There’s a principle in game design that’s referred to variously as rubber band or elastic artificial intelligence. While descriptions of the concept refer to it as “cheating”, and discussions of the technique can sometimes devolve into a litany of frustrated complaints, the idea clearly works. Put simply, an elastic artificial intelligence helps a game adapt to the varying skill levels of players, keeping a game competitive when it might otherwise simply let one player profit from an advantage to the degree that the game would become boring, frustrating, or simply unwinnable for all other players.

In a system like Mario Kart, which uses an elastic intelligence, computer-controlled opponents can vary their skill levels, and human-controlled opponents get advantages that make them better competitors. Adversaries get more adept as your own skill level rises, and in general, gameplay gets more balanced and more satisfying for most players, with the exception of the person in the lead. The leading player is merely kept from running away with the game every single time in a well-balanced game, or in a poorly-designed one, they’re punished for being good at the game.

This principle of balancing strengths in a game is pretty important for any system to remain engaging and entertaining for all players; The same idea of balanced gameplay is seen in any kind of multiplayer game. A recent Wired article on board game phenomenon The Settlers of Catan describes its analog implementation of a rubber band AI:

[T]he game is designed to restore balance when someone pulls ahead. If one player gets a clear lead, that person is suddenly the prime candidate for frequent attacks by the Robber, a neat hack that Teuber installed. Roll a seven—the most likely outcome of a two-dice roll, as any craps player knows—and those with more than seven resource cards in their hand lose half their stash, while the person who rolled gets to place a small figure called the Robber on a resource tile, shutting down production of resources for every settlement on that tile. Not surprisingly, players often target the settler with the most points.

In addition to deploying the Robber, players will usually stop trading with any clear leader. In tandem, these two lines of attack can reduce a front-runner’s progress to a crawl. Meanwhile, lagging opponents have multiple avenues for catching up.

The interesting thing is that, while a generation has grown up familiar with the idea of elastic balancing, it’s still a relatively controversial idea. In games like the Madden football series, where its effect is heavy-handed and obvious, it makes sense that players would balk. But some part of me feels like a lot of the people who complain about these features are the sort of people who object to letting their nephew win at Mario Kart sometimes, just so the game is more fun for everybody.

Interestingly, the “rubber band” analogy for this kind of game balancing is more apt than it might seem at first, when you consider the effect it has on keeping a game fun and lighthearted for a group of players. Rubber bands actually heat up when they’re stretched, but cool down when they’re restored to their natural state. So an elastic AI really is trying to function as a system that “cools down” a crowd of people who are playing a game.

I’ll leave it as an exercise to the reader to extrapolate on the political impacts we’ll see from a generation growing up seeing elastic artificial intelligence as an important part of keeping harmony in a community of people playing the same game. For the record, I’m a very good Mario Kart player, but I’m still strongly in favor of its use of elastic artificial intelligence.