Lorcan Dempsey's Aug. 8, 2010 blog posting on "Sorting out demand" is insightful and useful. His ideas are his 3rd top trend as presented at the 2010 LITA Top Tech Trends panel at ALA Annual Conference. http://orweblog.oclc.org/archives/002124.html
His argument about the shift (in libraries) from a focus on managing supply to "sorting out demand" is an economic one. Costs - in time, effort or money - for the user drives what services the library can provide and the "right" structure for the library. Libraries in the 20th century reduced the user's supply-based transaction costs by integrating the many sources of supply and bringing them close to the user. Libraries in the 21st century (or this early part of it)must also reduce user costs but those will be different costs as the supply transaction costs are falling due to the effect of digital format and the network. Dempsey mentions several examples of how libraries can provide services on the demand side: recommendations, contextualizing content for particular communities, connective services, tailoring content to purpose, and managing institutional assets. Each of these examples is interesting and promising, but not seem to be as compelling as the 20th century economic rationale for libraries.
I continue to think that libraries as we know them will be changed in ways very like bookstores and printed journals or newspapers; they will not serve the necessary local distribution needs as well as a globally networked provider. Some will survive because of local peculiarities or because they develop special services for their community--these may be the same thing. Research libraries will become more research museum-like, that is, more artifact-centric. But how a university, for instance, manages its institutional digital repositories and its licensed online resources may make more sense outside of the library. Yale has created a university-wide Office of Digital Assets and Infrastructure, that, though in its institutional infancy, is clearly the focal point on digital repositories at Yale, digital preservation at Yale, and discovery of resources across the university's many collections in libraries, archives, museums, etc. The changing economics will change the institutional structures. The more radical the changes in economics the more radical the resulting institutional changes.
Lastly, Dempsey closes his post with a link to Dan Chudnov's prescient 2006 post "help people build their own libraries" http://onebiglibrary.net/story/because-this-is-the-business-weve-chosen