With a title such as A Theory of Fun for Game Design, one cannot help but suspect strange tensions at work in Raph Koster's book. 'Fun' and 'theory' sit together somewhat uncomfortably, each shuffling their feet as if uncertain they have come to the right place. And yet, after relaxing, perhaps sharing a few drinks and anecdotes, suddenly it seems that the two are getting along famously - although they never quite manage to speak quite the same language. Such is the schizophrenic joy of A Theory of Fun - it combines a methodical approach with a playful freedom from excessive seriousness resulting in a book unlike any other on the subject of games.
The first thing one notices about the book is it's unsusual structure, as it has text on the verso pages, and (mostly) cartoons on the recto pages. For the most part, the annotated cartoons provide a superlight version of the main content of the book, while generously complete notes (collected as an appendix) provide more detailed annotations where appropriate. The result is a book that can be read in three ways - focussing on the cartoons and using the text for more details, focussing on the text and using the cartoons for illustration, and using the text and the notes in concert for a more scholarly approach.
Koster's view is that fun is another word for learning. However, in order to support this view, the author becomes forced into excluding any aspect of the word 'fun' that does not fit with his model. Koster is completely upfront about this - he suggests that taking a wider view of fun "renders the word meaningless", and therefore focuses only on a specific definition of fun as "the act of mastering a problem mentally". Since this is not how the rest of the world uses the word 'fun', I personally feel he would have done better to coin a different phrase rather than co-opting fun into a context that does not fit with how the word is conventionally used (this betrays my adherence to Wittgenstein's thinking on language yet again).
Still, since this linguistic sleight of hand is performed honestly and openly, there is not really a problem. However, I'm going to use the term Kfun (for 'Koster fun') to refer to the subset of fun discussed in the book in an effort to disambiguate the language.
The author explores kfun in a thorough and detailed fashion, looking at what games are good at teaching, differences in individual ability (solely through the perspective of intelligence and learning styles because of the focus on kfun), problems that result, and coming round ultimately to looking at issues of representation, ethics and ultimately the cultural role games may aspire towards. Because of the brisk pace, some issues are not examined in much detail, but conversely the book covers a vast multitude of topics making it an excellent primer on the subject of games.
Again, because of the volume of material covered, there are inevitable errors and ommisions. These vary from the trivial to the debatable.
An example of the former (trivial errors): Koster claims that Centipede is not really that different from Galaxian. But the attack waves in Galaxian are static, and nothing the player does affects them, whereas in Centipede, shooting the Centipede leaves mushrooms which alter the attack patterns in a dynamic and controllable fashion, resulting in a much harder game to master. (The obvious rebuttal to this sort of minutiae is "who gives a damn", of course).
An example of the latter (debatable errors): Koster claims "From a strict evolutionary point of view, cheating is a winning strategy." There is no reference to this in the notes, so I assume this is just his opionion - I believe he is in error. Although ecologies can support a certain amount of 'cheating', this does not equate to cheating being a winning strategy. Cuckoos lay eggs in other bird's nests (a form of cheating) - but this does not allow cuckoos to out perform other bird species - if there were only cuckoos, they would be royally screwed as brood parasites must inevitably be in the minority. Furthermore, it is widely accepted that in a social context the most successful categories are Tit for Tat and Tit for Two Tat - which outperform every cheating strategy. Koster's claim that "Duelists who shoot first while their opponents are pacing off are far more likely to reproduce" is flawed: four people attend a formal duel, the two duelists plus two seconds. It is hard to devise a cheating strategy for a formal duel that will allow the cheat to reintegrate into their society and breed.
There are also points in which I believe he misrepresents other people's theories. For instance, Koster equates Lazarro's Hard Fun with Csikszentmihalyi's flow in an inaccurate manner. I believe Koster might have been at the same Lazarro GDC presentation as I was - in which she inadvertently mentioned flow and provided an impromptu explanation while talking about Hard Fun. But this was not because flow refers only to Hard Fun. However, since Lazarro still hasn't written a book (I hope she will address this soon!) we only have her paper on the subject to refer to. Besides, Koster's conclusion that fun isn't flow is broadly accurate and his conclusion that flow isn't kfun is completely accurate, so in some respects it is not a problem.
However, these sort of issues are the result of the wide reach of the book which covers so much it is inevitable that a few imperfections creep in. These small and often insignificant cracks are more than compensated for by both a wealth of useful information, and super cute alligator cartoons.
The book does not really provide the theory of fun that some may feel is promised in the title - but then, I don't believe it is a coincidence that Koster chose to call it 'A Theory of Fun' and not 'The Theory of Fun'. Koster presents a theory of kfun, and he explores it thoroughly. It leads him to some conclusions that I find hard to agree with, because in accepting the value of the fun of learning (kfun) he rules out to some extent having fun just for the sake of it which I believe may be premature.
None of this actually matters, because at it's heart this isn't a book about a theory of kfun at all (although this is the central device by which it progresses). A Theory of Fun for Game Design is a highly personal and empassioned defence of the value of games and of the games industry. The book is framed in the context of a discussion between Koster and his grandfather, who asked if Raph was proud of what he did, and throughout the book the presence of Koster's children can also be felt. There is an intimacy to this aspect of the book which I feel is its greatest strength.
I believe there are two basic battles going on right now - one is for the defence of the games industry against those who would muzzle it or inhibit its growth (and this is where Koster ebulliently leaps to games' defence) and one is for the industry to expand its reach to meet the needs of a wider audience (which is where Koster's co-worker Sheri Grainer Ray has staunchly positioned herself). I personally feel more driven to fight on the second front, but I have great respect for those who have chosen to take to the field of the first.
Erudite, passionate and accessible, A Theory of Fun for Game Design is a unique and engaging book which suffers from a certain schizophrenia as to whether it is talking to the layman or the professional but somehow manages to make this confusion work to its advantage. It is essential reading for anyone interested in games, and an ardent defence of the value of games which is both apposite and timely.
A Theory of Fun for Game Design is published by Paraglyph Press, ISBN 1-932111-97-2.