Is it possible or even desirable to have one search interface that serves every need?
I have about 10 minutes to write this placeholder of a post. Hopefully, I'll get the opportunity to revisit this topic near and dear to my heart later.
I've often railed against naive librarians and administrators who insist we need "google boxes" as our only interface for every system, for every need, regardless of what is behind the box. In fact, we just
fought this battle had this discussion with our enterprise search consultants, but anyhoo.
This particular post was prompted by Martin Fenner's discussion of the new PubMed redesign. He rightfully (IMHO) points out that this one interface is supposed to serve clinical medicine, research in the life sciences, librarians supporting those two, journalists, parents of sick children, etc. It's also supposed to help find journal names, researchers, exact citations, genes, proteins (to be honest, I don't know what all of those choices on the dropdown do - in fact, I don't know what most of them do). He says that he thinks its leaning more towards life sciences research and away from clinical medicine now.
Pubmed is actually a great example. If it were possible to have a single interface, and NCBI had provided it, then the proliferating other tools that also search the data would not exist. Clearly, you need different information if you are in those different groups. You also need a different interface - and by interface I mean support in query formulation, results pages, and help adjusting your search - for the things besides journal articles by topic.
But if you go to the trouble to design different interfaces, how to you funnel people into them? Based on their query? Have them select (as if they will!)? As a way to narrow?
Maybe you don't. Maybe you provide an API or an index that lets other communities construct the search interfaces they need.
This wasn't always feasible, goodness knows, but seems to me it is now.
My short answers are no and no. Just like 'one size fits all' doesn't, one search interface does not serve every need. My thoughts are in line with Martin Fenner's; google-like search boxes will help quick searching, but won't really improve advanced searching.