[dap]

> If you take a group X, a certain percentage Y hang out in newsgroups and blog. He himself was part of that Y for Forth. Why would the percentage Y be significantly different for any given language?

There are lots of reasons I can imagine: people trying out newer languages have more reason to talk publicly about what they're doing, since they usually have a vested interest in the growth of that language. Organizations with tight rules on secrecy are often risk-averse in general and less likely to try out new languages. So new languages wind up with users that are more open, and they appear even more active. (I don't know if this is true -- it's just a guess. But your assertion that the percentages would likely be uniform across languages feels unlikely to me.)

I'm not sure the argument is that the percentage who hang out in newsgroups and blog is different, anyway, but rather that looking at newsgroup activity and blogs specifically selects out many of the actual experts. This has absolutely been my experience, and one of the most damaging assumptions I see is that surveying blogs and twitter feeds is enough to get a complete understanding of how people use a piece of software and what they think of it.