This morning’s highlight was Stu Weibel’s opening keynote address to the assembled conference attendees (yesterday included primarily workshop sessions and tutorials). Stu was asked to talk about DC’s past and future, and he gave many of us food for thought.

The first thing he did was typical Stu—he took a photo of the assembled group. I hope those will show up on his blog sometime soon (if so I’ll link to them). Will they look that much different from those he took 15 years ago, when Dublin Core began, or those taken at various points along the way? As is also typical, he asked how many people were returning participants in DC, and how many were brand new to the DC conference. Surprising to some of us, roughly one third of the group were new participants, a nicely healthy proportion of new people.

Stu asked some interesting questions, and gave DCMI some letter grades for performance in a number of areas. His first question, “Why didn’t we just stop after the 15 elements” suggests the possibility that nothing done since then (the mid-nineties) was worth the effort. He points out that a number of the assumptions made at that time have since been repudiated by experience—the Web is more than just a collection of document-like-objects that can be described in much the same way that we’ve traditionally described library materials. He got a laugh when he reminded us (including the three of us who were actually there in Mar. 1995 when DC was born) that we thought we could solve the syntax problem fairly easily—but of course our notion of syntax in those days was HTML.
Stu gave some personal assessments of the 15 years of DCMI in the form of grades:

For providing an international basis for the effort (A)
For including a diverse group of participants (B+)
For becoming a sustained and solvent organization (D)
Moving work from consensus to completion (C)
Establishing objectives and completing them (C-)
Documentation of decisions (B+) (I have to say I think he was somewhat over generous on this one, for reasons I’ll explain later)

Stu also pointed out the places DCMI got stuck along the way, among those were ‘tarpits’ (his word) of our own making—for example, data models (for which he gave DCMI an ‘F’). Those that were not our fault or not entirely of our making were things like the aforementioned syntax confusion (C+), which he believes stemmed from trying to do too much and getting overwhelmed (probably the rapid evolution of syntaxes had an effect, too). The ‘tarpits’ created by others included LOM (learning object metadata, promulgated primarily by IEEE), and INDECS (a now dead effort that was touted some years ago as a business model based approach).

Some other points Stu made which are hard to argue include the notion that cooperation with organizations with different business models is difficult, but such cooperation is critically important, for reasons around convergence of effort, identification of similar models and related technologies—all amplifying the network effect of what DCMI has done. Increasingly, cooperation is seen as an expensive value, particularly in terms of time and travel, and certainly DCMI has seen those issues having a big effect on conference planning and ability to build on past efforts. Stu also gave DCMI an ‘A’ for its standardization efforts, including the work to make DC a recognized standard via IETF, NISO, and ISO. He pointed out that these efforts were essential to allow DC to be adopted by government agencies and others that have requirements for such an endorsement.

On the Singapore Framework, Stu is dubious, in particular thinking of the four levels of interoperability [link] that include the Description Set Profile (DSC) and the Dublin Core Abstract Model (DCAM). He pointed out that lots of metadata is still used primarily at the lowest, human level, and is not yet useful to machines at the moment. This issue is particularly relevant given that the DCAM is currently under review, with opinions flying around (in the halls, on twitter, etc.) in ways they didn’t even when that document was new, and pretty much nobody understood it. Understanding is still an issue, to a great extent because specification and user documentation are not the same thing (something that the DCMI technical folks don’t seem to understand). Stu contends that the DCAM has failed, and felt that its authors still don’t agree about its motivations or implications. It will be interesting to see whether that assessment is widely held throughout the conference.

On linked data, Stu was clearly somewhat ambivalent, calling it “An aspirational target of great promise and unproven benefit”. Stu was around and in the fray when the RDF standard was still in diapers, and recalled for the audience how easy we all thought it would be then to bring its promise to fruition. At least a decade later, we’re still trying to do that.

In Stu’s opinion, the Web is the data model, and we shouldn’t deviate from it. He pointed out that with the issues of flexibility vs. constraints we are drawn in by the Siren Call of Flexibility but would be better off with more constraints.
On the positive side, Stu points out that the linked data bulge has brought us a strong commitment to identifiers, some useful conventions about vocabularies and syntax, some tools to build ontologies and models, and some expectations of utility due to broad adoption – network benefits, in other words. But we still need to worry about data quality, usefulness, and bridging the boundaries between the existing semantic communities.

Stu is skeptical of linked data as the new grail, but still thinks we’re on an exciting threshold—we need metadata more than ever, but we’re drowning in it. He quoted Tom Baker: “Data that cannot speak for itself will be more vulnerable to becoming irrelevant”.
Stu’s last points covered DCMI as an ongoing experiment in social engineering. He cited Malcolm Gladwell’s article in the Oct. 4 New Yorker where Gladwell asserts that social media are largely broadly disseminated, networked, weak tie activities, with low barriers, low commitment, low persistence. Gladwell contends that systemic change requires strong ties, hierarchical social structures, leadership organization, and f2f work. Stu believes that DCMI is a strong tie phenomenon, and its impact is amplified by this fact.

Be Sociable, Share!
By Diane Hillmann, October 21, 2010, 4:20 pm (UTC-5)

Add your own comment or set a trackback

Currently 3 comments

  1. Comment by Jonathan Rochkind

    Sorry, Diane, not Jon, fixed in my blog post. I really like that quote attributed to Stu, I agree completely, and am getting a bit frustrated at people trying to convince library catalogers that the only way to do data right is to do it the linked data way, something I don’t think is consensus amongst “computer folks” or is proven.

    We DO need to educate atalogers about how to do data right from a CS/software/”computer folks” perspective. That involves things having clear explicit models of what we’re recording; keeping semantics separate from encoding seperate from presentation; using unique persistent identifiers for entities (what entities? The ones in our model(s)). Etc. are indeed consensus amongst software engineers and computer scientists, and are decades old. (but not in the library metadata community, alas). The (significantly newer) linked-data/RDF/semantic-web approach is not the only way to do that, is in fact a very complicated way to do that that may not be the best way to do it, and I think attempts to ‘educate’ catalogers that the only right way to do data is the linked data way in fact just muddy the waters.

    Looking forward to seeing what you have to say about all this in your code4lib conf keynote, Diane. :)

  2. Comment by Jodi Schneider

    Thanks, Diane. It’s wonderful to have summaries like this one coming out of the conference. Especially with the context and perspective you add!

  3. Pingback by ALCTS RDA presentation « Bibliographic Wilderness

    [...] on this kind of “free floating assertion” — but I think the RDF project is, as Jon Phipps quotes Stu Weibel:  “An aspirational target of great promise and unproven benefit”  (Consisely stating [...]

Add your own comment



Follow comments according to this article through a RSS 2.0 feed