James Wallis levels with you

Caillois completeness

One of the things about the world of computers that won’t go away is that there are two kinds of computers, and the kind you like sucks. Apparently. Whether it was big iron versus terminals, or minis versus micros, or Spectrums versus Commodore 64s, or the Atari ST versus the Amiga, or PCs versus Macs, or machines that run your particular favoured brew of Linux and everything else, this is a debate that will never end. It will only grow more tedious.

The thing is, there really is an important line that divides computers, or at last computing machines, into two types. There are machines that are Turing-complete, and there are machines that are Turing-incomplete. A Turing-complete machine is one that given enough time can perform the same tasks as any other computing machine, and in the early development of computers this concept was a big deal.

You’ll be glad to hear that’s pretty much all I have to say on the subject of Turing-completeness: if you want to know more, let the Wikipedia entry confuse you. (If you have to ask who Turing was then you’re reading the wrong blog, though I can brighten your day with the trivia snippet that the great man was on intimate terms with my former public-school housemaster, which implies that Turing was either desperate or had appalling taste.)

I’m not here to talk about the finer points of Turing completeness. But the concept provides a useful benchmark for all of computing: does a new principle or new design live up to a certain, almost abstract gold standard? And that set me wondering: does a similar standard exist for games? Should it? Would this be helpful? Where would you start?

And is this important? Yes, yes it is, and particularly right now. We’re seeing a lot of ideas that originated in games appearing in the wider media, like achievement-point systems. These aren’t games, they’re game-like activities. They push many of the same pleasure-centres in the brain that games do but they’re not games, not in the classical sense. Is Mafia Wars (Zynga, more than 70 million players at last count) a game? You can’t win, you can’t lose, you just allocate some resources and spend some money each turn and watch the numbers go up, and feel the tiny dopamine hits. Fun, arguably. A game? Well….

Let’s go back to the beginning. Not the beginning of games, of course—in human development the invention of board-games pre-dates written language. And not the beginning of video games: that’d be like trying to create a standard for all music based on the work of Michael Jackson.

No, I’m talking about the beginning of games criticism. I’m talking about Roger Caillois. Roger Caillois (1913–1978), French philosopher and writer. His 1958 work Les Jeux et Les Hommes, known in English as Man, Play and Games, is probably the first serious examination of games qua games. Sure, Johann Huizinga’s Homo Ludens was more than two decades earlier but Huizinga was a sociologist and Homo Ludens is more about the phenomenon of play: why we play, not what or how we play. Caillois was, in a gloriously French way, a freelance intellectual, a free-thinker who hung out with the likes of George Bataille—another link between games-design and the Surrealists and Dadaists—and Borges. In short: Huizinga, fuddy-duddy university professor. Caillois, one of us.

Once you get past the faux-sepia and oddly homoerotic cover of the current English-language edition, Man, Play and Games is a helter-skelter ride through games and play across history, culture and species, stopping to examine fighting wrens, superstitious Parisians, children breaking things for fun, viewing the face of a god, and ants taking drugs. It is awesome; worth the price of admission just for the Chinese word wan, which means ‘the act of indefinitely caressing a piece of jade while polishing it in order to savour its smoothness’.

Caillois’s book does a number of interesting things. Primarily he breaks the make-up of games down into four primary constituent parts, like the Greek elements, and as a nod he gave three of them Greek names. All games, he said, are composed of these four parts in different proportions:

  • Competition (agon)
  • Chance (alea)
  • Mimicry (mimicry), role-play or let’s pretend
  • Vertigo (ilinx), the sense of losing yourself in immersion, to provoke a controlled amount of a normally scary emotion—what Caillois calls ‘voluptuous panic’. If you’re shaking your head, two words for you: zombie games.

If that sound a bit primitive, remember two important things. First of all, Caillois was breaking new ground. Nobody had done this kind of academic analysis of games before, so he was starting from first principles. Secondly, this was 1958. The very first video game may have existed on an oscilloscope screen in a laboratory in 1947, but it would be another thirteen years before anyone thought about trying to create the things commercially.

It’s not the agon/alea stuff that I’m talking about here. That stuff’s great, I teach it to my students, but it’s the first chapter of Man, Play and Games I want to bring to your attention. Caillois actually set out a list of six criteria for what a game had to be, do or contain, in order to be considered a game. Fifty years on it’s aged remarkably well. And while it lacks the conciseness of Turing’s definition of whether a computer is Turing-complete or not, it does a significant job of punching a fence across the territory and saying, ‘This. Here. Everything on this side is ours.’

So I propose a new standard for games or things that call themselves games. It’s not about quality or playability or the standard of their graphics. If there’s something that bugs you about a new design or principle or way of doing things then ask yourself: is this game Caillois-complete? Because if it’s not, the chances are that it’s not actually a game. It may be a software toy, not that there’s anything wrong with that. It may be an interactive exercise dressed up in the tropes of games. But just as a computing device that isn’t Turing-complete isn’t what we would understand as a computer, a game that isn’t Caillois-complete is lacking some part of the essential DNA that makes it a true game.

