Three Strategies for Effective Data Oversight

Computers in Libraries, Sep 2009 by Huwe, Terence K

Networked information as we have come to know it ought to be a dream come true for researchers who rely on large data sets. Yet all too often, access barriers form in powerful IT departments; this can stymie progress. IT managers often crave security, and that means centralized control, even to a fault. In contrast, information professionals carry a bias for "fair use" and "giving information away." These core values can be powerful change agents in organizations, particularly when professional cultures collide. When we apply our profession's principles of sharing knowledge, we can change hearts and minds, change the rules of the game in organizations, and perhaps even improve comprehensive user services. Our "potlatch" mentality of giving away what others might hoard means we often take the first step to bridge the straits that divide knowledge resources. The divide between the data lab and the knowledge center is a terrific case study for those who study how organizations can change. Bridging that divide is also a promising opportunity for digital librarians.

If we play our political cards right, the library and the data center can be merged strategically in ways that enhance access and study. Just as easily, a foray into data management may earn us new competitors we don't want or need. Local conditions define strategy, but I can think of three foundations upon which to build a vision for data oversight. To make the most of your situation and the opportunities it presents you with, consider these three steps as you take the leap into the number-cruncher's universe.

'Audit' Your Way to Alliances

When I write this column, I remind myself that not everybody works at sites that are quite as large as the University of CaliforniaBerkeley or as "deep" when it comes to data centers and data creation. In my local situation, the sheer number of players calls out for alliance building. So my first strategy is this: To succeed in building a data center or service, you must form alliances, and the more broad the alliance the better off you will be.

Here's why I'm drawing specific attention to the "Berkeley effect": This place is literally awash in data centers. They come in all sizes and shapes; some are hypersecure, others charge back for survey research, and still more just want to make things easier for their own department. Indeed, having your own data center is a bit of a "vanity license plate" for prominent professors. Deans and department chairs will dangle the promise of a new data center in front of a faculty member who is weighing a very plush offer from Harvard, as a way to "retain" Berkeley's brain trust. Here at the Institute for Research on Labor and Employment, we have added at least four of these centers, at various times, since the turn of the century. Data centers are a big business here, and they are highly decentralized. Yet decentralization can be a good thing - it can leave more space for academic creativity.

Decentralized settings favor alliance building. Even so, every now and then reality comes rushing in. The evolving dreams of successive generations of deans and directors (and even donors) mean that the process of decentralizing, recentralizing, and then decentralizing again often feels like a tidal force. An ever-growing universe of data centers is what is left on the beach of the scholarly landscape, as the organizational tides swell and contract.

This environment is fertile ground for alliance-building efforts by information professionals. Our culture of sharing oversight (as in shared cataloging) and building alliances (as in consortial agreements) has dealt us a winning hand. We know how to collaborate across organizational boundaries. While that may be true to some degree among statisticians, sociologists, lawyers, and economists, somehow I think we have an edge. How should we use it? The answer is simple to state but hard to implement: Form alliances based on mutual need and then play a leadership role.

Library-led alliances can bring order to "creatively chaotic" environments without disturbing the glories and wonders of academic research. When librarians take the initiative, they often can create shared understandings and identify kernels of agreement around which a comprehensive data strategy can grow. The actual steps are very clear cut too. First, perform an "audit" of your user community. Develop a list of every data center you can identify. Next, go meet the overseers of these sites, and spread some seeds of collaboration. Offer to create the wiki, run the listserv, and create the webpage for the effort. Now comes the fun part: Start developing a data plan that gives everybody a little of what they want but gives you, the digital librarian, oversight of the raw data, the value added data, the metadata, and the storage devices. Any questions?

'Share' Oversight but 'Own' the Warehouse

My second recommendation for building a robust data service function is more technical. Once your audit and networking is complete, it's time to take charge of the data itself. Try as I might, I can't convince myself that social scientists are going to ensure that their data sets will be with us in 50 years. We're going to need to step in and save this data ourselves. Ib do so, we need to be up-to-date technically. We also should be able to pull together the necessary storage media to compile the data warehouse.