This has been at the periphery of my consciousness for a while so I finally tried it today after reading a not so good article in Nature
that mentioned it (the article was a little misleading and really compared apples to oranges so I don't want to link to it but see v.428 p.683).
CrossRef Searching is actually pretty good. A group of 9 large publishers have collaborated with Google to allow free fulltext searching of their online journal archives and provide doi links. To try this, head to one of the publisher's sites. I tried Nature's link
. The ACM articles weren't going (because ACM is down?) but the other publishers were right on track and much more relevant that what an open internet search would yield. It's still not made for precision, however. I cringe when people refer to ScienceDirect as a database like Inspec or Compendex. This one doesn't have controlled vocabulary or field searching, either, so I would only recommend it to someone who doesn't have access to the high powered databases but will pay-per-view or ILL for articles.