Confessions of a Hypertext Author

Confessions of a hypertext fictionalist

Since some people are doing hypertext fictions, I thought I'd post some of what I've wrestled with in terms of process.

Once upon a time, I started by expecting to break down a hypertext just like I did a print narrative. I could perhaps use my hypertext preparations to create a print work, but the reverse didn't work -- for me, at least.

In a regular fiction, I had found that when I had some world-impression I wanted to render, I could find or invent some incident or series of incidents that would reveal salient aspects of that impression. Succeeding incidents or aspects of the same incident would reveal succeeding aspects of the idea. These aspects would accrete until they reached a point in which they started to inform each other in a way that might give readers more than the simple sum of the different parts -- the crisis or climax that we're all probably used to hearing about (OK, this is all thouroughly investigated by Viktor Schklovsky and Vladimir Tropp and Gerard Genette, among others).

Well, the failure or my failure to establish T_I_M_E and temporal sequence in hypertext loused that up. I eventually shifted towards looking at the same scene through the eyes of various characters, hoping to display the various aspects of the insight by moving from POV to POV rather than moment to moment.

I quickly found (but only slowly understood!) that I not only had to structure the story differently (I had wanted to anyway) but had to move differently from inspiration to finished product. After groping along for several years, I wound up with something of a system that I cobbled together and blundered into without instruction, and which I pass on just FYI, hoping for amendment.

I began to work more or less like the following. Some of it is standard fiction-class fodder, but some of it seems (to me at least) different enough to post. I suspect it will sound insanely obsessive, but this was a project of some 500 lexia and literally thousands of links, which took well over two years to write and code. I give here some account of the full process, and leave anyone interested to abandon the excess wherever possible.

1. I started with some idea, some image in which my concept seemed rooted or by which it seemed generated.
2. I found a representative incident that contained or represented all or as nearly as possible all aspects of the insight.
3. I lived with the incident, usually a long time, often drawing various diagrams, writing little snippets of character-thought or dialog, talking to and with the characters about many things, constructing histories for them and so forth.
4. I constructed a grid of critical questions I wanted to answer in detail before I began.
5. I wanted a character or characters to represent every aspect of the piece that I wanted to display. (I find no need for a 1-1 correspondence here. The fewer characters represent all aspects the better. Forget trying for complexity. If you're honest to your idea, it will hunt you down.)
6. As much as possible, every character should have something to say to every other character. At the very least, each character should have something to say to somebody. This means the character wants something from that other character. Any character who does not should go get his or her own story. Characters may have complicated ideas about what they want or how to get it, but desires themselves are pretty straightforward and universal. Desires make the heart go thumpthumpthump, and they relate crucially to the inspiration that made your heart thump -- whatever that might be.
7. I wound up classifying each block of text something like this:

• speaker, listener, person discussed, abstract theme, concrete motif, tone.

  I suspect this seems at once obsessively precise and too hopelessly crude to define character interatction. It is at least the latter, but I had become desperate to wreak order, and I figured as long as the artificiality of this substructure remained lost to the reader, I could use it as a temporary scaffolding.

8. I gave each block of text a name, as described in my previous entry "Webs and Nomenclature," and in the drafts, before I began to link them, I appended a longer file name to the end of my standard hexadecimal filename to help me remember what it was about.
9. Then I began to map the whole project more seriously. Who said what to whom would decide which lexia appeared onscreen at once. (Obviously, one often does not want this to happen; I had decided to do that in hopes of establishing maximal connection between characters. I suspect that if I'd had fewer characters -- say, 2 or 3 -- readers would have found the mix more comprehensible. I have come to believe that 3 may be ideal or at least easiest and 5 might be maximal, for reasons related to Walter Murch's analysis moviegoers' perception of sound mixes -- more if anyone's interested).

   See my previous entry on nomenclature for a description of site mapping.

10. Then I'd get a list of all things each character had to say to every other character.
11. I'd match the loose diatribes I'd written in various character voices.
12. I'd write in voices for characters who hadn't said what they had on their minds.
13. I would map, assemble links, remap, reassemble links.
• Proofing for broken links took a HORRIBLE amount of time, but then, I was writing many links by hand at the time.