almost as if some type of internal usability test is being done to select the better of the two pages.
CainIV - I think you're on to something here. I've observed for most of last year, if not even before, that Google was using multiple results for a query as a test to winnow down pages returned.
I've also been noticing that those indented clustered results (of two results for a query) that we used to see are much rarer now than they used to be (or perhaps they've vanished entirely), and I've assumed that Google's simply being tighter about its serp real estate.
It makes perfect sense that when Google is down to two pages, swapping them would be the logical way to do an A-B test and select only one.
(PS: I just checked some old screen captures and see that Google was returning results with 4 "sub"-links for a query back in Oct 2009. I didn't guess they were using those for testing then, though, but I'm now almost certain that they were).