Goldberg: Future Work

What sorts of interactions should we support?

• Exploratory, random, unpredictable. This suggests allowing names and visual cues to be turned off, forcing the user into a less focused exploration of the space. It also suggests the use of rating rather than choosing, and perhaps making a larger number of variations available at each point.
• Refining, developing. Although still exploring possibilities in a way not entirely under user control, this kind of interaction is more about refining something good into something better or generating a few good variations to be used together. Providing variation names and visual cues is more important here, and choosing seems better suited to this kind of interaction than rating.
• Writing, laying out, arranging. This is more the "traditional" model in which the composer takes the ideas and phrases (perhaps generated in one of the above modes) and arranges and assembles them into a complete piece. The Music Processor interface tends to fit this mode of work.
• Real time performance. More than one survey respondent expressed interest in using something like Goldberg during a recording or performance situation as an improvisational tool. To support this kind of use, we would need to make it possible for the user to have a very good idea of what Goldberg is about to do (perhaps by using names and visuals). Limits on the amount of variation may also be desirable, so that the performer will not be completely surprised. On the other hand, maintaining some amount of unpredictability would be seen as a benefit by many users who might imagine jamming with the computer interactively.