# Automate the Thing You Hate

> Pick the task you dread over the most frequent one. Dread is a decent proxy for risk.

Source: https://christwentyfour.com/blog/automate-the-thing-you-hate

The usual advice is to automate the most frequent task. I automate the one I dread. It has been the better guide.

## Dread points at risk

Dreaded tasks are fiddly, consequential, touch several systems, and happen rarely enough that nobody is fluent. That mix is how incidents start. Frequency-based sorting points the other way, toward tasks people already do well.

## Motivation is a real input

Automation loses to feature work on urgency. If you pick the task you resent, you will still care on the third afternoon, when the edge cases show up.

## Build it in three stages

1. A script that gathers the information and prints what it would do. Most of the pain goes away here, because the dread is usually about assembling context. It is safe to run any time.
2. The script acts, but asks first. You watch it be right in production, and you learn where it is wrong.
3. Once the confirmations are rubber stamps for a couple of months, remove the prompt and schedule it.

## Write for a tired reader at 3am

Say what the script is about to do, what it decided and why, and how to undo it. Then delete the runbook. Two working paths means you maintain both.