Language

Balance of Power first brought up the problem of language. The headline generator started me thinking in terms of sentence structure as a means of representing the game state. At first, I missed the mark: I thought of sentence structure as merely a clever means of expressing the state of the game rather than the events taking place in the game. Conventionally, game states are represented with little bar graphs or icons. In Doom, for example, the state of health of the player is depicted with a small iconic face showing varying degrees of health. The player on the brink of death is shown with head hanging and blood dripping from the nose. I myself have used such methods in many of my games. So I was quite pleased with myself for dreaming up this idea of using language to communicate the same idea. What I missed, at first, was the idea that I could also use language to report events. That realization came to me serendipitously and opened the door to a cornucopia of possibilities.

I further developed the use of language in game design with Siboot. The headline generator for that game elaborated the initial concepts used on Balance of Power. Much more important was the inverse parser I developed for input. Using language structures for both input and output changed everything.

Older readers may object that there's nothing revolutionary in this use of language; after all, text adventures have been using language for both input and output since the 1970s. But the parsers used in such text adventures are crippled by an internal contradiction. On the one hand, they seem to offer the player the prospect of communicating in natural language; on the other, they never come close to delivering on that offer. No matter how big you build your parser, no matter how large its vocabulary, the user will always struggle to find the right words to express his desires or, more likely, like the man in the badly tailored suit, learn to live with a crippled vocabulary. The best evidence of the sterility of the parser concept is the fact that parsers have been the great software dinosaurs of the last twenty years. DOS, a parser-driven operating system, has already gone extinct, and text adventures are now the preserve of a small and dedicated band of retro-gamers.

The inverse parser system, however, still has lots of punch left in it. I attribute its rarity to ignorance, not conscious rejection. Nobody knows enough about inverse parsers to dismiss them based on noun. I myself see plenty of design potential.

Once I stopped thinking in terms of "user interface" and started thinking in terms of linguistics, my view of game design, interactivity, and software in general began to reorganize into a clearer system with a solid foundation. My studies in linguistics continue to illuminate my thinking on game design, while providing me with crucially useful nuggets of truth such as this one: The words "Kathy" and "castrate" come from the same Indo-European root. The reader may wish to consult some of the references to linguistics books listed in Chapter 9, "The Education of a Game Designer."