Feedback loops are how systems learn

The idea

Any system that cannot observe its own output cannot improve.

Phase 23 was about building decision frameworks — structures that help you choose well under uncertainty. But here is a question those frameworks cannot answer on their own: was the decision any good?

An example

You decide to write for 30 minutes every morning. After two weeks, you feel like it is not working — but you have not defined 'working,' have not measured word count, have not tracked which sessions produce usable material, and have not adjusted anything based on results. You are operating in open loop: action without observation. Compare this to a writer who logs time, word count, and a 1-to-5 quality rating after each session, reviews the data weekly, and shifts the routine based on what the numbers reveal. Same 30-minute commitment. One system learns. The other just repeats.

Where it goes wrong

Believing that feedback loops are something you 'already do' because you notice when things go badly. Noticing is not a feedback loop. A feedback loop requires structured observation, explicit comparison against a reference, and a mechanism that channels the result back into future action. Most people have emotional reactions to outcomes, not feedback loops. The gap between 'I felt bad about that meeting' and 'I measured three variables, compared them to my target, and adjusted my preparation process' is the entire distance between open-loop and closed-loop operation.

Make it stick

Take the feedback loop you built in the exercise and formalize it: define the specific output you are measuring, the standard you are comparing against, and the adjustment rule you will follow when the measurement deviates from the standard. Write these three elements down.