Ten years ago, Raph Koster came to GDC in Austin to give a talk called A Theory of Fun, before it was ever a book that's fallen in and out of print multiple times -- it still sells 4,000 copies a year, which might make it the best-selling game book of all time.

Back then, MMO veteran Koster had just finished Star Wars: Galaxies, and the feedback was it wasn't particularly fun. Wondering if he'd lost touch with what makes games fun, he decided to look at psychology and cognitive science as his colleagues Dave Rickey and Noah Falstein had been doing, to explore the nature of play and what fun really is.

Science knows humans apply patterns to reality often unconsciously. Some behaviors, like nursing, we're born with, and others we learn over time. Playing games is a crucial education and practice tool over time not just for humans but for animals, who learn their basic survival behaviors by playing together.

"If you've ever seen a kid first learn how to walk, the look of joy on that toddler's face: It's fun. They're playing a game," suggests Koster. The brain releases endorphins in response to playful learning, and that basic concept is at the core of Koster's A Theory of Fun, which explores natural human patterns and systems to find what people naturally find compelling about games.

Some theorists separate "games" from "play," under the presumption that games are highly rule-bound while play is supposedly unstructured nd spontaneous, but part of what A Theory of Fun does was dismiss this idea. "When you're playing a tea party, it's just another system for you to learn," Koster says.

"If you're playing cops and robbers, or role-playing or making up another game with toys, it's a system with a lot more rules than Candy Land," he continues. "It has more rules, not less... games usually deal with small, constrained, tiny little rulesets you can write down. Ever tried to write down a ruleset for physics?"

Any system can be approached as a game, and since games are intentionally created to teach systems that modify the wiring in our brains, games can be viewed as an art form that re-wires people's brains. "We have the power, and that means we have to be responsibly... we actually get to engage in direct mind-control," Koster suggests.

The particular types of fun Koster is most interested in differ from flow states or pleasure-states like delight: "Art's challenging; art we have to work for. Something that's pretty and delightful... isn't the unexpected moment. That's delight, but it isn't what I would call 'fun'."

Koster sees fun as very dependent on the neurotransmitter of reward, dopamine. "Dopamine is really interesting because it specifically enhances learning and memory. Specifically it relates to predicting rewarding outcomes, which funny enough, is a lot of what we play games for," Koster explains. "It is a teaching signal to the brain. It gets dumped in you when there are unpredictable situations as well, in order to encourage you to solve them. It also decreases inhibition."

In other words, dopamine is associated with the thirst for knowledge: "Maybe fun isn't 'learning,' it's 'being curious about life,'" Koster suggests.

People do play for other reasons besides fun: To focus meditation, to explore a story, to gain comfort instead of fun per se, or for "deadly serious" practice to win a tournament. These are valid reasons to play games but separate from Koster's theory of fun.

"A lot of people hate the idea that we can reduce all of this to something so mechanical," suggests Koster. "I hate to say it, but the more science that has come out over the last ten years, the more this entire thing has been validated. There's more and more evidence to show we do in fact engage in significant, difficult learning with games, that gamers are predisposed toward learning, that games have real therapeutic value... it's all come true."

But that creates, now, a funny issue with the word "game." Abstract games that are nothing but challenge, art games that have no challenge at all, yet all are called "games." What, then, does that mean? According to Koster, game design means the creation of systems, not any of the visual or created elements.

"Every game consists of being presented with a problem, preparing to start it -- setting up the chess table -- a topology in which the problem exists, because shooting at a space invader from behind the shield or behind the field is a different problem... and a core mechanic," says Koster. "Then you get told how you did."

Look at Portal, for example; there's the macro-level of beating the game, a smaller level of beating one stage of the game, all the way down to the subtleties of positioning the gun and understanding the game's grammar. This "atomic" view of games helps explicate and illustrate the gap between what a game is, and the game's surface (what Clint Hocking refers to as "ludonarrative dissonance").

Of course, many designers are running over games with a fine-toothed comb. Independently of Koster, Dan Cook came up with "skill atoms" in his Chemistry of Game Design; Ben Cousins measured the amount of time you spend in the air jumping in a wide array of games and found that an optimal time exists. Designers research games closely, define their science, and diagram them.

Yet what is the black box at the core beneath it all? There are only four core mechanics in games, Koster theorizes: Solving problems heuristically. understanding other people and social relationships, mastering your physical relations, and exploiting the natural human difficulty in estimating probability.

At games' core, they're entirely about math -- but as someone with a Master's degree in poetry, Koster has a hard time accepting this. "It seems to me that math has real problems expressing a whole bunch of stuff. How do you write a game about the taste of a peach? How do you touch the ineffable?"

Yet so many art games -- Rod Humble's The Marriage and Jason Rohrer's Passage -- were derived directly as responses to A Theory of Fun. Koster sees a spectrum with accessible entertainment at one end, and art that requires literacy at the other.

Entertainment is conservative and familiar, while art is risky, challenging patterns we don't yet understand. It enforces -- sitcoms help us do social norming and understand how our culture works. It provides the delight of pattern recognition. But art is challenging and offers new systems to master (a bit of info you can use if you ever get into a "games as art" debate).

More and more we create games that create lots of surface and very little "black box," games that become button-presses leading to events, one after the other. "It's so much easier to express art through story and movie-making than it was through game mechanics," he says. But does that mean games like Dear Esther are really games?

"It might be we're creating a new kind of entertainment that isn't 'game design'... we might need a new name, because a designed game is an interactive experience, but not all interactive experience are designed to be games. And maybe there's such a thing as "ludonarrative consonance," where some associations -- like uni-directional platformers and the meaning of life, or colonialism and MMOs -- just naturally fit.

"Am I seeing everything as systems because that's the way the world is and that's what games are? Or... am I approaching it all this way because games trained me to see everything as systems in the first place?" wonders Koster. "Because we design either through intention or accidentally by omission, we are changing a brain."

But the things that make us the most happy are the things that games do really well: Social connection, gratitude and generosity, optimism and striving for goals.

In the end, if fun is joy, and the grand pursuit of happiness, that's enough for Koster.

Gamasutra is at GDC Online in Austin this week. Check out our event page for the latest on-site coverage.