The Tragedy of the Commons, Revisited (Again)
In 1968, Garrett Hardin published his landmark article, "The Tragedy of the Commons", in Science. Hardin created one of the enduring metaphors of our time to lay out the tension between the long-term good of the group, embodied in the commons, versus the near-term advantage of any single member. In Hardin's example, if each individual pursues his advantage by adding a cow to the herd that grazes on shared meadow, the result will be cumulative over-grazing and degradation of the commons -- and a net loss to the group. The article has had a profound effect upon the environmental movement, but more importantly, Hardin's articulation of the issue in terms of the tension between cooperation and individualism illustrates a fundamental theme, which resonates across the intellectual spectrum from how advanced research is organized to the content of the technology itself. This tension is, moreover, central to the whole notion of a library, which is both cooperative and uniform -- a shared collection and associated technical services that are more or less standardized across institutions -- and user-specific, that is, the public services typically summed up as "reference." But the distinction is neither hard nor fast: how a reference librarian finds material for a user is profoundly affected by how a cataloger describes and classifies information.
The at-times muddy distinction between what is shared and what is not seems to me to be useful notion for understanding research in digital libraries. Elsewhere, I have argued that "shareability" is a defining characteristic of an infrastructure. It is not hard to argue that digital library research is really about applications that sit on top of shared technological infrastructure -- particularly when viewed from the perspective of the end-user. Digital library research, moreover, has in many cases resulted in a suite of independent albeit related technologies, and a casual look at stories in D-Lib can yield a view of friendly but clearly distinct specialties. For example, the technical relationships among scanning (Conway, February 1996), metadata (Weibel, July 1995), and user behavior (Van House et al., February 1996) are not instantly obvious. Yet results from one have implications for the others: how people look for information affects what is selected for digitizing and how it is described.
Clearly, there is much that is specific to the many disciplines and subdisciplines -- computing, linguistics, engineering, information science, cognitive psychology -- that converge in digital libraries. However, and as discussed by Dan Schutzer this month, there is much that is shared: naming; network connectivity and efficiency; privacy, security, and encryption; to name just a few issues. Solutions to many of the concerns central to digital libraries may well occur at least partially within the infrastructure. For example, Carl Lagoze and his colleagues have outlined a way to manage intellectual property through the infrastructure and the applications: "terms and conditions" are attached to the architecture of repositories, the digital objects they contain, and the disseminations of those objects. This repository architecture then sits on an infrastructure that authenticates the parties. The ordinary end-user -- a physicist on the track of pre-prints, perhaps, or an Austen scholar looking for the BBC interpretation of Pride and Prejudice in video -- only sees the terms and conditions that apply to the object sought; all of the rest is middleware. But a systems manager, vested with the responsibility of maintaining and updating a set of repositories, perceives both applications and infrastructure differently.
The point is not merely that advances in digital libraries require advances in the information infrastructure or that one user's infrastructure is another user's application. Rather, aspects of digital libraries themselves will also become part of other dimensions of the information infrastructure and the services it will support. This is true not merely at the technical level of repositories, digital objects, and interfaces but also among the services offered end-users. As illustrated in Schutzer's story, storage and retrieval are fundamental operations in electronic commerce, and both are fundamental to both analog and digital libraries. Thus, libraries themselves become part of the infrastructure as well as sets of services offered end-users, and the notion of "special libraries", which now denotes a set of resources appropriate for a specialized user community, may likely become a functionally specialized collection of data and operations that support another level of end-user applications. After all, end-users who want to know their account balances probably do not care how the data are recovered from an underlying collection -- although they care very much that the information is reliable and confidential.
Hardin foresaw a tragedy in part because the commons is ultimately a finite resource. But the economics of advanced technology does not always obey the models of classical economics, which have dealt well with fixed resources, scarcity, and competition. Indeed, an over-abundance of information resources is motivating much of the research in digital libraries and suggests that scarcity is not a concern -- although it has been argued that human attention, or "mind share", is finite and hotly-contested, and the resources to manage information are definitely limited. Still, the tension captured in Hardin's powerful analogy between what is individual and what is shared remains a useful one, and reminds us that recognizing the commons sometimes depends on who is leading the cow.