Confused? Refer to this blog post for details...
I tried the 1st satisfactory iteration of the piece (source material may be found on the blog post linked above) in a setting with an audience. Granted, this particular time, the whole process was a bit rushed, but regardless the experience opened up another line of inquiry for this composition: what about the people watching? I had originally intended for the participants themselves to embody both the role of performer and audience, blithely suggesting that the experience of participating is the content of the piece. I have revised this line of thinking. How can I incorporate a passive audience into the experience of the piece, while keeping the emphasis on the process?
I began thinking about what impressions I'd like this audience to come away with. The topology and dynamics are a factor I'd like to communicate. More important than even these structural ideas, however, is the understanding that the performers are interacting with each other in a complex and challenging way. I decided that a program note written in prose may be helpful to this aim, but perhaps a more elegant solution could bring this to light. What if I notate the ruleset like a map?
This notation is much less useful to the performers, who require less information about the system as a whole, and more information about their specific task. Furthermore, the layout is too confusing if one is to react quickly using this schema. To this end, I will keep the individualized rulesets as cards with trigger word - section number pairs. Furthermore, the full source text could be projected behind the performers so the audience member could follow along, if they wanted. The performers would have the text on music stands in front of them, along with their cards. More tests will determine whether a single source text works better than 4 copies in this setting.
For version 2A, I expanded the source text by elaborating on the material, preserving the outline format. I added more material, from other class notes. Finally, I felt comfortable enough with the format that I wrote some new material specifically for the source text. I imagine further additions to the source text to be more poetic as I become even more comfortable. The outline format allows a certain poetic that balances the representational with the abstract. Since the source text was bigger, I was able to find more significant trigger words. This should make the task easier. Also, I edited the way I selected trigger words. Now, for each person, I select one of the top 4 most frequent words in each section. As a result, each section is guaranteed to activate each person. I made sure that each person has the total same number of trigger word instances in the entire text. In the next iteration, I'll take into account the sum of trigger words pointing to each section, across people. This way, I'll make an even more level playing field. To generate these rulesets, I'm using a matrix of trigger words, with the vertical axis corresponding to the different sections of text, and the horizontal corresponding to the performers. Instead of payoff, as in typical game theory matrices, I'm substituting trigger words, which are valued according to their frequency. So, to sum across a column is to determine the likelyhood of a single person getting triggered throughout the game, and to sum across the row is to determine the likelyhood of a section of text to be activated. I've been doing these by hand, and I'm totally content working this way, because I've developed a few software tools to reduce the busywork, and also I like having the ability to intervene. I may eventually generalize this algorithm to a set of constraints, which would act as a fitness function and totally automate it, but that would be more to impress my geeky friends than to actually help me compose.
I haven't tried this one out yet, so if anyone wants to get three other friends together and give it a shot, that would totally blow my mind. I'll cook you dinner if you document it.