Document-level controls to capture and communicate meaning


Semantic authoring is like car restoration on Full Custom Garage. The markup is the frame, and the CSS is the sheet metal. Pound at will. Make the data dance and drive change.

 --Test subject

prodoc provides two primary ways to influence the meaning of information: data structures and visualizations.

The basic process for making sense of new data?

  1. Do existing structures work? If so, adapt, otherwise...

  2. Add markup to give the data structure

  3. Pick meaningful names

  4. Add visual differentiation

  5. Adjust the look and geometry to look for patterns

  6. Rinse and repeat until the data becomes actionable

  7. Add a switch or something to change the appearance in real time, if it helps

Sometimes the process starts by making the data look pretty before creating custom data structures. The ability to incrementally add structures and style rules to the base platform reduces the incremental cost of small projects. Successful experiments get moved from the laboratory document to prodoc.dtd.

In practice, document-level experimentation has been key to the evolution of prodoc. The full case study document is the actual file that started the development of the <music/> element.

Some of the design alternatives were driven by knowledge loss. The old markup didn't make sense. Create new markup that makes sense for current thinking. A few data structures only stabilized after multiple documents with partial solutions were created, sometimes across years. A couple of systems only went into production after the best features of 3 or 4 solutions from different perspectives were integrated.