Jerry McDonough, another invited speaker at the ICDAT, kindly commented that my talk sparked 3 new ideas for research projects in his mind. I highlighted what I like to call “parallel descriptive technologies” in libraries, archives and museums – each community now has the claim of having created a complete suite of standards for describing and disseminating content. However, if you look closely, what has really happened is that each community has defined an optimal way of describing one specific type of content (objects of material culture, bibliographic materials, archival collections).
The main argument I advanced was that rather than think about descriptive practice as confined to certain types of institutions, we should think of descriptive practice as guided by the materials at hand. People would think it rather odd if a museum used CDWA / CCO to describe the books in its library, while nobody takes offense if a library uses MARC (or its more XML savvy sidekick MODS) / AACR2 (RAD) to describe objects of material culture. If we want to build more cohesive aggregations of content, I’d submit that libraries, archives and museums will have to agree on the same suite of standards for the same types of materials. The cohesion achieved throught this discipline would also serve users well. It has worked for books – now let’s make it work for objects of material culture.
I also argued that the main sticking point in all of this isn’t data structures such as CDWA or MARC, but data content standards and vocabularies. One reasonable data structure can be mapped to another reasonable data structure, but cross-walks don’t achieve interoperability if the parties involved use different conventions for arriving at data content such as personal names or dates, or different controlled vocabularies to tell them whether the object in question is an “andiron” or a “firedog” (to use a time-honored example remembered from a Murtha Baca talk). And, to come back to Jerry, here’s where one of his research ideas comes into play: he contended that since data content standards such as CCO, AACR2 (RAD) and DA:CS are rules-based, the output created by applying them should be susceptible to computational transformation. An intriguing idea, don’t you think? And now remind me: what were your other two research ideas, Jerry?