January 25, 2005
A Theory of Fun Reviewed
A Review of A Theory of Fun for Game Design
written and illustrated by Raph Koster
Paraglyph Press
2004
244 pp.
$22.99
In an illustrated essay that is somewhere between a meditation and a manifesto, Raph Koster works to justify games to a general audience by characterizing them as learning experiences that can be tuned to challenge us in new ways. The book, based on a 2003 talk at the Austin Game Conference, is, unfortunately, short on real argument; Koster has thought out his positions in the book, but he usually neither backs up the claims he makes with much discussion nor follows through to investigate their implications. It’s interesting, though, that Koster has tried to make A Theory of Fun for Game Design itself a playful learning system, by juxtaposing text with diagrammatic or cartoon sorts of discussion, for example, and by providing copious endnotes with digressive comments and references. On the recto there are some gems: a nice chart showing the evolution of the 2-D shooter, drawings of game patterns for some of those shooters, and amusing cartoons in which teens brag about, among other things, beating the last level of Ulysses.
Despite the Ulysses reference, this book isn’t an academic book or thesis – and, unless you judge a book only by its title, you’d have trouble confusing it with one. Koster’s style and the frequent appearance of Alligator and Gratuitous Penguin don’t really suggest that you should shelve this alongside Wittgenstein’s Tractatus. As with the slides from his 2003 talk, A Theory of Fun doesn’t make the sort of contribution that Rules of Play, “Half-Real,” or “Videogames of the Oppressed” have made. As a primary, stand-alone introduction to game studies or game design, it is really only suitable for young readers. It does have some value to scholars of computer games, though. As mentioned above, the book’s style and format sketch a more ludic way to pursue “serious” exploration of video games, one that has some advantages. Also, the book serves to represent Koster’s own views on video game design, circa 2003-2004. Those studying Star Wars Galaxies: An Empire Divided (of which Koster was creative director) should certainly know about his perspective; it will also be of interest to general students of the MMORPG and those who deal with Koster’s other (and future) game design work.
Koster’s basic premise is that game play is puzzle solving, and that games teach us mastery. He explains at length that to do so, they should be neither too easy nor too hard, which probably does not count as a remarkable insight. But there is some more interesting discussion. Koster distinguishes particular types of challenges in existing games, such as the speed-based challenge to complete the level as rapidly as possible and the thoroughness-based challenge to get to all the secret areas. And, ultimately, he is interested in how challenges could be varied by designers to address different sorts of mastery, both to better appeal to those who are currently non-gamers and to help teach current gamers different sorts of things than they already are being taught.
Ah, yes – the terms at the core of the book, such as teaching and mastery, do seem more educational than ludic. This doesn’t trouble me; I’ve found the riddle, which is a sort of teacher that conveys a new perspective on the world, to be a good figure for a different sort of game. According to Koster, fun itself is “the act of mastering a problem mentally,” (p. 90) and is distinct from aesthetic appreciation, physical effort and mastery, and social actions. As Koster said here on Grand Text Auto, he does indeed fit a footrace into his theory, explaining in an endnote that there is some cognitive challenge there, but mainly using it as an example of what his theory isn’t about (the physical dimension of a game). His book’s take on games is relevant to the other examples I mentioned, too: A roulette-like game is also discussed – it’s treated as a repeated game that teaches us about probability in the long run. Rhythm/music games are cognitive challenges, too, in that they teach patterns. While this cognitive challenge may not be all there is to fun in the usual sense, and may not be the only or main sort of fun that we need to consider for all games, it is certainly an important part of computer game design, and it’s an aspect of games we can focus on and try to improve. Koster’s theory of Koster’s sort of fun is quite lively and interesting, a broad exploration that points to many implications of this formal and cognitive aspect of gaming. Koster goes on to discuss the juncture of the fun of games and the aesthetic (and fictional) qualities of them a bit, for instance, and to write about social and political issues ranging from game ratings to the demographics of gamers.
