July 12th, 1999
The progress you have made over the last decade is impressive. From a struggling oddity you have matured into a medium of expression to be taken seriously. I personally do not enjoy your work, but this in no way detracts from the respect I accord you. You have a rich and robust medium to work with, and much to be proud of.
Yet I grimace at the ceiling you have placed over yourselves. I do not refer to the lack of spectacle in your work; as far as I am concerned, ungarnished text is just as expressive and powerful as any number of polygons per second; it goes through a different section of the brain and reaches down into our mental bowels just as effectively. No, my concern is more abstract; it is with your structural limitations.
As I understand it, all hypertext is uses a branching tree structure. Computer scientists use the confusing term "graph" to describe such a structure. Nevertheless, I will borrow their terminology to state my case succinctly: you are using unweighted graphs when you should be using weighted graphs.
Let me explain that last sentence in real English. Your branch points are all treated equally; there is never any bias based on context. You cannot set up a branch point that is likely to veer left if the reader has shown a sense of humor, or right if the reader has demonstrated gravitas. Your work cannot truly react to the user; it is, in every sense, just as dead as the printed page.
I do not say this to denigrate the printed word. Dead it may be, but it retains plenty of emotional power even when dead. My argument is that, if we could but raise the Word from the dead, breathe life into it, then would it not command even greater power? Socrates in the Phaedro caught the difference, arguing that the Living Word (discourse) was superior to the Dead Word (writing) because it could explain itself, defend itself, and respond to its listener.
"But how can we give the Word the breath of life?" you ask, quite reasonably. My answer is, to give it the power to respond. Wave your hand at a lizard; it scurries away. Call your dog; it comes running. Cut the stalk of a plant; watch it wither. Give it water and watch it flourish. All life responds to, reacts to its environment, and in doing so, acts upon its environment. Life interacts. The printed word doesn’t.
Here we come to the most important question I can ask you: what is the artistic essence of the computer? Is it the ability to edit text, as in a word processor? Come now, that is low-thinking, mired in matters of efficiency rather than art. Is it the multimedian combination of text, graphics, and sound? Hell, my teachers were showing us filmstrips, audio slide shows, and movies when I was a kid; there’s nothing new there.
It is certainly a common error to misjudge any new medium by gauging it in terms of existing media. The Celtic druids rejected writing on the grounds that it weakened the memory. They could see it only as a means of recording ideas. Its ability to transmit ideas over distances and time spans didn’t occur to them, because they had never had any such technology before. Cinema was crippled in its early years by the notion that it was a kind of "recorded theater". It took decades for moviemakers to realize that the essence of the medium is the camera and its ability to jump all over the narrative universe, cutting from wide scene to closeup, bouncing backwards and forwards in time, unconstrained by pedestrian considerations.
I claim that the artistic essence of the computer is its ability to make informed decisions. This wondrous ability has a name: computing. That’s why we call it a computer. It’s not a texter, nor a graphicser, nor a sounder. We already have technologies for those: books, photographs, and tape recorders. The essence of the computer is that it computes. Hypertext doesn’t. Does anyone sense a problem here?
July 16th, 1999
Mark Bernstein of Eastgate Systems wrote me pointing out a series of errors in this essay. First, he bridles at my assertion that hypertext uses a branching tree structure, pointing out that most works of hyperfiction have cyclic structures. He also observes that many hyperfiction works have "guard fields" that permit the software to take different branches based on prior user choices.
I readily conceded Mark’s points -- he’s absolutely right and this essay attacked the problem with too broad a brush. The core concept that Mark and I agreed upon -- I think -- is that there’s a need for something richer than guard fields in computing branching decisions. I argue that we need many more variables available for taking into account during the branching decision, and much richer algorithms. However, Mark points out that there’s plenty of room for non-narrative hyperfiction, and such material may not require so much algorithmic richness.