I've been a bit backlogged but finally got time to read this very interesting entry on Paul Lamere's Music Machinery blog (via Boing Boing). Mr. Lamere plotted the beat deviation of a bunch of music. If the drummer keeps perfect time (or there is no human drummer and it's all sequenced), the graph would look flat like this (the Y axis is milliseconds, so this is a tiny, tiny variation):
But if they float around the beat, like John Bonham (my childhood hero--I taught myself to play drums listening to Zeppelin), you get something more like this:
While Mr. Lamere started this project looking to see which drummers were playing against click tracks (effectively listening on headphones to a metronome while they are playing--more info in my book, of course), I think the bigger, more interesting outcome of his research is just the presence or absense of significant beat variation and how it relates to types of music.
I've long been interested in this very idea, and have been arguing in my field in recent years against the use of rigid, time-constricted control technologies (SMPTE Time Code, etc) in live performance. But I've also been questioning myself--does anyone care if a performance is rigidly locked down to time? Mr. Lemere has found an interesting technique that might be applied to that question, and I'd be particularly interested to study this in live performance too. Any PhD candidates out there, in need of a research area? (I've got a bunch of other related questions too.)