In lieu of an abstract, here is a brief excerpt of the content:

  • Knowledge Discovery in Chess Using an Aesthetics Approach
  • Azlan Iqbal (bio)


Computational aesthetics is a relatively new subfield of artificial intelligence (AI). It includes research that enables computers to "recognize" (and evaluate) beauty in various domains such as visual art, music, and games. Aside from the benefit this gives to humans in terms of creating and appreciating art in these domains, there are perhaps also philosophical implications about the nature and "mechanics" of aesthetic perception in humans. We can, potentially, learn more about ourselves as we replicate or simulate this ability in machines. My original intention or objective was to develop an aesthetics model or method for computers to recognize beauty in the game of chess since it is greatly appreciated by human players1—there is even an entire subdomain devoted to it in the form of chess compositions2—and because it has other promising applications, such as improving computer chess heuristics, for example, in solving complex chess problems.

My intention was also that this model should correlate positively and reasonably well with human player aesthetic assessment. Both were achieved with better experimentally validated success than previous approaches.3 A computer program called CHESTHETICA (see appendix) was developed to aid in the complex calculations involved. It can identify beautiful combinations4 in large databases of chess games based on the aesthetics model.5 [End Page 73] In the course of that research, new discoveries about the game were made by the program and noted. A discussion on the philosophical basis of computational aesthetic recognition in chess, specifically chess themes,6 is presented in the following section. Each discovery along with an explanation about the aesthetics approach that apparently led to it is presented in the two sections after that. Each is also contrasted against what we currently know to demonstrate its novelty. This is followed by a discussion about the relevance of aesthetics in extracting new knowledge. The article concludes with a summary of key points and a note on further work.

Computer Recognition of Beauty in Chess

Prior to performing the experiments that were intended to gauge beauty in chess, I derived several principles of aesthetics pertaining to the game from the relevant literature. Notable sources include work by the psychologist Stuart Margulies, who derived eight "principles of beauty" from the judgment of thirty expert players;7 the four "elements of beauty" as described by chess grandmaster Jonathan Levitt and international master of chess problem solving David Friedgood;8 the eight brilliancy characteristics described by Damsky that are typically used to award prizes to the most beautiful games in certain tournaments;9 and twenty-one chess composition conventions.10 It is notable that in all these references, the term "beauty" was used to refer to the things described. All this information was distilled into a list of eight aesthetic principles, which includes the use of chess themes.11

An important question might therefore be, how do we know that what is being gauged computationally in those themes is actually beauty and not some meaningless pattern?12 There are at least three arguments that can be presented in favor of the former. First, one could say that a theme, in and of itself, having been substantiated in chess literature as a constituent of beauty in the game recognized by players and chess problem composers, should be considered aesthetic regardless of how it is assessed, even if this means attributing a somewhat arbitrary fixed value to all possible instances of it; this is not an unaccepted or even uncommon approach in computational aesthetics.13 The idea here is simply that, given a reasonable level of competence in the game, chess themes, in general, are likely to be appreciated aesthetically by most players and composers. Combinations that lack themes are therefore less likely to be considered beautiful.

Second, as was the case in my own research, computationally detected themes were actually assessed dynamically based on their "effectiveness"—also substantiated in chess literature as being a prerequisite or relevant to beauty14—and "aesthetic features" or properties within the theme itself. "Effectiveness" means that there must be some form of achievement as a result of executing the theme; for example...


Additional Information

Print ISSN
pp. 73-90
Launched on MUSE
Open Access
Back To Top

This website uses cookies to ensure you get the best experience on our website. Without cookies your experience may not be seamless.