That doesn’t mean it’s bad, or it’s broken, or we should turn our noses up at it. It’s just Caillois-incomplete. And when we think about games, that’s a useful critical tool to have. What does Caillois say a game should be? He gives six points, which I’ve retyped here with his short descriptions of each one. (This is the last two pages of the first chapter of Man, Play and Games, pages 9-10 in the University of Illinois edition—one hell of a way to start a book): A Caillois-complete game is one that’s:

  • Free: ‘in which playing is not obligatory; if it were, it would at once lose its attractive and joyous quality as diversion.â’
  • Separate: ‘circumscribed within limits of space and time, defined and fixed in advance.’
  • Uncertain: ‘the course of which cannot be determined, nor the result attained beforehand; and some latitude for innovations being left to the player’s initiative.’
  • Unproductive: ‘creating neither goods, nor weath, nor new elements of any kind; and,except for the exchange of property among the players, ending in a situation identical to that prevailing at the beginning of the game.’
  • Governed by Rules: ‘under conventions that suspend ordinary laws and for the moment establish new legislation, which alone counts.’
  • Make-Believe: ‘accompanied by a special awareness of a second reality, or of a free unreality, as against real life.’

I’ll break these down and talk about what each of them mean in the twenty-first century in my next post.

(And this is a link to somewhere you can buy Man, Play and Games. Not an Amazon link because, as many people have described over the last couple of weeks, Amazon right now is being a playground bully. Try the Book Depository instead. Excellent prices, free shipping anywhere, not a playground bully, and a refreshing focus on, you know, books.)

Categorised as: game design


  1. Carlton says:

    Thanks – it’s a nice piece about Caillois. Surprisingly few game designers seems to recognise the contribution he’s made to the field.

    He has lots to make us think. The characteristic that I believe many will find most challenging is the idea that games are wholly unproductive either materially or presumably, cognitively. That’s especially difficult for those of us interested in game-based learning.

    Still, like you, I’m continually amazed at the number of activities described as ‘games’. Caillois’s definition will make many of those producers who hope to exploit the cache of ‘games’ uncomfortable!

  2. Ian Sturrock says:

    I’ve argued for a while now that most MMORPGs aren’t, in fact, games, though I’ve not actually read Caillois. Certainly they’re not Caillois-complete.

  3. John Scott Tynes says:

    Ian, that’s a surprising assertion. Can you elaborate?

  4. Ian Sturrock says:

    John, OK:

    In reference to WoW, Warhammer Online, and Guild Wars, which are the only ones I really know —

    1) “Free” is iffy. As well as actually “playing” the games, there is a whole farming side, which is very nearly obligatory if one is to enjoy the actual gameplay, but which in and of itself doesn’t seem either enjoyable, or gameplay.

    2) They’re absolutely not circumscribed in time. You can log on and play at any hour of the day or night, and there will always be something to do, though how gamelike that something is, is anyone’s guess.

    3) A combination of walkthroughs, and more experienced players leading parties in-game, mean that for the vast majority of players, there’s no uncertainty, and little room for innovation. Admittedly, if you’re a trailblazer at a new MMORPG, this is less true, and you could deliberately avoid walkthroughs & assistance for a more gamelike experience — but neither of those applies to the vast majority of players.

    4) It’s not unproductive. You’ll be disappointed if you end your MMORPG session with your character’s stats & possessions unchanged, and it DOES produce new stuff.

    5) and 6) These ones do apply to MMORPGs, though of course they’re insufficient to make MMORPGs games, in and of themselves.

    Some of the same points apply to tabletop RPGs, but only to some tabletop RPGs — a minority, I would say. Most RPGers play either extended campaigns, or one-shots, rather than campaigns that last potentially their whole lives. The impossibility of permanent character death in MMORPGs, as compared to its likelihood in tabletop RPGs, is also significant — arguably, when your character dies, you’ve ended one game.

    • admin says:

      Ian, I think you’re reading more into Caillois’s brief descriptions than is actually there. I’ll go into this in more depth in future posts (when life gives me a moment) but while most MMOs fail ‘circumscribed in space and time’, they get a pretty clean pass otherwise. Gold-farmers get a special exception rule for Unproductive, though.

  5. Jody Macgregor says:

    I guess the definition of ‘free’ means that any game being played for review becomes Caillois-incomplete during that time. If I have to play Final Fantasy Crystal Chronicles: The Crystal Bearers so I can write a review by Thursday it stops being a diversion. Arguably it stops being unproductive as well, but that depends on whether what reviewers earn can be called ‘wealth’.

  6. […] through but the ending left me a bit unsatisfied. (tags: animation videos vimeo pivot short 3d) Cope » Caillois completeness James Wallis on what makes a game a game, in the hopes of offering a critical tool. Ticks all the […]

  7. Tom P says:

    Maybe it’s useful to think of MMO*’s as environments/platforms which support games, raids etc. as well as non-callois complete activities; gold farming, socialising etc.

  8. Hi James-

    Folks here at GameHouse (my new gig) are reading your post.

    Didn’t you once publish a gaming magazine in one issue of which someone opined that The Sims isn’t a game, either, but more of a toy? That’s what Farmville seems to be, an interactive, virtual toy.


  9. […] Cope » Caillois completeness Interesting set of metrics for judging the game-ness of something. (tags: games) ← Links For Thursday 11th February 2010 […]

  10. […] James Wallis proposes a new standard for distinguishing between games and … non-games (eg. software toys.) […]

  11. […] James Wallis lauds Man, Play and Games. Wallis dispenses such high praise, in fact, that I am actually a little upset […]

  12. […] inimitable James Wallis has bravely salvaged his blog and started posting again, pointing this week to Roger Caillois’ book Man, Play and Games, and summarising some of its main tenets. In […]

  13. […] James Wallis, Callois Completeness […]

Leave a Reply to Carlton Cancel reply