Flickr Commons and the Smithsonian

The situation covered in this article was pretty intriguing: based on the Library of Congress' success in putting some of their photographs on Flickr Commons, the Smithsonian decides to do the same. The authors of the article were at the head of the project, and here they recount the success of their efforts.

They provide some background information on their own digitization efforts. Digitization itself is headed by the different museums and research centers comprising the Smithsonian. The Smithsonian Photographic Services handles the requests for the majority of physical photographs, but the authors state they aspire to virtually collect all the photographs dispersed through various parts of the Smithsonian in a single digital repository under a comprehensive, unified metadata scheme. The idea is that such a repository will dramatically enhance the use of their holdings.

The Flickr Commons project was seen as a first step in this direction, in part because it would help indicate how the Smithsonian could present a single face to the public in spite of its highly distinctive, individuated collecting institutions (presenting a unified, impressive front to the public seems to be a recurring perceived benefit of repositories), and in part because the Smithsonian would begin to learn how to normalize its metadata when it began pulling images from its various departments.

Without quoting their statistics, the Smithsonian finds the project a success. Traffic is thorough and is boosted with each new submission of photographs. The institute is researching ways to employ the tags and comments into their own site's search functions and into the object catalog records.

An approach like this to digital curation partly hinges on the nature of the digital objects, which I would describe as humanities objects. This is a simplification (their photographs are data in the technical lineage of photographs, and many depict social and natural phenomenon such that they're "data" too, among other exceptions), but the subject matter lends itself well to the folksonomy Flickr provides. This is similar to the creative texts marked up in the IVANHOE project Jerome McGann discussed. In both cases the objects being described don't suffer from overlapping or contradictory descriptions; in fact it seems the only appropriate way to mark them. Considerably uncontrolled community engagement with the objects is seen as the best way to their digital aspect. Certainly scientific data can support layered metadata that describes different levels of a measurement, but these terms and layers are carefully circumscribed. In the case of more open or ambiguous digital objects markup approaches like IVANHOE and Flickr seem to fit well (although granted that IVANHOE does not appear to receive much traffic these days).

Interesting too is that the Smithsonian essentially gets to have it both ways with Flickr. They have applied machine tags to a group of photos, the Belize Larval Fish Group (it doesn't look like they're displayed to users though). These tags have a simple machine-readable format ("taxonomy:genus = Sphyraena") that other services can harvest. This achieves the kind of controlled specificity typical in scientific datasets, but with the same tool (Flickr).

What's challenging about digital humanities objects is that while traditional tools like XML or Flickr's machine tags will work well for storage and access, actual value-adding markup depends upon a different kind of tool that relies on a lot more human agency. I think these tools are still being developed as it's difficult to apprehend what's desired. Mass computation on Big Science datasets one could argue is a difference in degree; digital tools for humanities might be a difference in kind.