Are perpetual betas really such bad things?

Strangely, a lot of the feedback I'm getting from well-versed technical people about Web 2.0 doesn't fare well for the concept of releasing software in perpetual beta. Is this really such a bad thing?

A canonical example I cite is the dictionary that ships with Microsoft Word. Once MS rolls out a product update, that's that, until the next version on CD comes out or the user downloads add-ons. I've pretty much given up using Shift+F7 to look for meaning and antonyms, now using Dictionary.com and its family of wordsmith sites. Those aren't exactly the first URLs that comes to mind when thinking of Web 2.0, but they do leverage the fact that being web-based, they can make changes to the core application as often as they want and instantly re-release the software, with most users being none the wiser. And they don't have to worry about cross-platform ports.

We can only assume that the Windows dictionary is mirrored on the Mac version of Word.

Maybe developers don't like the label implying products that are eternally unfinished, but the quality control aspect is the key. Flexibility, QA and choice are what make the concept a new driving force in software development. I'm puzzled why so many people are rejecting the theory.