Info Grid 2005 - Tuesday 27th - Monday summary on evolution of libraries online

I found Carl Lagoze's talk on the next-generation NSDL particularly interesting.  Basically he talked about two stages in the evolution of the understanding of digital libraries.  In the first phase, people were just striving to get content online.  At that time they maybe couldn't even imagine that within a few years, the Internet would have vast amounts of information available.  In some ways, people were proceeding from an imperfect model of the library:

The formal part of a library is books come in, get catalogued, and then people use the card catalogue to locate books on the shelves.

This was replicated in content repositories (digital libraries, phase 1).

But this missed the whole informal, social aspect of the library.  You lose both the community and the context when you just have raw indexed content online.

If I can turn a phrase, in a library, people congregate around other people, not around the card catalogue.

So in phase 2, they are attempting to bring back the contextual and social aspects of the library.  This means that what some people may consider frivolous applications: blogging, photo sharing, instant messaging, recommendations, bookmarks... all these aspects of person-to-person communication need to be integrated.  As well as automated "wisdom of crowds" technology: "the person who read this paper also read these papers" etc.

The good news is, there is no need to reinvent the wheel - the whole point of the Service-Oriented Architectural (SOA) methodology is that your role becomes as much one of integration as of creation.  Take the best of the online services, and integrate them into your site.

The other part of the SOA paradigm is, in a way, that small, simple, imperfect but quickly developed and released pieces of technology can be more successful than complex, perfected tech.  "Good enough" can be great.

I'm actually a bit concerned that we may get too caught up in elaborate Web Services architectures when simpler protocols may in some cases work better.  But we have to balance that against getting trapped in silo solutions.  The key is to architect so that components are loosely coupled.  Herbert Van de Sompel showed how doing a good internal architecture with strongly separated components means that those components can then be re-used in external applications and distributed environments.  In effect, he treats his local network as if it is a distributed network.