Don't think that you're too small to get people to clean up. I mailed the editors at The Register after getting fed up with having to clean up their RSS for my personal homepage one time too many. I explained that I appreciated the service they're providing, but that they weren't generating valid RSS per the specs (and provided URLs to same).

Got a nice reply back from (IMSMR) their technical editor that they couldn't do much as it was generated straight from their content managment system or some such, but that he'd send a note to the authors asking them to be careful what they used in titles. There's still the occasional glitch, but I have noticed that I don't see as many errors as there use to be.

As for fixing things, I use HTML::Entities' decode_entities routine to check. If that doesn't pass on something I run it through s/&(\S*)/&$1/