An issue during the time you are discussing, USENIX had evolved into "two foci" between the practitioners (which included both FOSS community and LISA types) and the more academic-oriented folks looking for respected places to publish papers/develop their tenure files.
I think this is a long and accelerating trend, and not just at conferences. There simply are no venues for "engineering" papers or presentations any more, which doesn't bother me directly, but bothers me very much indirectly, because I love engineering papers and have to read academic papers, ummm, very selectively. (In particular, anything labeled "formal semantics" just gets skipped.)