I'm torn about developments in search functionality these days.
Yes, the one-search-box interface is here to stay and if libraries and information resource developers don't get on the band-wagon, we will all miss the boat. Users want simplicity and nothing's simpler than one box you can type whatever you want into and results and thrown back at you. But, given current technology, the single search box can NOT provide better results than a more functional interface with more options. Searching is about communication: the user is trying to tell the system what he/she wants and the system responds with what it thinks is appropriate. Too little information and the system either doesn't know what to do or simply guesses. And, too often, too little information means the user is assuming a whole bunch of information. The system doesn't (or more accurately, the system designers don't) necessarily have those same assumptions.
One objection to this might be, "But what are 'good' or 'bad' results? Can you make that kind of judgment? Aren't all results either useful or not? And isn't it just the user that must decide this in the end?" This is all valid. Usefulness IS the value in results. It's what makes them good or bad. This is not a case of moral value but utility of the results. And yes of course it's the user that must decide in the end but, by definition, the user cannot perfectly judge results. They're searching because they don't have the answer. Hopefully, while putting pieces together, they can make a more knowledgeable assessment of the results but in some cases the user may just be judging them on whether they LOOK like their good. So there are limits to how much the user is able to be the final judge. But there must be a judge otherwise what are we doing by providing search results (or any kind of library resources for that matter)? There must be some kind of assumption that results can be objectively determined as useful or not based on user input otherwise there's no point in developing a computer system to take such input and spit out results.
The other objection is that the library is not here to make things purposely difficult. If there is an easier way of doing something, why shouldn't we provide it? Those against the single search box may describe providing one as 'pandering' but aren't all efforts to make things easier a kind of pandering? Where's the line beyond which such developments become a negative thing? And again, this is basically true. We should be making things easier. Particularly in this age where putting up barriers may serve only to scare users off. But there are plenty of cases where simply giving the users what they want, even giving your paying customers what they want is inappropriate. In education, students consciously want a perfect grade. We may suppose that what they want in the end is an effective education but if you gave every student the choice between a free A+ and an well-earned B-, the vast majority would choose the former. I have no doubt about this. Wouldn't you? Especially in this increasingly competitive education and job market. But educators do not simply hand over perfect grades despite this desire because that would defeat the purpose of teaching. On a similar note, we cannot simply hand over free search results when it's clear that this is not the best way of searching. This results in more garbage and less quality and, in my opinion, leads either to a misuse and misunderstanding of the information found, or more work on the part of the patron.
What single searching should do when used properly is simply backload the work farther down the line, which may work better in the long run (i.e. the novice user gets an idea of the info available early on which can then inform the search strategy) but unfortunately gives the impression that what pops up first at the top is what they were looking for all along. A savvy user should recognize that more work is necessary but wouldn't the savvy user be fine with a more advanced search interface to begin with? If they know enough to recognize good and bad results, they should know enough to aim toward good results in the beginning.
I guess my point is that single search box interfaces are necessary but imperfect as we now use them. Perhaps this is not really an argument with anyone serious about the topic at all.