Treating my ranked climb like an SLO

At work a service doesn't need to be perfect. It needs to meet its objective, and the gap between perfect and that objective is the error budget: room to ship, break things a little, and learn.

I'm trying the same idea on ranked.

The objective

Each ladder on my challenges page has a start and a target, like Gold IV to Diamond IV. Progress is where I am between the two. I update it by hand after each session, and I don't round up.

The error budget

Losing is inside the budget. Playing tilted isn't. The rule I'm testing: two losses in a row ends the session. In SRE terms it's a burn-rate alert. The budget isn't gone yet, but it's burning too fast to keep deploying.

The postmortem

After a bad session I write three lines: what happened, what I'd do differently, and one thing to practise next time. Blameless, mostly.

Is it overkill? Probably. But measuring before deciding is the most useful habit my job has given me, and it turns out it travels.