Gentlemen, stop your engines!

For fifty years, computer chess has pursued an original goal of Artificial Intelligence, to produce a chess-engine to compete at the highest level. The goal has arguably been achieved, but that success has made it harder to answer questions about the relative playing strengths of man and machine. The proposal here is to approach such questions in a counter-intuitive way, handicapping or stopping-down chess engines so that they play less well. The intrinsic lack of man-machine games may be side-stepped by analysing existing games to place computer engines as accurately as possible on the FIDE ELO scale of human play. Move-sequences may also be assessed for likelihood if computer-assisted cheating is suspected.

Deeper model endgame analysis

A reference model of Fallible Endgame Play has been implemented and exercised with the chess-engine WILHELM. Past experiments have demonstrated the value of the model and the robustness of decisions based on it: experiments agree well with a Markov Model theory. Here, the reference model is exercised on the well-known endgame KBBKN.

Model endgame analysis

A reference model of Fallible Endgame Play has been implemented and exercised with the chess engine WILHELM. Various experiments have demonstrated the value of the model and the robustness of decisions based on it. Experimental results have also been compared with the theoretical predictions of a Markov model of the endgame and found to be in close agreement.

Self-play: statistical significance

Heinz recently completed a comprehensive experiment in self-play using the FRITZ chess engine to establish the ‘decreasing returns’ hypothesis with specific levels of statistical confidence. This note revisits the results and recalculates the confidence levels of this and other hypotheses. These appear to be better than Heinz’ initial analysis suggests.

KQQKQQ and the Kasparov-World Game

The 1999 Kasparov-World game for the first time enabled anyone to join a team playing against a World Chess Champion via the web. It included a surprise in the opening, complex middle-game strategy and a deep ending. As the game headed for its mysterious finale, the World Team re-quested a KQQKQQ endgame table and was provided with two by the authors. This paper describes their work, compares the methods used, examines the issues raised and summarises the concepts involved for the benefit of future workers in the endgame field. It also notes the contribution of this endgame to chess itself.

The Kasparov-World Match

The Kasparov-World match was initiated by Microsoft with sponsorship from the bank First USA. The concept was that Garry Kasparov as White would play the rest of the world on the Web: one ply would be played per day and the World Team was to vote for its move. The Kasparov-World game was a success from many points of view. It certainly gave thousands the feeling of facing the world’s best player across the board and did much for the future of the game. Described by Kasparov as “phenomenal ... the most complex in chess history”, it is probably a worthy ‘Greatest Game’ candidate. Computer technology has given chess a new mode of play and taken it to new heights: the experiment deserves to be repeated. We look forward to another game and experience of this quality although it will be difficult to surpass the event we have just enjoyed. We salute and thank all those who contributed - sponsors, moderator, coaches, unofficial analysts, organisers, technologists, voters and our new friends.

Strategies for Constrained Optimisation

The latest 6-man chess endgame results confirm that there are many deep forced mates beyond the 50-move rule. Players with potential wins near this limit naturally want to avoid a claim for a draw: optimal play to current metrics does not guarantee feasible wins or maximise the chances of winning against fallible opposition. A new metric and further strategies are defined which support players’ aspirations and improve their prospects of securing wins in the context of a k-move rule.

Discarding Like Pieces

The guideline of ‘discarding like men’ to estimate the merit of a chess position is well known. This note compares it with some statistics.

Self-play: statistical significance

Heinz recently completed a comprehensive experiment in self-play using the FRITZ chess engine to establish the ‘decreasing returns’ hypothesis with specific levels of statistical confidence. This note revisits the results and recalculates the confidence levels of this and other hypotheses. These appear to be better than Heinz’ initial analysis suggests.

Skilloscopy: Bayesian modeling of decision makers' skill

This paper proposes and demonstrates an approach, Skilloscopy, to the assessment of decision makers. In an increasingly sophisticated, connected and information-rich world, decision making is becoming both more important and more difficult. At the same time, modelling decision-making on computers is becoming more feasible and of interest, partly because the information-input to those decisions is increasingly on record. The aims of Skilloscopy are to rate and rank decision makers in a domain relative to each other: the aims do not include an analysis of why a decision is wrong or suboptimal, nor the modelling of the underlying cognitive process of making the decisions. In the proposed method a decision-maker is characterised by a probability distribution of their competence in choosing among quantifiable alternatives. This probability distribution is derived by classic Bayesian inference from a combination of prior belief and the evidence of the decisions. Thus, decision-makers’ skills may be better compared, rated and ranked. The proposed method is applied and evaluated in the gamedomain of Chess. A large set of games by players across a broad range of the World Chess Federation (FIDE) Elo ratings has been used to infer the distribution of players’ rating directly from the moves they play rather than from game outcomes. Demonstration applications address questions frequently asked by the Chess community regarding the stability of the Elo rating scale, the comparison of players of different eras and/or leagues, and controversial incidents possibly involving fraud. The method of Skilloscopy may be applied in any decision domain where the value of the decision-options can be quantified.

The data-mining of studies database 'HHdbIV'

This article clarifies what was done with the sub-7-man positions in data-mining Harold van der Heijden's 'HHdbIV' database of chess studies prior to its publication. It emphasises that only positions in the main lines of studies were examined and that the information about uniqueness of move was not incorporated in HHdbIV. There is some reflection on the separate technical and artistic dimensions of study evaluation.

Turing, Kasparov and the future

The 'Turing 100' Conference in Manchester was the main event of the Turing Centenary Year in 2012. This is a report and reflection on Kasparov's popular talk. Within it, he explained how Turing and influenced computer chess, his career and the chess community. Kasparov also played Chessbase's 'TURING' emulation of Turing's second paper chess engine, here labelled 'AT2'. Quasi Turing-tests, computer contributions to world championship chess, and suspected cheating in chess are also mentioned.

Tablebases and Tables

This item explains the origins of the words 'database' and 'tablebase' in the chess endgame context - and recommends the name 'Endgame Table' (EGT).

Qualified Statistics

This note caveats standard statistics which accompany chess endgame tables, EGTs. It refers to Nalimov's double-counting of pawnless positions with both Kings on a long diagonal, and to the inclusion of positions which are not reachable from the initial position.

Rook versus Bishop

The focus here is on the influence of the endgame KRPKBP on endgames featuring duels between rook and bishop. We take advantage of the range of endgame tablebases and tools now available to ratify and extend previous analyses of five examples, including the conclusion of the justly famous 1979 Rio Interzonal game, Timman-Velimirović. The tablebases show that they can help us understand the hidden depths of the chess endgame, that the path to the draw here is narrower than expected, that chess engines without tablebases still do not find all the wins, and that there are further surprises in store when more pawns are added.