Unfortunately, the database that we use as a backend doesn't like it if you feed it improperly formed utf-8 unicode characters (i.e. it just chokes). I hadn't realized that people would generate these, but Stephen Nuño indirectly pointed out that this is certainly not the case.
The fix is quite simple, it's just a perl regular expression to replace high characters with their html entity equivalents (the numerical html entities are quite helpful for this). Now one can enter such characters into the title, name, and post, and there won't be problems. I just wish that I had thought of this earlier. Still, the fix is quite pro-active and should prevent any recurrance of this sort of problem. (I've also added it to the comments and trackbacks, as well.)