Bruce Esrig and "cognitive load" of XML

At CMS 2006, Bruce Esrig of Lucent spoke on the topic of Cognitive Load of XML.

Bruce is a deep thinker. (He's also a charter member of the DITA OASIS TC.) This was a very thoughtful and thought-provoking presentation.

I was ready for a ppt on "how to overcome" this cognitive load, but Bruce makes the opposite case - that this cognitive load makes XML more attractive, both as a way to provide content to our audience that demands highly flexible and highly available information and also to the content developers. XML authoring requires a new combination of conceptual skills and practical expertise in envisioning information design and creating the actual content. He nicely ties these thoughts into chunks, templates, standard content, guidelines, and collaboration.

This goes a long way to explain my own observations with my Lotus team at IBM. Before DITA, we used an in-house "HAT" authoring system, based on templated word processing. We had information types and topics, but not the strict typing and map-driven architecture that DITA and XML provides.

I had fears that with our migration to DITA, that our authors might complain about being forced to follow the strict XML topic types and other structures that DITA and our XML authoring tools enforce. However, the opposite occurred. Our writers and editors to a person quickly became very engaged with the authoring environment and the new challenges. Instead of worrying about things like heading levels, indents, and italic-bold constructs for command strings, they quickly moved on to looking at larger structures and topic relationships, how best to chunk and group for reuse, how to take advantage of conrefs, all the rest.

So, maybe there's something to this notion of "cognitive load" having benefits, and providing part of the explanation for DITA providing "better writing, better content."

Thanks for the favorable press!
 
We found that experienced authors appreciate the structure that is provided in a sufficiently-specific authoring language. New authors find a rich authoring language somewhat daunting, and can come up to speed faster if they have the guidance that a specific language provides.
 
When we revised our internal XML language to make it more general, the experienced authors objected strenuously to the loss of guidance that came with loosening the language. We provided templates that contained the essentials of the old structure, but there was still a strong demand for certain language features.
 
If we had specialization, as the DITA framework does, we could have kept both the new generalized structure and certain specialized structures (whichever ones were compatible with the new generalized structure).
 
In terms of "cognitive load", a simplified argument in favor of DITA is that DITA offers benefits by allowing language designers to be prescriptive about what authors must do.
 
Looking at the situation more closely: If the authoring community is diverse, some requiring generality and some specificity, DITA has a unique ability to address both sets of needs in a single, consistent framework. This is done by building one or more prescriptive languages from simpler building blocks. When this is done, the specificity of each more-restrictive language is founded on a common conceptual base.
 
Bruce Esrig
Information Architect
Lucent Technologies

XML.org Focus Areas: BPEL | DITA | ebXML | IDtrust | OpenDocument | SAML | UBL | UDDI
OASIS sites: OASIS | Cover Pages | XML.org | AMQP | CGM Open | eGov | Emergency | IDtrust | LegalXML | Open CSA | OSLC | WS-I