One True System

So I've been posting about cyborg systems on tychoish, about the informal logical systems we use to interface our lives/reality/thoughts/work into digital systems to organize what we accomplish with our computers. It's a topic of some interesting to me, and I'm going to provide a simple piece of advice in this post:

Try your damnedest to only use one system. For as much of your data as you possibly can.

This might be an odd piece of advice, you say, coming from someone who proudly supports the Unix philosophy of "use tools that do one thing well." I think, however, that the question of tools and systems are fundamentally different. Having tools that do lots of things (poorly) means that you have ineffective tools. Having more than one system to organize your data means that things get lost.

Joseph Spiros, a friend of mine from way back, wrote an essay that I think serves as a pretty good introduction to some of these ideas. Read Prelude to Haven and come back when you're done.


Great. Lets continue.

Having one system that houses everything is a great boon to our representation. If you know, from the very beginning, what kind of data you're going to be dealing with, and develop some sort of organizational system based on this knowledge then it's really hard for files to get lost, for things to be double classified and without reference. Basically, for any given thing, there should be only one given place that you should have to look for this.

This is of course pretty difficult to do. So maybe I don't mean "only have one system for everything" but, rather "have one system for any given thing." More concretely:

  • Only have one email account. If you get mail at more than one address, use a client that allows you to view/send email from more than one address (eg. gmail) or forward multiple accounts to one address. This way, if you're looking for an email there's only one bucket to look in.
  • Organize data by either projects, subjects, or kind, but not more than one of these categories. Projects would be spheres of your work that form a body onto themselves: if you wrote books, a project would be a book and you'd collect notes, drafts, and versions related to a book in one "pile" [1]. Subjects can get dicey (as they require you to sort your data into a given number of subject-based piles and then be able to recall that sorting again. Kind-based organizations require you to keep all your notes in one pile, drafts in another, final copies in another which can grow unwieldy depending on subject, but greatly decreases the chance of misfiling.
  • For data manipulation standardize your practices on specific tools: tasks go in one place, project files go in another, references in another, and so forth. If you track two todo-lists that don't synchronize with each other, then the chance of loosing track on some things in one when they should have gone in the other system. One "OK" system is superior to 2 or more excellent systems.
  • Use search tools to your advantage, but avoid the google-method of relying search to be smarter than the index of files. Search is good, and there are a lot of times when searching for something is helpful (lost files, finding a specific quote), but often times they take a lot of system resources to build a gross index that likely doesn't contain the kind of information your looking for.

And that's it. One more step on the road to a better, more fulfilling cyborg experience.

[1]I've deliberately avoided using terms like "folders," or "categories," or "tags," as these abstractions aren't in and of themselves useful metaphors.
comments powered by Disqus