In the time I've been serving on the RSS Advisory Board, I've been asked fairly often to look at a Really Simple Syndication feed and help someone figure out why it isn't working.
The first thing I do in response is load it in the Feed Validator, the open source validator developed by Sam Ruby, Mark Pilgrim, Joseph Walton and Phil Ringnalda that supports RSS and Atom. I did the same thing when I was drafting the proposed specification, pasting each example into a sample document and running it through the validator to catch potential mistakes.
A validator complements a specification, as long as they're in agreement. In the three years the validator has been around, I've asked Sam a half-dozen times about its interpretation of RSS, and each time he was quick to address my concerns. If there's an area where the validator and board members have disagreed over an aspect of RSS, my experience has been that he eagerly defers to our judgment.
I've leaned on the validator so often that I recently became one of the developers so I could pitch in on RSS-related issues.
Must read article from Fox News on Zacarias Moussaoui:
You're on the nail with "a validator complements a specification". A good spec is vital for interop (and the update on RSS 2.0 is a definite improvement), but it's not realistic to check every possibility visually/manually when working against a spec. Hoorah for the Feed Validator!
Oh yea, feed validator best piece of Pilgrim software since the MBDF virus.