What Joel's point is, I believe, is that generic umbrella series/brands are bad for user/customer perception (such as his "For Dummies" example). And as the entire Stack Exchange ecosystem is going to cover a massive range of topics it makes more sense to separate them as the good perception of a Stack Exchange cooking site isn't going to help the perception of a Stack Exchange mechanics site.
That's not to say I'd put up with crap on a mechanic's QA site even if the cooking QA site was good, but previous experience with Google search absolutely helped sell the Gmail switch.
I'm not saying that there is no such thing as a bad umbrella brand, just that there is such a thing as a good umbrella brand (at least for me). Springer (which also covers a massive range of topics!), and possibly also O'Reilly, are examples of umbrella brands that I perceive as good. In fact, I can think of far more good umbrella brands than bad ones for books; as a bibliophile, I rely on my knowledge that I can trust certain publishers for quality, format, content, &c. to help me make my buying decisions quickly.
I can't think of many examples of good or bad brands for websites—except CPAN and its derivatives, which are (again, to me) a very good umbrella brand indeed. (I can think of CTAN and JSAN right off the top of my head; I'm sure that there are others.)
Although with stack exchange it doesn't matter to much, because the sites all have the same look. So that works the same way the domain name would.
Likewise with these StackExchange sites; some of them I had no idea they were stackexchange sites until I actually visited them. It took quite a while to click on their links after seeing them around the net in some cases. Had it been superuser.stackexchange.com I would have immediately known it would probably be a high-caliber Q&A site.
I guess I'm an outlier on both points?