Online Catalogs and their Peculiarities

A recent article in The Chronicle of Higher Education looks at problems in online library catalog databases and some efforts to fix them.
This is a topic of ongoing interest to me, but I can't say I think the article addresses the problem with sufficient depth, and the comments on the article suggest that I am not alone in thinking so.
We're presented with a specific case: a grad student gets poor and frustrating results when using University of Virginia's Virgo catalog. We aren't told what the Virgo search options are or where in the interface the student typed "Thomas Jefferson." We are told, however, that new and better software is taking over.
I haven't used Virgo, so I don't know how it's set up. But I've used Melvyl, Gladis, PittCat, PittCat+, Library of Congress, WorldCat (and is this or is this not the same as OCLC?), and various other library databases in the US, the Czech Republic, and Great Britain. I'm not a librarian and I don't consider myself an expert on library software, but we could say I have some experience in using and getting used to different catalogs.
In most catalog software I've used, the user chooses whether to put "Thomas Jefferson" into search fields such as author, title, or keyword. This is usually but not always pretty effective. Things aren't always properly catalogued. For instance, when I returned from the Czech Republic, I spent a lot of time looking for materials at UC Berkeley--generally books for which I had full bibliographic data. I was surprised to discover that often the library didn't appear to have a book when I searched by title, but that an author search would bring the book right up. I think the reverse was also true, and that sometimes it was more productive to locate Berkeley books using Melvyl, even though I'd normally Melvyl to see if any library in the UC system (within easy driving distance) had a book.
Some of the new catalog software, however, has you just type in "Thomas Jefferson" without specifying whether you want books by him, about him, with his name mentioned in the keywords, or with his name in some other random location. University of Pittsburgh's new PittCat+ does this, forcing the searcher to spend absurd amounts of time winnowing down the categories to find him as a subject. When I first saw PittCat+, I thought I would like it because it does allow you to narrow the search in ways that I had seen on OCLC, but I rapidly concluded that since you can't start with a focused search as in PittCat "Classic", PittCat+ simply wastes the user's time. This was a complaint I heard repeatedly from librarians and faculty. We learned rapidly that PittCat "Classic," with all its faults, was way quicker and easier to use than its replacement.
I've heard people argue that Google-style searching is the new direction for library catalogs, and I've heard people decry it. Well, surely it should be possible to have both Google-style searching (useful for fairly obscure things) and more focused initial searches. After all, if you want books about Thomas Jefferson, the problem is more likely to be how to narrow down the search, since there must be a plethora of books dealing with various aspects of Jefferson--biographies, histories, political studies, works on plantations and slavery, etc.
One of the people to comment on the Chronicle article argued that what we really need is a return to Library of Congress headings. He or she apparently taught these headings for years and imagines that they are easily learned. I beg to differ. I wouldn't say Library of Congress headings are useless, but back in the early days of online catalogs, I tried searching with them, since in those days UC Berkeley's online catalog used them (maybe it still does, in some hidden place). I'd dig around in the big red books, trying to figure out where in the hierarchy my topic might be. I understood the concept of hierarchy, but--not having taken the commenter's class--I found it impossible to guess which aspect of a topic might be higher on the hierarchy. If I had wanted to look for Czech art, for example, I would have been uncertain whether Czech, Czechoslovakia (this was before the Czech Republic existed), or art would start the string. Fortunately Boolean searches pretty much wiped out that problem. Not that Boolean searches are always simple, but the basics of Boolean searching are easy. Czech and art, or Czechoslovakia and art.
I haven't yet familiarized myself with the catalog here, but I've already had one disturbing experience with it. I wanted to put a volume of the commonly used anthology Art in Theory on reserve for one of my classes. I searched for the title Art in Theory and was shocked to find that apparently the library didn't own any of the three volumes. I therefore (after scanning the pages needed and posting them online) requested that the library order the series. To my great surprise, yesterday I got an email saying that we do actually own the volume in question (apparently not the other two) and that it has been put on reserve for my class. Well, I'd like to know where this book was hiding in the catalog.