I think this exchange has been extremely valuable since it illuminates an area of genuine difference in the views of catalogers vs. programmers.
I tried to elicit a real-life example with my request for how someone should go about finding information on a specific topic. This happens thousands of times every day when people ask a reference librarian to help them find specific types of information. If you are only mildly interested in a topic, it's one matter, but if you are trying to write a paper for class, or keep from looking like a fool when talking on topics with others who may be experts, you take a more serious approach.
Students have tremendous problems with this as they progress in their studies at university. Research has shown that some 80% or more people rate their searching abilities as "very good" or "expert". And they may be, for a mundane task such as finding the height of Mt. Everest, getting somebody's email address, or finding and buying a new Ipod on Amazon.com. (More or less what librarians term "ready reference") But once they are confronted with the task of finding information for a class paper--even on extremely simple topics such as the one I gave you--they discover they are helpless and don't know anything at all. They don't know where to begin; they don't know how to end; they don't know anything except to type different words into a box and it's not working.
This is when they come to the reference librarian for help, and in my experience, they are more or less in a state of shock and totally panicked. This makes the librarian's job *especially difficult* since your number one task is to calm them down. Still, I think a lot of their panic is from suddenly having to face the undeniable fact that in a realm where they believed they were experts, it has become frighteningly obvious that they don't know what they are doing at all.
One thing that traditional librarians always had--and yes, I am bringing up that dreaded *past* again! :-) -- was the control in a library catalog. They knew that within specified limits, and those limits were clearly laid out, they really could find *all books* by Dostoyevsky in the local collection, no matter in what language or how his name appeared in the item. This is something you--let me be very clear about this--you *absolutely, positively, cannot do* using only full-text tools.
So, the library catalog has always been arranged conceptually (an error shown in the statements of many non-catalogers who discuss our "textual strings" and ignore their conceptual purpose), and therefore, within known parameters, you can find "everything" about the concept "cats" in a library catalog. See for example, this search in LC's catalog with some very nice suggestions for other searches:
Browsing in this way shows clearly what is available in the LC catalog on cats and even provokes the searcher at times with the unexpected, e.g. Cats--Caricatures and cartoons, or Cats--Humor, or Cats--Psychic aspects. This is how a catalog is arranged: by concepts, and how it is supposed to work. When I have shown people how it works, they find it quite powerful and extremely fast. Also, with our conceptual controls, we can separate the concept "cats" from other conceptual entities that are not related but use the same text, e.g. Cats & a Fiddle (Musical group), Cats & dogs (Motion picture) Cats & Jammers (Rock group) and so on. Try searching Cats in Google (it would probably be best to turn on the Safe Search, but perhaps not) and compare it to the subject search in LC and with an open mind, consider if the Google result is anything that could be useful or merely chaotic.
Google and other full-text tools do not have this kind of control, and librarians miss those controls a lot, I assure you. This is why experienced librarians often look askance at the new tools.
What I just outlined is becoming lost knowledge, especially so among the public, but also increasingly among younger librarians. But I submit that ignoring these controls that *are not replicated anywhere else* or dismissing them out of hand actually limits our imagination as to what can be done.
The method I sketched above (which becomes far more complex in practice) is hopelessly obsolete in many ways that I won't go into here, and this I readily admit, but the control it provides is not.
The future I foresee that would be best for our users is not to abandon the controls found in a catalog but to adapt them for the current "user needs" (whatever they are determined to be, and not those determined in FRBR), perhaps dumping some things, and working with the additional access provided through ever-improving full-text searching capabilities.
Perhaps someday, the conceptual work done by human catalogers can be done automatically by machines. But nothing I have seen makes me think it will happen anytime soon. If our controls should vanish, and this may be a possibility with the economic crisis and the generalized view of "traditional cataloging = continued obsolescence," it will impoverish us all.