A great opening keynote from honorary fellow Simon Singh, for his work on documentaries and books such as Fermat's Last Theorem.  Some of the elements that stuck: the importance of storytelling, the importance of trust between the filmmaker and the subject, and the line between simplifying materials for general consumption versus distorting footage to misrepresent opinions.

Later in the day I went to a combined session on singlesourcing and Minimalism. Both represented real efforts to grapple with the empirical side of the subject, measuring user comprehension and performance with variations of cohesion and media. A few lessons here from the single-sourcing session, with Lynn Gattis:

• When single-sourcing at the paragraph level, cohesion matters greatly, and you need to make sure the terms in the paragraphs reinforce each other and the heading they expand upon (not a surprise, and one of the reasons that DITA focuses primarily on reuse at the topic level, where cohesion is less of an issue, since each topic represents a primary unit of reading/use, unlike a paragraph, which is usually understood in the context of surrounding content).
• When assembling content for new purposes, overviews and summaries may need to be rewritten. Again, this fits well with DITA, where overviews and summaries can be separated out into separate topics, which become specific to the collection or map, without affecting the usability or reusability of the common components/topics between the overview and summary.
• Transitional statements were not a primary concern for cohesion/usefulness - consistency of content and terminology was, like the content of the paragraphs reflecting the terms introduced in the heading/title.
• Single-sourcing can't fix bad writing - if you reuse bad content, you've just multiplied your problem.

Still a lot of work to be done here, and one of the things that confused me was the degree to which the study focused on cohesion as a question separately from the question of single-sourcing: while the study established that coherence does matter to readers, it did not establish what kinds of coherence (beyond the paragraph-to-paragraph ones) were at risk with single-sourcing. In casting the net too broadly (getting rid of almost all visual cues for the non-cohesive versions of the document), the specificity of the study to single-sourcing concerns seems at risk.

The minimalism presentation from Henrietta Shirk focused on differences in exploratory behavior with different types of Minimalist information, including exercises, self-directed work, and viewlets. The introductory material on Minimalism was admirably clear and including a bibliography, which is always nice to see, but I was left somewhat confused as to how the behaviors of the users differed, and whether the information was meant to support different kinds of exploratory behavior, or were themselves the subject of exploratory behavior: in other words, the study was clearly placed in context, but I missed the conclusions. That said, the work is in progress, so things may become clearer in a future update.

Quibbles aside, it's great to see a focus on research at the STC, especially on two topics so important to the profession. I'd like to see more of this in the future, and hopefully with practice and attention there will be more such presentations, with increasingly refined theses and results.

Finally, I attended an informal springboard session with Ant Davey where we discussed the possibility of a DITA specialization or set of specializations for research reports, as previously suggested by Ant on this site. It certainly looks like a good match for DITA, potentially with some quite minimal specialization; my first thought was to use a set of structural specializations, one for each of the major types of report content, and then use either a specialized map or a custom document type to set sequence and nesting rules for the overall report. With the new chunking capabilities coming in 1.1, the interoperability within DITA between compound documents and map-based approaches should improve substantially, and I'm less wary of the compound document approach than I might have been in the past.

--MP in Minneapolis