Jonah: Right on. The web, as a research tool, is still in a primitive
state, unless (and in some cases even if) you can afford expensive
subscriptions to services like (in my personal dream scenario) MathSciNet.
The web reminds me of Roget’s Thesaurus: a terrific resource in principle,
but never actually much help in practice. (I think there have been about
three occasions in my life when I got something useful out of Roget.) The
web hasn’t been the same for me since I lost ProQuest. This is a sort of
poor man’s Nexis, which anyone with a NYC library card used to be able to
access free of charge. Then it got chopped in some City budget-cutting
exercise.
Even if you count ProQuest, though, looking back over my own stuff, I think
my best pieces have come from browsing around in actual libraries, or from
the immense amount of random junk in my own head, or from meeting & talking
to interesting people. The web is as yet no substitute for KNOWING STUFF,
LISTENING TO PEOPLE, and READING.
Incidentally, even when there is a decent open-access research source on the
web, it generally has a really crappy search engine. Why is this? Truly
bodacious search engines are there for the asking, with phonetic-match
searches, nearest-spell searches, wildcards, and the rest of it. They are
standard black-box modules, nobody has to write his own search engine any
more. Yet people do, and they are HOPELESS. The one that drives me
craziest is Gutenberg. You have to put
in an author’s name EXACTLY THE WAY THEY HAVE IT ON THEIR DATABASE,
otherwise you’ll get no hit. Grrrr. Abebooks
similar.