Yeah, as I go through the list it's interesting to see the items where PG is outlining a broad problem (i.e. #7) versus something where he is throwing out hints around a solution that he's thought of. For this one, I kept coming back towards something like applying the Kayak/Sidestep model to airline miles. But I'm not quite sure how that would work, so I decided to go a completely different course. I agree with you, though, there's probably a much better html-parsing-based idea that could come out of this one.