OCLC being a catalog aggregator, already knows an awful lot about the possibilities of serendipitous discovery when you take information out of context. A recent example is the Identities function, still in beta, but discussed at Lorcan Dempsey's weblog in several posts, among them this one: Lorcan Dempsey's weblog: The anxiety of influence. He notes the power of this type of exploration, this kind of discovery, of who or what is most influential in this case, but for me, this points to the fundamental importance of the many projects underway today that provide the fodder for these new ways of using what's been with us for a long time.
This is part of my wonder about the complaint about Google Book Search, that scanned books, searchable at the word level, strip those books of their context -- one would think that they no longer existed in any form on earth except as disaggregated words in a database somewhere, but actually, they still exist in full contextual forms. Today those contexts include, among other things, shelves in buildings. Yes, they actually get returned to the very buildings from which they came (surprise!). But digitizing them makes them amenable to so much unimaginable connecting-up. What will we learn from them tomorrow, in addition to what we learn from their text and their context today?