The abundant endnotes offer URLs and mentions of several useful books for follow-up reading, and will prove useful to some. They are a bit heavy on glossing terms whose use is fairly clear from context, but that bit of the book at least points out to the wider world of scholarship and writing. (There are a few mistakes in there, though, as might be expected: For instance, Exidy’s Death Race was not “the first instance of a game being adapted to a movie,” (p. 236). The 1976 game was “based,” pretty vaugely, on the 1975 movie Deathrace 2000.) While the cartoons carry much of the humor, I found plenty of little things to like in bits of the main text as well: Koster’s description of his lengthy bouts with Laser Blast – a game that has no obvious reason to fun, but which I also played obsessively – his nod to Floyd in Planetfall, for example.
But, I want to turn this review to the “dark side” for a moment, and discuss one particular paragraph I found problematic, because I think it captures of a few of the ways in which the book let me down as a reader – as an academic reader, anyway. You can blame me for misusing the book, if you like.
As preface, I’ll note that Koster usually offers tentative characterizations of the most important concepts he deals with – “fun” being an exception, and even that is not really defined until page 90. After skipping through a few other designers’ more detailed definitions of “game,” he writes that games are “puzzles to solve, just like everything else in life,” but he goes on to distinguish games from everything else by explaining that “the stakes are lower with games.” (p. 34)
Koster describes at the beginning of a paragraph on page 38 that we don’t think we can drive a car just because we know the rules of the road and what the controls do. (Of course, driving a car is not low-stakes, so it isn’t a game, but let’s continue.) Board games, on the other hand, “have fairly few variables, and so you can often extrapolate out from the known rule set.” He goes on to offer what he calls an important insight for game designers: “the more formally constructed your game is, the more limited it will be.” (Emphasis in original.) This is a claim, offered without real argument or even an explanation of what it means, that just seems immediately either wrong or meaningless. Chess is just about completely formalized, yet it has provided centuries of not-very-limited novelty and play. Salen and Zimmerman’s lucid and thoroughly argued concept of games is that their formal systems provide the only possibilities that are there in the game at all. And, what does it even mean to have one game be more formally constructed than another? Should such a game have more rules? That wouldn’t make it more formal, of course, that would just make it more intricate. To make a game less formal, we’d have to somehow have “informal rules.”
There’s also not much of a hint about what Koster means by “limited” by the middle of the paragraph, but the following sentences provides a clue about that term and also about “formally constructed”: “To make games more long-lasting, they need to integrate more variables (and less predictable ones) such as human psychology, physics, and so on. These are elements that arise from outside the game’s rules…” So a less limited game would be one that is longer-lasting, and less predictable. A few problems, though: (1) Creating such a game involves things that come from outside the game’s rules – these are exactly the things that the game designer doesn’t control. (2) Well, wait: actually, physics is “designed in” to computer sports games, and Jesper Juul, in his dissertation, makes a strong case that real-world physics should be considered part of the rule set of real-world sports, albeit an inflexible part. This means physics isn’t really outside the rules, it’s just something we chose to include by designing a sport instead of, for instance, some verbal game. (3) What does it mean to “incorporate human psychology?” How can any game played by a human not do this? How does a board game such as Risk not incorporate human psychology? Or is Koster referring to some other sort of board game, even though the rules of Risk are simple and one might be able to extrapolate Australia strategies and such from them?
Now, it’s possible that Koster’s quoted insight means something – it’s even possible that it’s truly insightful, and maybe his audience in Austin grokked that during his original talk. But the meaningful sense of it really isn’t clear in A Theory of Fun, and that characterizes many of the arguments (or the things that should be arguments) in the book.
