Information Repository Systems

(I've taken these notes in response to a conversation we've had on jabber, as I think something less ephemeral than an IM conversation --tychoish)

I think there's a need for some sort of information management tool for humanities academics and people who work with non-numerical data. Basically, there's a lot of information that we'd like to collect that there's no good way to store in a format that would be accessible and useful in a more enduring future.

It's easy to get vague with the term "information," so let me provide some concrete examples.

We collect abstracts to articles, references (citations), links to things on the web, quotes, notes, ideas, fragments of drafts, notes, pdf files, images, sound bites. "Texts" in the largest possible definition of the term.

There are tools that seek to organize information. We talked about Devon Think but both agreed, that while it's nifty we both agreed that it mucked about in our file system in away that wasn't entirely helpful. While the searching functions were useful, the file-system abstraction was less than helpful.

Thus the goals of this project, for whomever would like to undertake it would be:

• Use existing tools and systems.

  Build on top of the file system. Everything should be in one directory without hierarchy. Use file extensions to denote object type. Users should be able to know what's in a file based on the name. Tab completion in bash/zsh should be useful. File names, including extensions should be < 40 characters.

• Reuse Existing tools

  Tools such as git, xapian, grep, emacs, emacs remember mode, ikiwiki, sygn, tumble manager, jekyll, etc. should all work with the tool, because they already exist, are quite powerful, and by using (and wrapping) standard tools we make it possible to spend more time concentrating on creating the best user experience, and in creating the most valuable tool for the users of the software.

• Focus on providing multiple views and perspectives on a given piece of information.

• The capture of information should be minimal touch.

  We need to automate the hell out of the population of objects in our database, so that people don't have to think about the future selves that will be accessing this data.

• Avoid building a monolithic tool.

• Target a specific kind of researcher and thinker, rather than a everyone. If the tool works well enough it may generalize itself (in the way that org-mode has).

• Data should be stored in transparent formats.

And so forth. Now all we have to do is build it...