John Perry (Stanford philosophy prof, and the best lecturer I have ever had) has an insightful and humourous essay on what he calls structured procrastination. The idea is basically that habitual procrastinators don't "do nothing", they just do "a something" that's not supposed to be their top priority. I don't know if this is true for everyone, but this is how I operate. Perry basically says the trick is to do "a something" that's worth doing (e.g. not a something like flipping through irc channels, which is how I tend to procrastinate).

When I first read this essay a couple years ago, I realized the most productive times of my life weren't when I took 14 units and avoided "extracurriculars", it was when I took 21 units and tried to do a million things at once. Sure, I failed to do some of those activities well, and sure I got bad grades in a couple of the classes...but on the whole I got a lot more done. Do you measure your success by what you fail to do or what you do? I think the natural tendency is to be risk averse and say that high success means low failure.

Maybe an organized go-getter type person can get away with that: only start tasks you know you will accomplish. But for structured procrastination to work you have to be willing to do badly at some things so you can shine at others things. Of course, you will piss off organized people by doing this, because they will prefer consistent mediocrity to spotty "brilliance". And unfortunately, I think with systematic measures like grades people do tend to measure success as a lack of failure... Despite claims to the contrary, I'm pretty sure grad schools (and colleges) prefer all As to a grab bag of grades garnered by taking really hard courses.

A good example of structured procrastination at work is gnome-blog. I was feverishly productive for about 4 days while writing gnome-blog. Why? Because it didn't matter, and hence was an easy to work on thing to AVOID working on Storage's NL parser (which is at a very difficult point right now... working through it little by little). Now what I really need to do, is find something more onerous than the NL parser. Because I know that I probably could get the NL parser working with 3 days of productive coding... Maybe I can get the NL parser done while avoiding consolidating my student loans?