A substantial portion of my professional output is writing. I don’t feel like I do nearly enough of it, so to encourage myself to do more, for a few years I kept an internal dashboard which had one number on it: total words shipped to the Internet.

Kevin Kwok encouraged me to share it.

I’m having trouble visualizing that number.

The Harry Potter series has approximately 1,080,000 words in it.

Is that a good metric?

No and yes!

The actual metric I optimize for is the product of the number of software people I’ve helped and the median concrete gain from the help.

The two things that impact this more than anything else are project selection (one essay continues to outperform multi-year stretches of my career) and just getting up and doing the work. I don’t have a repeatable technology for good project selection, but the World of Warcraft player in me knows exactly where my buttons are. Show me a number, I’ll make it go up.

This seems to be, all things equal, a positive number to make go up. Some of the words are aesthetically speaking better than others, some essays are more impactful than others, sometimes I wax loquacious to no purpose, sure, I understand all of those objections, but if you told me whether I wrote 25k words or 200k words next year I’d be able to supply instantly whether it was an effective year or not.

What is included?

Essays on this site, Bits about Money, my previous newsletter, Hacker News comments, tweets, Complex Systems transcripts, etc. There are many varieties of “etc” and this number is a little sensitive to how many of them my script can conveniently capture. (Which makes the 7 significant digits a bit of a lie, but the important part for me is that I can make the number go up.)

Pertinently not included: one-to-one emails (which would add at least a million words here) and anything I don’t own (which includes work for consulting clients or employers).

How frequently is this updated?

Last update: 11/20/2024.

I update this every few months, when I get around to it. I have a script but getting the data sources together takes an hour of work spread out over about a day. (Shakes fist at Twitter archive pipeline.)