Beyond Digital Incunabula

Like Gregory Crane mentioned in the article assigned as part of last week's reading entitled "What Do You Do with a Million Books?," digital libraries are limited by preconceived notions based on the paradigm of print libraries. Our digital libraries, he says, "remain filled with digital incunabula." In the article "Beyond Digital Incunabula: Modeling the Next Generation of Digital Libraries," Crane and his colleagues explore these "incunabular assumptions," and use the Perseus Digital Library as an example of the ways in which digital libraries can move beyond such limitations.

In the first section of the article, the authors identify three notions that have carried over from paper libraries and have limited the design of digital libraries. The first is that scholarly writing is based on well defined forms in which the units of information (the journal article, the book chapter, etc.) are relatively coarse. This has translated into the digital world as PDF or HTML files that mimic their print predecessors in format. Secondly, metadata in a digital library replicates the card catalog of the print library, and thus has all the same constraints. Finally, the static nature of print libraries has carried over to digital libraries -- just as no new knowledge is created by interplay of books in a print library, digital libraries are not being exploited to create new knowledge "by learning from their collections and from their users."

The authors next identify three characteristics that they believe distinguish "post-incunabular" digital libraries from their antecedents. One characteristic is finer granularity. More advanced digital libraries are beginning to break down digital objects into smaller and smaller chunks, making it easier for users to get to the information they are looking for without having to wade through a whole lot of content in which they have no interest. Another characteristic is autonomous learning. The documents in a digital library should constantly be looking for resources with which to enrich themselves, as, for example, by scanning for new secondary sources and new editions of primary source material. The final characteristic of a post-incunabular digital library is decentralized, real-time community contributions, as exemplified by Wikipedia. The article continues, using elements of the Perseus DL to exemplify each of these three characteristics.

The greater consequence of these incunabular assumptions is the continued "hegemony of library, author, and publisher" over the needs of the user. In order to counteract this, digital libraries should offer the user the ability to customize their environment. The user should not just be able to change the page layout and appearance, but should benefit from more substantive forms of customization. For example, in the Perseus DL, the user is able to select the textbook with which he or she learned Latin or Greek, and thereby be shown which words in a given passage are likely to be new or unfamiliar. The user should also benefit from the digital library's ability to analyze his or her behavior and background and on the basis of this offer automatically generated content. On the basis of four initial questions, the Perseus DL is able to predict, based on past question patterns, which of the words that user is most likely to query.

This article is another instance of the conviction that we saw expressed in the readings from last week, namely that the next generation of digital libraries should be greater than the sum of their parts. The idea that digital libraries should not just provide content, but also add value to that content, pervades the literature and the discourse on the subject of the future of digital libraries. We need to find practical ways to make the content of digital libraries useful in a way that print resources are not, and to do this we need to think outside of the paradigm of the print library and to view digital objects not as copies of their print counterparts but as distinct entities with unique properties. The example of the Perseus DL provides concrete examples of the way in which this is actually being accomplished.