For this reason, while I did find things to like about A Theory of Fun, it frustrated me as an academic reader. If Koster’s claims were really meaningless, I wouldn’t be so troubled. The frustrating thing is that I sense (from the book and from Koster’s discussion on here) that these claims do mean something. Perhaps the casual format and the (successful) attempt to make the book fun took their toll on the precision and clarity of the discussion. Or, maybe Koster went too far along the path of explaining gender differences and social reactions to game violence, important topics where he does investigate the implications of his ways of thinking about games. It just happens that some of the arguments, examples, explanations, and implications that would have made the book valuable to me as a scholar are, unfortunately, lacking.
That said, A Theory of Fun certainly can be read as a fun book (in my sense of fun). It’s definitely a better book than I could make out of one my talks. And, it’s a welcome contribution from an experienced MMPORPG designer and director, one which might play a different role in the industry and with general readers. As Will Wright says in the foreword, the book is an attempt to build bridges between academics and industry, and to probe how we can usefully talk about games. I’m certainly grateful for that. I’ll hope, though, that Alligator and Gratuitous Penguin will return for a sequel, and that Koster will be willing to take the thoroughness challenge next time.
January 25th, 2005 at 4:12 pm
I’m working on my own review right now, and agree that it’s difficult to get a handle on what he really means sometimes. Given the intended accessibility of the book, some explanation of what “formal abstract systems” and mechanics are would be nice. Are these just the coded rules? Do formal abstract systems always equate to mechanics?
I also have some problems with the theory of fun, but I will say this: I think his discussion on responsible game design is some fantastic stuff, and pretty much makes Theory of Fun required reading for just about anyone involved in game development, journalism and criticism. Not just for its content, but the way it’s written and presented.
January 28th, 2005 at 3:24 pm
Knowing Mr. Koster’s bent towards “sandbox” style games (like Ultima Online) rather than “theme park” style games (like World of Warcraft), my guess is the statement
the more formally constructed your game is, the more limited it will be.
refers to formal structure on the “activities” of the players — the specific hoops that need to be jumped through to make “progress” or at least cause something new to occur.
This doesn’t refer to rules — in chess you have pieces restricted by specific rules and a specific goal, but the method to get to checkmate is wide-open and not predetermined by structure.
I also get the impression he thinks limiting is a bad thing, and wants to disparage the World of Warcraft method of handing off quests on a silver platter. On the other hand, the players seem to *like* this more direct approach.
Part of the perception here is that because there is inflexibility in *direction of approach*, the structure itself is inflexible. This ignores the possibility of flexibility of *technique*.
Let me shift to second person here and take the Thief and Splinter Cell series as examples.
In Thief, you are not a fighter. You are limited in some areas in that you *must* be sneaky. However, you generally get many different methods of direction, of approach, so perhaps you are sneaky by entering the second story window or perhaps you are sneaky by dodging the guards up front.
In Splinter Cell, you can sneak or fight. You have more flexibility of technique than in Thief. The world structure itself, however, is highly linear, and you aren’t going to be able to vary which encounter you come across next.
I have heard people disparage Thief and praise Splinter Cell because they focused on technique. I have heard people disparage Splinter Cell and praise Thief because of their approaches to linearity.
I don’t suppose it’s too controversial to suggest that both methods of allowing player freedom are useful, even if particular personality types seem to prefer one over another?
January 30th, 2005 at 6:32 pm
There are a number of good reviews and ongoing
January 30th, 2005 at 10:27 pm
The statement has zero to do with World of Warcraft, trust me. :P Nor was it referencing sandbox versus themepark style worlds. It wasn’t about online games at all.
Rather, Nick’s on the right track when he cites Jesper’s treatment of physics. Many games do NOT design physics into their ruleset. (In fact, most sports games will design in only a fraction of actual physics). You can have also games that simulate a broad array of physical activities without simulating physics.
Physics has a lot of rules. We might like to conveniently label all of them under one word, but the fact is that physics is complex, really. It also permits all sorts of interesting behaviors, if fully implemented. Were sports games to REALLY include phsyics, you could set fire to other players. But they don’t, and you can’t.
