwang occurs when that value for the side-to-move is less favourable than it would have been if his opponent were to move. There are exactly three different categories (Roycroft, 1990).

Traditional Zugzwang requires a drastic change in value, such as a winning position being a draw for player A when A is forced to move first. Less drastically, we may define weak Zugzwang if the distance-to-mate for player A is larger when player A is forced to move first. Note that this definition makes sense even if the game is always a win for player A. We give a simple example from the endgame KRK, namely Kb6, Rc6 for White and Kb8 for Black. If Black is to move, he will lose in one move (0. ... Ka8 1. Rc8 mate). If White is to move, it will take two moves to win (for instance 1. Rc7 Ka8 2. Rc8 mate). For every position P on the board which is a win for White independently of who plays first we define a number k(P): ``k(P) := (distance-to-mate if Black is to move) minus (distance-to-mate if White is to move)'' Then P is a position with weak Zugzwang if k(P)is negative. In our example, we have k(P) = -1.


More, and More Perfect Prose
John Nunn

[17(2):78-80]   (Text still missing ...)




EDITORIAL


Twinkle Chess
I. Samuel Herschberg and H. Jaap van den Herik

[17(2):49-50]   Before Science became scientific, there was a simple rule: ``Nature abhors a vacuum''. This, it was justly remarked later, explained nothing and one had to wait for scientific Science until it became clear, by much more involved reasoning, why nature abhorred a vacuum only to the tune of a column of ten meters of water or three feet of mercury.

In the same spirit, now that we all have gone scientific, there is not only a definite interest in the phenomena, in what happens in the world, say the world of computer chess; at least as much interest attaches now not just to what is observed, but to why it should be observed. For anything newly discovered, a spate of theoreticians rush in, doing away with such simple things as the horror of a vacuum, explaining it. Explaining it? Yes, if one prefers involved arguments with Science to unscientific statements of bald facts.

As is abundantly clear from this issue of the Journal, the spectacular success of FRITZ 3 at speed chess is one such bald fact of life, of which Grandmasters may well feel a horror. Explanations, though, are far to seek and the Grandmasters, soundly trounced, rush most unscientifically into excuses. Even the designer, Frans Morsch, concurs: it is really beyond those Grandmasters to click a mouse smartly. Not so, we maintain.

Other explanations offered for the program's success share the weakness of being in the nature of excuses. The losers argue that the computer is relentless, persistent and graceless, ever ready to pounce, inexorably punishing the slightest oversight.

There is no denying that some truth may be in what the losers adduce, but your Editors believe that the victims in speed chess have blamed the wrong agent. With unwonted seriousness, this Editorial will argue that human beings, even the élite of them known as chess Grandmasters, have wilfully placed themselves at a disadvantage by consenting to play speed chess at all.

It all boils down to the answer to a double-barrelled question: what can one do in the twinkling of an eye? The answer to that query is, when one is a human being: very little. What can a moderately slow computer, such as the Pentium, do in the twinkling of an eye? The answer is: validly evaluating about 50,000 nodes. And here lies all the difference.

Theories of human visual perception are not conspicuous for the consensus they have reached in explaining what goes on when we humans use our eyes. Yet, they all are agreed that it takes roughly half a second before we truly perceive what we see. This is the experimental latency which we express poetically by `the twinkling of an eye'. The computer has no such delay, no latency gaping between perception and action.

Now consider a speed game at five minutes for 40 moves (i.e., at 7.5 seconds a move) and the much-vaunted hardware which will allow, in 1994, 750,000 nodes to be searched in that interval. As compared with 1989, only five short years ago, this is what DEEP THOUGHT searched for a move and DEEP THOUGHT, even then, we all know, was a formidable opponent. Next, let us enquire into a human being with considerable visual latency pitted against the then DEEP THOUGHT. Due to his unavoidable latency (and the psychological stress it no doubt induces) we human beings are now at a disadvantage. What is more, it turns out that we voluntarily increase our handicap as we reduce the time allotted to a move, a reduction which seems imminent in next year's version of the Dutch Rapid Computer-Chess Championship.

The engines going for ever-more-rapid chess need fear their opponents less and less for the simple reason that, the faster the moves, the greater the deleterious impact of the latency on the human being. We are not surprised, therefore, that Kasparov, who at six minutes to the computer's five had crushed his opponent when he put his mind to it, promptly lost when, in a public show, he saw his own time allotment reduced to four minutes. He who delivers himself to the computer by imposing high requirements on his own visual agility must not be surprised if the computer mercilessly devours the champion who has foolishly surrendered his edge.

And while we are on this subject of speed, where, in the extreme, the human being is without a sporting chance against a program, let us try to agree on a terminology. In three months of consulting the chess literature on which we ultimately depend, we have encountered too many terms for the faster-than-normal execution of moves, none of them satisfactory. Rapid chess? What rapids are you shooting? Speed chess? Does anyone remember the instruments once known as high-speed printers, if you please, not just speed printers, churning away at two lines a second? Speed chess, again? Speed is a drug many people unfortunately have become addicted to. Blitz games? Some of us, old enough to be aware of the Second World War, find that ``blitz'' has most unfortunate connotations: after the Blitzkrieg miscarried, London was blitzed. Seeing the importance of the time we spend in twinkling our eyes, may we suggest ``twinkle chess''?



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