Didn’t see that question back then. Agreed, form is one of the things often not discussed in the electronic domain. Having a classical background myself, I’m also very much interested in the topic. I think one of the reasons it’s so hard to go through it systematically is that the working processes with the computer can vary so widely. For that reason, general answers, particularly when based on formal ideas from the instrumental domain, are often not adequate.
On the positive side, with computers, we can test musical form empirically, which is a great chance! Finally, it’s all about judgment and trusting your ears. To check the formal consequences of an algorithmic idea, test it out. I’m sceptical about numerical proportions delivering a sustainable base for formal decisions. There’s nothing magical about the golden ratio or other numbers. They may work or not; their status primarily stems from their long tradition in art history.
I can briefly sketch how I appreciate form: I let it depend on the changes of timbre that are possible with specific setups. For critical parameters, one could define changes gradually or discretely. My favourite vehicles are Envelopes, within SynthDefs, or in combination with Patterns.
Although I had it in mind, I didn’t mention the term “form” in this thread’s post. For me, the selection of parameters for timbral control and the formal decisions are closely related. But of course, that’s just my choice.