Personal Musings on a Successful NSDL
The NSF’s early digital-library research initiatives [Larsen & Wactlar 2003] yielded very promising results regarding discipline-specific content and/or presentation, as well as generic library functionality (especially various forms of search), but left unanswered, in my view, the following three questions:
- What capabilities—available in digital libraries, either now or in the future—will be fundamentally important for large-scale enhancement of teaching and learning?
- To the extent that these key capabilities can and should be common across (all STEM [8]-related?) digital libraries, how best can such commonality be supported?
- What is the best means for digital libraries, where primary emphasis often has been on Internet-wide accessibility, to build enduring relationships with end users?
I think we neglected these questions early in the NSDL effort so as to implement (utilizing OAI, the Open Archives Initiative) a quick solution to the perceived need for basic interoperability. To some degree, we assumed that the answers would emerge as the NSDL became functional and reflected the varied strengths of its component libraries. In hindsight (and I do not mean to discount NSDL’s OAI achievements) I believe it would have been more effective to approach interoperability with a deeper understanding of the need for it in the educational context implied by questions 1 and 2. Doing so, even if it delayed the initial “release” of NSDL, might have engendered higher levels of cooperation among the NSDL grantees and, more important, might have actually accelerated NSDL use by educators.
A related factor was that my Unidata experience led me to emphasize the role of “community” in the Core Integration effort for NSDL. In retrospect, I did not recognize that: despite significant agreement on the general need for a digital library focused on STEM-education—similar to the kind of agreement that existed among meteorology departments about the need for Unidata in 1983—lack of agreement on specific needs, and the means for meeting them (as might have grown from answers to the three questions), made community-building difficult (more so than in Unidata) and perhaps premature.
Exacerbating this problem was the fact that “community” in Unidata corresponded rather closely to “users,” whereas the basic plan for NSDL construction placed more developers than users at the community table. (This is the primary reason I articulate question three.) Businesses occasionally succeed under the “build it and they will come” model, but a strong focus on end-user needs is typically required earlier than later.
Looking to the future, I suspect that answers to the three questions posed above are within reach. Of course one possible answer to the second question is simply that minimal commonality is required, and that NSF generally should support discipline- and age-specific specialization as the primary keys to educational enhancement. However, I believe that the most educationally transformative digital libraries are likely to have a great deal in common, basing my view on indications of progress such as the following:
- The integration of AAAS “Strand Maps” into DLESE [Sumner et al 2005]—and subsequently into NSDL—sets the stage for new forms of educational library use, where best practices in science learning help shape users’ experiences of browsing and resource aggregation.
- The replacement of the NSDL “union catalog,” which comprised proxies for individual library resources, with an RDF based system that supports characterization of relations among resources, sets the stage for digital libraries that foster understanding about the highly constructive nature of learning and of knowledge creation.
Restating the above, I envision an NSDL that fosters meta-cognition about learning. Libraries have always been places for knowledge construction (think about a student’s or a researcher’s personal collections of 3×5 cards, for example), where one’s mental model is transformed not merely by the absorption of ideas from individual resources, but also by awareness of the contrasts and similarities among such resources, augmented by one’s own notes. In the spirit of the recent OAI-ORE specification [Lagoze et al., 2008], such meta-cognition might well be modeled and encouraged via “aggregations.” Indeed, educational constructivism might serve as the basis for an NSDL “hedgehog” concept.
I feel certain that suitable infrastructure—extending NSDL’s central core, such as by embedding capabilities to exploit the ORE standard—will be crucial for realizing this transformative power on large scales, especially if it can be realized in ways that truly complement and utilize the strengths of legacy libraries at the nation’s educational institutions (where strong end-user connections already exist). There are examples of such complementarity at the level of component NSDL libraries, so NSDL writ large may be positioned to capitalize on the pattern, though modification may be required in the K-12 arena.
Finally, I have confidence in present NSDL leaders, and I hope they find this essay constructive. Two of the leadership traits I have highlighted (disciplined thought and disciplined action) are in greater evidence than when I was involved. Further, I think that even partial answers to the three questions posed above may help NSDL gain the focus implied by the other two leadership traits (a persistent core and primary alignment with one of three value disciplines). Underlying my optimism is the hope that the NSF will remain a supportive and patient sponsor, recognizing that major collaborative efforts necessarily travel rough roads in their early stages.
[8] STEM is an acronym for science, technology, engineering and mathematics.