The problem here is really one of RSS — there is no differentiation between entity-escaped HTML and plain text in any node, nor a certified XHTML content quality.

It does totally blow. Quite frankly, if you are using Rome Fetcher, we are already second guessing update times vs Last-Modified vs Etags vs TTL and UTF-8 vs Win-1251 encoding and a THOUSAND other things that people screw up all the time, even with Atom it is hard. Add to that (in my area) Apple and Yahoo not honoring capitalization or trailing “/”s in namespaces, and the number of workarounds in Rome grows to a truly horrendous level.

Rome does a best-fit guess well. If you need some additional code, I will gladly share some of my wrap arounds, but frankly, there are still just some shortcomings in the specs, and a MILLION bad feeds to try and deal with.