Changing a parameter is the weakest way to change a system
💎CORE ARGUMENT
Most of what an organisation calls a decision is the turning of a dial. A date moves, a budget grows, a target goes from ninety to ninety-five. Dials are easy to reach and visible to everyone, which is why we reach for them. They are also the weakest way to change how a system behaves, because the system that produced the old number is still there, and it will produce the new number's problems with the same reliability.
Above the dials sit things that are harder to see and far stronger: who gets what information and when, what the rules reward, and what the whole arrangement is for. Change one of those and the dials tend to move by themselves.
The slipped launch is my example in both directions. Moving the date from 18 February to 11 March was a parameter change, and it was necessary. It bought three weeks. It did not explain why we had been surprised. What would have prevented the surprise was an information flow: a test at full volume, early enough to change the decision. Nobody would have needed to be braver.
After the reorg Greta proposed two lanes on the roadmap, one for each old team, with a ratio between them. A ratio is a dial, and we would have argued about it every quarter. We chose one list with one order, which is a change of goal: it says the merged team is for one thing.
I keep this at developing because dials sometimes do the job. A parameter near a threshold can tip a system, and now and then a date really is the whole problem.
How this note changed
- 26 January 2025: First version, from the essay. All theory, no example.
- 23 March 2025: Added the slip as a dial that was needed and not sufficient.
- 10 August 2025: Added the two-lane proposal after the merged roadmap was published.
🧱BUILT FROM
🌍WHY IT MATTERS
When someone proposes a number, I ask which rung it is on. If the answer is a dial, I ask what would have to be true higher up for the dial to move on its own, and I spend my argument there.
☑️NEXT STEPS
- Sort the retro's five changes by rung
- Name the rung in every proposal I write this quarter
The list of places to intervene, read from the bottom: constants and numbers last, information flows, rules and goals far above them.
💎 The structure of a system explains more than the people in it is why the dial disappoints. 💎 I can listen to a system but I cannot control it is the caution about pushing even at the high rungs.
Parameters adjust the size of flows inside a fixed set of loops. The loops themselves, the information that feeds them and the goal they serve decide how the system responds. Changing a number leaves the response intact.
Dials are the only lever most people are handed. A product manager can move a date on Tuesday; she cannot move a company's goal. A weak lever within reach beats a strong one that is not.
My four-rung compression of the list, and the decision to run the merged team from a single ordered roadmap.