Law 47's Judgment, last line:
*"There is no substitute for strategy and careful planning. Set a goal, and when you reach it, stop."*1
Read it twice and the tenses come apart.
Set a goal — past. Before any of this started. When you reach it, stop — present. In the middle, hot, winning.
Those are two different people. The one who set the goal didn't know it would work. The one being asked to stop has just been shown that it does.
⚑ And nothing available to the second person could produce the instruction. In that moment every signal argues for continuing — the resistance is collapsing, one more push looks cheap against what it might return, and the estimate you'd use to check that is the estimate the winning has distorted.
So the sentence isn't self-control advice wearing a plan. It's a pre-commitment device wearing the costume of self-control — and for anyone who started without naming a stopping point, it supplies nothing at all. You can't pass a mark that was never placed.
[RESONANCE] I've been treating "know when to stop" as a quality of character. Some people have it. I mostly don't, and I've been vaguely intending to acquire it, in the way you vaguely intend to become a person who reads before bed.
This says the quality doesn't exist. There's no faculty to develop. What there is, is a number written down by a version of you who was not yet winning, and who therefore had access to a piece of information — what would actually be enough — that gets destroyed by success and does not come back.
That's a relief and it's also a much sharper obligation. The work isn't in the moment. It's earlier, when nothing's happening, when writing down what you'd be happy with feels premature and slightly embarrassing.
⚑ The version of me who could answer that question honestly only exists before the thing starts working, and I have been letting them go by without asking.
The vault's Law 47 page carries this — Do Not Go Past the Mark You Aimed For — but the interesting part is what happens when you set it against the other stopping instruction in the same chapter, which is a completely different animal.
Wellman, on cross-examination, 1913: "be ever on the alert for a good place to stop."2 You cannot know in advance which question will break the witness. There's no pre-set mark. You have to detect the peak while performing and act on it before the next sentence — already loaded — comes out of your mouth. See Always Stop With a Victory.
Two remedies, both called stopping at your mark, and they have almost nothing in common:
Pre-commitment protects against the campaign that runs three months too long. It's cheap, it's done in advance, and it works precisely because the deciding self is absent when it fires.
Live detection protects against the eleven extra questions. It's expensive, it happens in real time, and it requires a monitoring process running alongside the performance using the same attention the performance is consuming.
And here's the thing worth keeping: most people are decisively better at one than the other, and the failure modes look identical from outside. Someone who runs a campaign eighteen months past its return and someone who talks for four minutes after making their point are both described as not knowing when to stop. They need opposite interventions. The first needs a number on a piece of paper. The second needs a habit of shutting up, and no number will help them because the moment can't be scheduled.
Nothing in Greene separates these. He states both, four pages apart, under one law.
Named tension: the pre-commitment reading makes stopping a design problem — you solve it before the situation exists, and the in-the-moment self is deliberately given no vote. The live-detection reading makes it a perception problem — you solve it only in the moment, and no amount of advance planning substitutes. Both are in Law 47. They imply different training, different diagnostics, and different people being good at them, and treating them as one virtue is why "know when to stop" has never been actionable.
Promotion criterion: this becomes a real page when I can state the sorting rule — what distinguishes a situation whose stopping point is nameable in advance from one whose isn't. First cut: can you specify the win condition before you begin? Campaigns, negotiations with a defined ask, projects with a deliverable — yes, pre-commit. Conversations, performances, anything where the decisive moment is produced by the other party — no, and the pre-set mark is worse than useless there because it gives false comfort.
That cut needs testing against cases where it's genuinely ambiguous, and I can't think of one yet, which usually means I haven't looked hard enough.
Next action: apply the sorting rule to three things currently running. If any of them has no nameable win condition, that's the one where I've been planning instead of watching — and the planning has been the comfortable substitute.