Saturday, September 29, 2012

Thought Experiment: Note-Taking System Design

(Caveat: this is just me thinking out loud.)

I've seen people do really remarkable things with AI-assisted note-taking software, and I bought one of these because it's just a beautiful, brilliant product concept, but I use Scrivener for screenwriting and general note-taking, and I've realized I could probably replace it quite happily with Markdown, Fountain, and some good shell scripts.

The structure I use in Scrivener is really basic: a project has multiple subdirectories of semi-categorized notes, and two subdirectories containing screenplay fragments -- the subdirectory full of files I'm going to keep, in order, and the "maybe later" subdirectory, which serves the same purpose as the Trash in OS X or acts_as_paranoid in an old-school Rails app, giving me the ability to retrieve screenplay fragments I've "thrown away" if I later discover new uses for them. The fragments I like get processed into one big-ass PDF file, aka a screenplay.

The only part of Scrivener which I can't replicate by hand in Unix and which I care about remotely is Scrivener's corkboard UI mode, which allows you to mimic physically moving index cards around on a board. I don't actually use this, ever, but I can see the usefulness in it. In every other respect, Scrivener feels like good workflow design tied to middling code and a bloated UI, so I probably will indeed replace it in my workflow with Markdown, Fountain, and a bunch of scripts.

(As an aside, I've written several very bad screenplays, and one which I feel is actually worthwhile. I think more worthwhile ones await me in my future, because I wrote the worthwhile one using a new approach, which, in addition to solving the "my screenplays mostly suck" problem, also brought me away from the terrible "work like a banshee for a few weeks and then collapse from exhaustion" workflow that I've used in creating all my other, crappier scripts.)

I think overall the most sensible model for note-taking is simply to set up local installations of MediaWiki, the software powers Wikipedia, but it seems like it might be too much hassle.