People tell me that Chris Crawford designed some classic games. I'll have to take their word for it, because I only know him as the mind behind the laughably awful "interactive storytelling" system Erasmatron, which was released with much self-important fanfare in 1997, to the general derision of the Interactive Fiction community and the indifference of everyone else. Undaunted by this reception, as only true visionaries are, Crawford worked another nine years on his brainchild and has now come back with its latest iteration: Storytron.
Storytron's website is heavy on redundant hype, telling us:
"...you can interact with incredibly authentic computer-controlled Actors in an adventure that combines the best aspects of traditional stories with the unprecedented experience of controlling the story's flow by being the protagonist."And shortly afterwards:
"It's hard not to be attracted to it; an artistic medium which offers the rich emotional experience of a story, coupled with the empowerment of actually being the protagonist."But as yet, the website is short on concrete detail. After pages of vague promises, we are at last presented with the first gruesome screenshot of Storytron, which, out of concern for the delicacy of my readers, I hide behind this link. View at your own risk, but by all means view it.
A rather unpleasant sight, I hope you agree. For players of Erasmatron, the image of "Knifer" will no doubt recall nightmares of Gramma Mara (what is it with Crawford and big ugly faces?), but I am more horrified by the incomprehensible flowchart at its side, which recalls nightmares of a thousand software presentations. What is this thing? Part of the design tool, perhaps? Closer reading tells me that no, this is part of the game itself. It is, in fact, the text of the Storyworld, the language of Storytronics. You don't read a Storytron story, you follow a flowchart. Characters speak to you in business protocol diagrams, and ugly bikers threaten you in UML.
The UML-like diagram is an example of "Deikto", "a revolutionary miniature dramatic language", which Crawford has developed to solve the eternal problem of computer-based interactive storytelling: machines just find natural language too damn difficult. Deikto, according to the website,
"is small enough for the computer characters to be fluent in, [...] and versatile enough to express any dramatic occurrence possible in a given storyworld."By way of example, the website offers a diagram much like the following:
with the explanation:
"In this sentence, the player is saying: 'I advise Mary (with moderate urgency) that she should go... (the word "where?" denotes that the player has not yet finished the sentence).'"
Crawford knows that his language is crude and ugly, but doesn't seem to care:
"Deikto may not be the most inspiring language to write sonnets in, but the singular beauty of interactive storytelling is not in its representation - it is in the richness, depth, variety and drama of the interactions it allows."I'm sorry, but the beauty of storytelling -- interactive or no -- is very much in its representation. In fact, it's almost entirely in its representation: in the language in which the story is told, in the use of the language to create structure, pace, imagery. There is no story which makes up for its poor use of language with "variety of interactions", just as there is no good story that can be told in inarticulate grunts. The best storytellers are not the ones who come up with the best interactions, but the ones who can best represent them. The best storytellers exploit the imprecision of human language -- its ambiguity, its layering of meaning, its ability to express more than what it signifies -- in fact, the very things that make human language so difficult for computers to understand -- to make more beautiful stories. "Richness, depth, variety and drama" come not from the story, but from the storytelling language.
All Deikto can do is describe the bare skeleton of a story, the bare facts of character interactions. I could imagine such a language being effective in certain limited areas -- stories where the communication between characters is severly constrained for some reason, such as a story set on a world of Deikto androids. Perhaps the following could be the climax of a treacly ET-style narrative, as a tearful Deikto android, with his dying breath, utters the following:
Indeed, the excellent 80s computer game Captain Blood pulled off much the same trick. The player communicated with aliens through an interface ostensibly designed for the purpose, composing 'sentences' in a simple icon-based language. This language helped tell a reasonably humourous and dramatic story. But one reason there wasn't an avalanche of Captain Blood clones was that the game milked the idea for pretty much all it was worth: the icon-based language just wouldn't work in any other setting. Players could buy it as a means of communication between human and alien, but between human and human? Hardly. The same characterisations from Captain Blood, in the same language, would look coarse and laughable when transported to any purely human setting.
The very idea of using a machine-comprehensible language for general-purpose storytelling is misguided. Art, including the art of storytelling, is a human domain. It's intimately bound up with the concerns of being a living organism; an organism that wants to survive and reproduce, that must exist in a complex society of similar organisms, that knows it was born and knows it will die. And the language of art must be human language. Human language embodies our concerns; it even mimics our concerns. It mimics our evolution, it evolves with us; it mutates, it combines with other languages, it gives birth to new words. It grows with an individual as he does. In a sense, each individual speaks his own language, has his own linguistic signature: his own vocabulary, his own phrases, his own intonation. Writers exploit these properties of human language. They use its extensibility to coin new words, they use its mutability to create shades of meaning, they use its uniqueness to aid characteristion. And they use its sound: human language arose from and is still grounded in vocalisation, in a way that machine languages never were.
There's no reason a machine should be interested in any of this stuff; indeed, most of these properties are even undesirable for machine-based communication. A machine language needs to be precise, unambiguous, fixed, immutable, based on an unwavering common standard. These properties, which make computer programs so powerful, precise and reliable, and enable computers all over the world to communicate, make for pretty lousy art. No good poetry will be written in C++, no love story in first-order logic, and certainly no good fiction in Deikto.