Over there, I'm whingeing again [So, going with your strengths then, are you? - Ed.] about the U.S. National Center for Biotechnology Information's venerable PubMed literature search engine, and how maybe, just maybe, Google Scholar is a better bet. Dear old PubMed is a tool that many (most? all?) scientists, at least those of a biomedical bent, are very used to using to find published articles. Unlike Google, it sits on top of a curated, and keyword-indexed, collection of relevant literature. Also unlike Google, it is completely unable to search inside articles using free text queries, and appears to occasionally be somewhat less than adept at finding things.
The post is here: PubMedically failing.