Could have been Ted. I don't remember that particular session, but sounds like something he'd present on.

>Once on a conference, you may remember actually, was it Ted Roche who had a session about extracting data from HTML... and out of the 45 minutes, some 30 was about how to clean up the mess left by Word (as someone up the chain decided to use it for generating it in the first place) so the actual data could be extracted.