Get the latest tech news

Doing is normally distributed, learning is log-normal

There are few things I think about more than the essays on gwern.net, and there are few with as satisfying a theoretical payout to contemplate in my orb as his essay on “leaky pipelines”, aka log-normal distributions. The skulk: Say you’re working on a Laravel web app. You’re about 90% sure you know how to start the app. You’re 80% sure you know how to handle the infra you’ll need to get it online.

Software estimation is such a mess in part because it has trouble recognizing that, at least, just-in-time learning is at least non-normally distributed. Everything we know about traditional project management, from Waterfall to Gantt charts to estimation practices, are on some level based around the idea that each individual step in the chain is bell-shaped: A process taking twice as long as it normally does might be 2 standard deviations out, aka in the bottom 2.5% of outcomes. Not only do you face the possibility of paying a five- or six-figure sum for someone who’s useless, you have very little ability to estimate as an employer yourself when they will become useful.

Get the Android app