How the heck does something like this happen?
It would appear that the workunit had a quorum of 1 but an initial replication of 3--not sure why that is, on a sieve workunit no less. At any rate, it appears that first I returned my workunit (_1), then another guy got the _2 and returned that. He took twice as much CPU time as I to process the workunit, which is a little strange since my computer is a slowish AMD Phenom (first generation) and his is a Core 2 Quad Q9400, though not necessarily indicative of a problem.
Somehow, after he returned his result, mine was marked as invalid. AFAIK, the only way a sieve result can be marked as invalid is if it returns a bad factor. I'm pretty confident in my machine's stability (it's not overclocked and is well under the thermal spec temperature), but just in case, is there a way to tell whether my result returned any factors? (Presumably, invalid results wouldn't count towards the total factor count on the "Your Account" page.)