I went to my psuedo-cousins for Thanksgiving, and (over)ate the proper amount. Food was delicious. (Repeat, etc.)
Before I left, I started memtest86 running. I've been having kernel oops'es every night pretty regularly, and google said it was either bad ram or a bug in the rmap code. (Or both!) Since testing for bad ram was easy, I did.... Sadly, when I got back I had some results that seemed to imply that I had a bad address or data line on one of my chips. I moved the dimm containing it to the second slot and then told Linux not to use the affected RAM (and about 400 other MB). I'll call Kingston in the morning, and hope they'll take it back under warranty. I considered trying the BadRam patches, but there aren't any up-to-date patches, and I'd much rather have properly working RAM.