[I first published this in March 2008, and still find it useful to think in terms of concentration and diffusion. I was reminded of the entry and prompted to 'recycle' it - despite its age - when I read the following sentence in a recent blog entry by Nicholas Carr "Although a network can be a means of diffusing power, it can also be a means of concentrating it".]
I find Web 2.0 increasingly confusing as a label; no surprise there. This is not just because of its essential vagueness, but because I think it tends to be used in a couple of very different ways. Where this happens there is bound to be some confusion. Schematically, I will use the labels 'diffusion' and 'concentration' for these two ways.
diffusion is probably the more dominant of the two. Here it covers a range of tools and techniques which create richer connectivity between people, applications and data; which support writers as well as readers; which provide richer presentation environments. What tends to get discussed here are blogs and wikis; RSS; social networking; crowdsourcing of content; websites made programmable through web services and simple APIs; simple service composition environments; Ajax, flex, silverlight; and so on.
concentration is a major characteristic of our network experience, which often involves major gravitational hubs (google, amazon, flickr, facebook, propertyfinder.com). These concentrate data, users (as providers and consumers), and communications and computational capacity. They build value by collaboratively sourcing the creation of powerful data assets with their users. The value grows with the reinforcing property of network effects: the more people who participate, the more valuable they become. And opening up these platforms through web services creates more network effects. These sites also mobilize usage data to reflexively adapt their services, to better target particular users or to identify design directions. Of course, these platforms are very closely controlled, and there is an interesting balance of interests between openness and control at various levels in how they manage resources (see for example my discussion of the Amazon and Google APIs).
Interestingly, if you trace Tim O'Reilly's writings on Web 2.0 since the publication of his major defining article you see an emphasis on what I have called 'concentration' come through. (See my note on an interview with Tim O'Reilly by David Weinberger, on which I draw above, and also see O'Reilly blog posts here and here.)
Now, of course 'concentration' and 'diffusion' are often complementary approaches. The major Internet hubs 'diffuse' their benefits through service and data syndication, apis, participation, etc, but their value often derives from successfully driving network effects through wide participation and consolidation of data. In fact, many of the 'diffusion' techniques work best when associated with concentrating applications. Think of tagging for example. People have incentives to tag their resources in Flickr or Librarything in ways that may not obtain in the library catalog. Scale matters in the context of the social value created in these services (of course, in these examples, folks are also tagging their own resources). You cannot simply add social networking to a site and expect it to work well. Think of all those empty forums.
Much of the library discussion of Web 2.0 is about 'diffusion', about a set of techniques for richer interaction. It is appropriate that libraries should offer an experience that is continuous with how people experience the web.
However, there is a very important way in which the library experience is not continuous with the web. It remains fragmented: it does not have the characteristics of the concentrating, gravitational hubs which characterize so much web use, and are so much a part of O'Reilly's Web 2.0. Fragmented by database boundary, by service boundary (e.g. connecting a discovery experience gracefully to a fulfillment experience through resolution), by library boundary. We are now familiar with the comparison between this fragmented experience and discovery on the web. And we are also familiar with discussion of how the library presence is weakly represented in the major network presences.
However, think also of the library management environment. Think for example of places where data needs to be concentrated to create value: aggregating user data across sites (e.g. counter data), or aggregating user created data (tags, reviews), or aggregating transactions (e.g. circulations, resolver clickthroughs). Motivations here are to drive business intelligence which allows services to be refined (e.g. how does my database usage compare to that of my peer group), to develop targeted services (people who like this, also liked that), to improve local services (e.g. add tags or reviews). These are examples where scale matters, where data may need to be concentrated above the individual library level.
And, we are seeing for fee services emerge which address this need. LibraryThing, for example, syndicates its user-generated tagging to libraries. I am not sure that ScholarlyStats provides a service which compares usage across libraries; it would be interesting to know if there were demand for such a thing.
This then touches on larger questions about sourcing decisions (in what combination of local, collaborative, and third party do libraries acquire their service capacities) and about concentration of library presence (in what combination of library or library and third party are services offered).
For example, I discussed Georgia Pines and OhioLink the other day as examples of groups of libraries collaboratively sourcing a concentrated library presence which increases their gravitational pull.
And libraries are beginning to think more seriously about sourcing services with central web presences. Think for example of the decisions made by the National Library of Australia and the Library of Congress when they chose to use Flickr for significant image projects. NLA is seeking to expand the coverage of PictureAustralia; LC is seeking to collect tags from viewers. In each case, the library wants to benefit from the concentration of users and data that Flickr has created on the web. And to suggest another example, Andy Powell has been raising some intriguing questions about how repository services should be sourced in ways that, again, map onto peoples' experience of the web: would a consolidated network level service be more motivating than a serious of institutional presences? (see here and here). Social networking or other services, he suggests, might flourish at this network level in ways that are not feasible at the institutional level.
When we discuss Web 2.0, there is a temptation to think about blogs and wikis, RSS and a Facebook application, and to stop there. There is also some useful thinking about how to expose web services or data in ways that they can be remixed into other applications. However, Web 2.0 is also about concentration, concentration of data, of users and of communications. We need also to think about how libraries reconfigure services in an environment of network level gravitational hubs, driven by network effects. This will involve greater concentration of library resources in various ways, and also - probably? - greater reliance on other web presences to deliver their services.