I am in early planning for a project that will both involve heavy knowledge building in the application, and also export to experimental presentations, some graphical, some experimental hypertext. Because much work will be done in-app on a desktop, and require ‘being in the zone’ of the subject at hand, I want it to be delightful, where key annotative information is clean, clear, elegant. I’ll be using named links extensively and likely have lots of information in Text that breaks the flow.
The last time we did something like this, I used rich text. I’ve been working with written documents with complex ideas for too long to leave good old text with fine fonts and layout. That is to say that I find markdown/asciidoc/MIF just frikken ugly and in the way of the ideas. I know I will get responses from folks who use MD like breathing and can ignore the inline metacharacters placed there to help the real reader. But that’s not me.
My choices are:
— stick with my cranky boomer notion and use styled text in clever ways to capture the annotative information, and possibly hiding local interpretive information in attributes, plus use stamps and highlighters and other tricks. This could involve different faces, styles, even bespoke fonts. At some point this becomes lumpy and itself obscures real reading.
— go with some MD-like convention, taking advantage of the community’s skills. That is, embrace, extend, and leverage the junk… and somehow ‘hide’ it or make it less prominent.
— do something radical, on top of all the other radical things we are attempting, maybe hiding annotative information at the end of named links, maybe even in notes that execute external code internally, like the JS map items.
Much depends on what my helpers have the patience to support for my peccadillo, but much will be governed by the publishing strategy (and other export to code), and whether we rely on Export Code, one of the clever Pandoc or similar pipelines, some bespoke exporter, or a combination.
In anticipation of the expected ‘what are you trying to accomplish?
-
The primary rationale for Tinderbox is as a knowledge enrichment environment, the best on the planet for this requirement, with a lot of very clever folks engaged. I’ll spend a ton of money scouring the scientific literature and building a pipeline to get neurobiological information fed in. I want to use this to understand what is known well enough to refine and structure that in a formal hyperstructured model that reveals what is not known. I’ll have a few experts on board as well, so collaboration is a separate problem. We should be able to browse and enrich what’s there as if it were a superwikipedia. So I can’t be stumbling over the noodlemaker when what I want is to eat.
-
All of that is to create both: a) a human-readable model that can be exported for experimental browsing on the web, with some display experiments long in incubation. b) a machine readable model to feed a reasoner that our startup is building, now at MVP, for reasoning over unknowns and across domains.
These will conflict. Wisdom humbly requested.