There's a topic that's been sitting on my mind for a long time now and it probably deserves a bit more than a single blog post. I'll take a stab at addressing the overall issue though. A common saying in cataloging is that you catalog only the item at hand. That is, you're not expected to go running off to various sources or do in-depth research on each item. There's usually many reasons given for this but one of the main reasons boils down to one of practicality. Most libraries received too much material to give each item this treatment.
However, I believe this rule is becoming increasingly irrelevant. Searching and research is becoming dramatically cheaper in terms of manhours with each passing year. Given certain seed information for newer publication such as isbn, author and title it should be relatively easy to have automated searching done of the publisher onix servers, webpages related to it, Wikipedia entries, LibraryThing, Shelfari, and more. This searching can be performed and gather information for the cataloger to use to enhance the record. At the very least our systems should be smart enough to be attempting to guess appropriate authority records and offering analysis of similar records and suggest subject headings.
A good example of potential actually exits right now with the wealth of organized information for one particular medium, cds. Cds themselves can have their track information as cd-text. Barring that, there's many ways that the songs or cd can be used to pull up information from a wealth of online resources such as freedb, musicbrainz, discogs. This really deserves a post all of it's own, which I may get to one of these days.
For example involving printed books not long ago someone asked on how to create a record for two books (The Talisman and Black House) that came in the set. She claimed as far as she knew that they were unrelated. Turns out that they shared the same central character (Jack Sawyer), had re-occurring characters, and were written by the same two authors. Had she checked LibraryThing she would have found the books classed by readers as a series. Wikipedia even notes how the second book tied into Stephen King's Dark Tower series. There was no reason to even have her go to the browser, the act of putting in the isbns should have pulled up some of this information.
So lets abandon item in hand. It no longer takes a walk across several floors and possibly hours of effort to find useful information on several books. It could be as easy as doing what we're already doing, with a little smarter software and a little bit more flexibility in our cataloging procedures.