I don't mean to be rude or smug or anything, but  it seems that Eno uses the generative model to write his lectures.  He seems to have several different ideas/references that he rearranges and recombines to make the same points in different ways again and again.  This perhaps points to a problem with the generative system as a creative strategy.  As he says himself there are no real surprises only subtle variations based on your inputs.  This means that a generative piece can remain consistent and self contained like any other art work, but rather than capturing a moment in time, it lives in time.  Maybe this is not really a problem, but it forces us to look at the art object in a different way.  If algorithms can really be expected to challenge the traditional art object than we can't simply present little chunks of them.  These arbitrary selections of potentially infinite or really long sequences are just like any other dead, captured objects or sounds.  I don't want my ipod to play 20 minuets of 'its gonna rain' I want it programmed in there so that each time I hear it I've plugged into a different section of its endless march through time.