Thief, on the other hand, was one of the first 3-D first-person "stealth games." The player's objective was to get through a series of missions without being detected. The innovative concept, which drew heavily on the tension created by what the player could not see, required advances in technology: to make the simulation more realistic, Thief's designers had to work out the subtleties of light and shadow, creating mathematical models that represented how well a character could hide in the darkness. Were the surrounding guards unaware or suspicious? Was the shadow too small to conceal the character? Thief was also one of the first games that tried to faithfully simulate the behavior of sound in a three-dimensional space, accounting for echoes and the damping effects of walls.

It's no surprise that Looking Glass's MIT alumni would create a game that modeled the physics of sound and shadow, says ­GAMBIT's Tan. Video games, he contends, are simply a natural extension of the experimenting being done in labs across campus. "MIT's always been about simulation," he says. "There's always been a sense that you can replicate a reality." And people naturally want to play around with those simulations, he says. "Interacting with that simulated reality [becomes] the entertainment."

Once the simulation has made the leap to entertainment, it can still invite the users to explore, experiment, and even hack. "Think about games as a box of tools," Tan says, "a box of verbs and nouns." This vision appeals to the engineering mind-set common at MIT. "A game is something you tinker with, not something you receive as an entertainment experience," Tan says. Thief succeeded because the quality of its simulation convincingly immersed players in the world of the game.

But the more a simulation caters to the engineering mind-set, the more overwhelming it can be. Many game projects that originated at MIT have failed to find an audience, says Tan, "not because they're not fun to play, but because they're extremely complicated for non-MIT players to understand." When a simulation includes too many factors, it loses the accessibility that incites players to pick it up and start messing around.

Although Looking Glass got it right withThief, the company was also guilty of creating simulations that only a narrow group of users could appreciate. Tan points to another of the company's creations, a flight simulator that modeled the experience of flying so accurately that you nearly had to be a pilot to use it, he says. He believes that while the realism of such games was impressive, it held Looking Glass back from greater commercial success. The company folded in 2000, but its games shaped the sensibilities of many modern game designers.