I recently switched email programs. I had been using Thunderbird, pretty much for a single feature: its client-side Bayesian spam filtering. Now that Laszlo is using a new mail server and we're off of god-awful Exchange, I can run spamassassin against my IMAP account. Spamassassin is awesome, but even better is my new mail client: Pine.

'Course Pine isn't new, although relatively recently it's been turned into a kickass IMAP client and ported to run native on the PC. There a bunch of reasons I really like Pine, but chief among them are:

• It's totally keyboard based.
• It fast to start and fast when it runs.
• It's extremely customizable. (I mean ridiculously so.)

The common theme of these features is that Pine does not attempt to be simple. It's got a bunch of crazy key sequences, a zillion arcane settings, and can be tricky to set up. Now that I have it running though, I'd say that (conservatively) I'm twice as productive with it as I was with Thunderbird. For instance, I find that at least once a day, I need to find a message that I know is from a certain person that contains a certain subject. Pine can do this with ;TF<name>;NTS<subject> -- literally seven keystrokes in addition to the search tokens. By contrast, in Thunderbird this involves opening a little dialog, filling out a form (good luck keyboarding all the way through it -- all of the Mozilla keyboard support is spotty) and then waiting. Yup, can't explain it, but Pine is like 10 billion times faster at sorting and (complex) searching than Thunderbird, even though Thunderbird is the newer software.

The irony of this doesn't escape me. My job is ostensibly to make pretty, consumer friendly software that's a lot more like Thunderbird than like Pine. In fact, combine the choice of Pine with my well known predilection for VIM and the fact that I use cygwin bash rather than Windows as much as possible, and you could draw the reasonable conclusion that I'm some sort of retro software enthusiast.

Most of the people who choose this sort of software (and for once here I'll include the Emacs people) tend to explain this in terms of the mouse. While it's true that I try to avoid the mouse as much as possible, I think there's something deeper here: it's no accident that I prefer a bunch of software programs that were initially developed over twenty years ago.

When Pine and VI were first created, most software users were also experts who themselves wrote software. As such, they created software for themselves: software that could be finicky, hard to set up, and painfully over-customizable, but software that ultimately did what the experts who used it every day wanted it to do.

If you think about it, this is pretty different from the design goals of most current software projects. The questions that most usability testing asks are: where would my grandmother find the spell checker? How would my blind, mentally challenged neighbor know where to click? It's all about "software for the rest of us" but the funny thing is that there's an increasing number of us who don't lag; we lead. What software are we supposed to use?

We should make sofware that challenges us to be more thoughtful, more skillful, and more productive. I'm asking for more than the bicycle/tricycle thing here: I'm asking for software that gives us new ways of working; that's as smart as we are; that leads instead of follows.