I wrote in my post on the one true system about the informal systems that we use to interface the way we interact with knowledge and information in the "real world" with the way we represent that information on our computers. Exploring these systems lay at the core of the cyborg question, but today's essay [1] is more about how our logic systems adapt as we use computers and as the kinds of information we need to store change and grow.

As near as I can tell there are a few kinds of "systems review" that we tend to do. Theoretically if you develop a system that's flexible and that accounts for all of your future information needs, then you shouldn't have to modify your system very much. Theoretically this is a good thing: better to spend time "doing things," rather than thinking about "how you're going to do things."

The sad truth is that this doesn't work out very well pragmatically: we change our work habits, and our information changes, and our projects change, and our informal logic for interacting with our computers fails to address the problems, and eventually everything spirals out of control. This is pretty abstract, but every time you see someone with hundreds of icons stacked on your (or someone else's) desktop, or you find yourself with hundreds of unsorted (and unread!) email messages, or you have to hunt through half a dozen places for a PDF you are witnessing the symptoms of a flawed system.

The only way to address this is to review your "systems," and make sure that you capture any problem before information spirals out of control:

  1. Have an overflow bin, but only one overflow bin. This is important, but counter intuitive. By overflow bin, I mean something where unfile-able items are placed. This hopefully alleviates the tension to file away information that hasn't been fully processed, or that doesn't fit into your system, or might be ambiguously filed [2].

  2. Do major reviews of your system only infrequently. By major review, I mean, think about how you use your information, what has worked, and what hasn't, and then use this as a model for revising your system. Don't do it regularly, even if you know that something isn't working. Think of this as something that you do only about twice as often as you get a new computer. As part of this major review process:

    Keep a regular journal when you aren't in the process of updating procedures. Track about what works and what doesn't. Often I've found that I have ideas about how things should change, but the changes aren't the kind of thing that I could reasonably change during normal work. These insights/problems are useful, eventually even if they aren't always immediately relevant. So record them for later.

  3. Do Minor reviews regularly. Look in the "overflow bin" from item one and see what's falling through the cracks, file things that do need to be filed. The GTD folks call this a "weekly review," and while GTD-task processing is only part of "the system." [3] It depends on what kind of information you're managing, but staying on top of and in touch with your "stuff" is important.

  4. Be sure to "touch" your information regularly. While I'm in favor of keeping information even when it's not apparently useful (you never know), I also agree with the idea that information is only really useful if you use it. I've often found myself falling into the trap where I'll stockpile stuff "to read later," which of course rarely happens. Avoid this and browse from time to time.

I mean in the end, I'm just a guy who writes more than he should, and has a pile of digital information that's probably a bit too big, but this is how I do things, and I think the lessons I've learned (and continue to learn) may be helpful to some of you. Reviewing and thinking about systems before hand is, if nothing else, instructive.

Onward and Upward!

[1]I use the word "essay" under the terms of its slightly less common meaning "to make an attempt," rather than the terms that describe a genre of writing.
[2]Borrowing from the Python programing language motto, somewhat, "Every bit of information in your system should have one, and ideally only one, obvious location." Now of course, we can categorize information on many different axises, so the key isn't to pound data/information it's to build your system around a consistent axis.
[3]The system, for me, represents everything from the way we store bookmarks on line, to notes that we collect as we work, to tasks and other time-sensitive data, to the way that we store resources like PDFs and papers. Though we don't have "one" system for all these things, and we're not likely to revise them all at the same time, on some conceptual level it's all the same thing.