If I practice something, if I am trying learn a structure by rote, I am an naturally inclined toward the automation of the performance. But I am REALLY not interested in standing on a stage miming to a backing tape.
I have been working on a truly live approach to performing electronic music for years. Ableton Live was a godsend, especially with Reason piped through. I still find Live limiting in that it can only have one time signature per track and has unsophisticated quantising, but other than that it does more or less what I want. I discovered it when I was attempting to create my own software to do this sort of thing - but where my software was very free and visual (in theory..), Live is all about quantisation and locking everything to a strict beat.
I think a lot of the success of my 'Installation' experiment came from the fact that the music was truly live. It always started from scratch. It may have taken a while to get going sometimes but even then it was obvious that I was DOING something. It was not artificially limited by a decision to automate one thing but not another - I was making the software work to its strengths.
Not all gigs, however, are three hours long. I am back into the statutory 45 minute slots again, in front of healthier audiences, but thrown back into that panicked confusion over how to prepare. My practices, when not dominated by software and hardware configuration or audio file organisation, frequently end up with me jamming over some random little tune I played to test a patch, and turning into material that was lovely then but never becomes a fully structured 'track'. These tangents are where a lot of my live sets come from, but they lack that recognisability that can only come from working with established music that people have already heard.
I don't know if I'm fighting a paradox or not. I could distill my live set to a perfectly executed backing track but then it wouldn't be live. Or I could do what I'm doing now: not practice much and rely on naturally occurring hooks to latch onto and work with. That's risky. Risk looks great, but it's... well... risky - there will be gaps where the quality flounders. Or a mixture - some prerehearsed stuff - perhaps the more vocal stuff - deviating into jamming when a hook emerges.
Pah - I don't think I'm getting any closer to an answer. This entry branched into a 'what makes a great gig?' listing - I'm going to post that list next in the hope of distilling its essence into something I can use...