We can think of physics as being an “included ruleset” similar to an included file in programming. (That is exactly how commercial physics engines work in games, after all). They provide a wide array of behaviors in a “blackbox” fashion–we only know the API, the interface to reach the behaviors. I consider them “outside the game rules” because they are never explicitly codified.
From a game design point of view, a game which makes actual use of physics is “including” a vast array of rules into its game without explicitly defining them. There is an assumption that a common frame of reference is shared, and that therefore everyone will know what’s up.
Physics is just one example of the sort of “imported ruleset” that games frequently use. All head-to-head games import human psychology as an additional ruleset defining how opponents make decisions. If you can communicate with the other player in any way (such as by seeing their face), a whole host of new and poorly defined game rules come into play. These generally complexify the game experience, without really complicating it since we are used to dealing with things like facial expressions.
An example of this at work would be bluffing in poker. The difference between a bluff done by a computer based on an algorithm and a bluff done by a human you cannot see, versus a bluff done by a human you cannot see and a human you can, is quite impressive. Poker played against humans (versus poker played against a computer) is a far richer game.
In the book I reference game design atoms briefly. In the list of core elements of a game mechanic I cite “a challenge to overcome.” Formally constructed games tend to have defined all the challenges in advance. Less formally constructed games have more variables. Games which import rulesets which are full of ambiguity have a lot of highly unpredictable variables.
The easiest way to do this is to make another person the source of the challenges. The way to limit the ambiguity is to formally limit the possible challenges they can offer; to expand it, reduce the formal strictures on the challenges presented.
Hence the statements in the book.
I am beginning to think that I either needed to write a simpler book or a considerably more in-depth one. :) Certainly most of the reviewers seem to be asking for more in-depth.
February 5th, 2005 at 12:26 am
Slashdot, lagging behind us as usual, has a review.
February 7th, 2005 at 1:07 am
Really? I’ve heard a bit about adaptive algorithms in poker playing recently which make some interesting variations, while I’ve heard a lot of coworkers discussing online poker as basically a way to make money statistically. I don’t play poker myself, but from what I’ve heard it sounded like the two sides’ playing styles could blur when the algorithms are interesting enough, and the players are strong enough at their math.
Anyway, that wasn’t your point. And maybe you were thinking of no-limit poker. ;)
October 24th, 2005 at 2:25 am
[…] to pop up all over the place. An interesting one is Nick Montfort’s review over at Grand Text Auto. Generally speaking, I think arguing with rev […]
November 1st, 2005 at 2:32 pm
[…] uage yet again). This very much echoes a similar discussion on Grand Text Auto here and here (and my response to their review is here), and discussions I have had w […]
March 4th, 2006 at 11:31 pm
My two cents… I used this book as a starter text in my “Video Game Culture and Theory” course, and followed it up with Brenda Laurel’s Utopian Entrepreneur and finsihed with Jesper Juul’s Half Real. It was a lower-level English class designed without any prerequisites. I filled in with a few other stand-alone articles, and had students write game reviews, a “close playing” and a researched term paper.
I knew I had to start at a pretty basic level. I started out asking them to look at War Games and Tron and a few other games-related movies, and I had them watch/play the famous Strongbad on classic video games. We also had a unit on new games journalism as a bridge between game reviews (the only kind of writing about games I expected the students to be familiar with) and games criticism. The “culture” came before the “theory” in this class. It was a three-week January course. A few students dropped the course when they saw the workload, and the ones who stuck it out said they would have enjoyed it much better if it were a regular full-term class. (I would have been able to assign more reading then, and also perhaps spent some time as a class in EverQuest.)
In retrospect, I think I should have had them buy an Atari Classics CD as a required text, since some of the students Googled for the titles of classic games, ended up finding and playing Flash or Java emulations that weren’t terribly faithful to the original, and played them without a sense of what the emulations were missing.
I do teach some upper-level new media courses. I would definitely use Koster’s book again for the lower-level course, but I would problably just put the library’s copy on reserve if I were to use it again for an upper-level course. It’s a good introduction, but it’s not written for an academic audience.