Winston Lord worked for days on a report for Henry Kissinger.
It came back with one notation across the top: "Is this the best you can do?"1
So Lord rewrote it. Polished it. Resubmitted.
Back it came, same question.
He redrafted it a third time. Same question again.
At which point Lord snapped: "Damn it, yes, it's the best I can do."
Kissinger's reply:
"Fine, then I guess I'll read it this time."1
Notice what is absent from that exchange, because the absence is the entire technique.
No feedback. Not one word about what was wrong, what was missing, which section was weak, what a better version would contain.
Kissinger did not read the first three drafts. He says so — "I guess I'll read it this time" — and the admission is the punchline.
He could not have given feedback. He had nothing to give feedback about.
This is a different mechanism from everything else in Law 4, which is why it earns its own page rather than sitting inside the Louis XIV material.
Louis XIV's silence conceals his intentions and extracts information from anxious courtiers. Warhol's silence lets an audience construct profundity. Both are about what the silence causes people to think of the silent party.
This is about what the silence causes someone to do.
The six words contain no information and issue no instruction. What they do is refuse to close the loop, and an unclosed loop on your own work is intolerable in a specific way — you cannot stop, because you have not been told you are finished.
So the writer keeps going, and the going is the point. Kissinger extracted three additional rounds of effort at a cost to himself of six words and zero reading.
Work out why Lord could not simply reply yes on the first pass, because that's the trap and it's beautifully constructed.
Is this the best you can do? is not a question about the document. It is a question about the author, and it is unanswerable in the affirmative by anyone conscientious.
Nobody who has worked for days on something believes it could not be improved. Say yes and you are claiming a ceiling on your own capability that you don't actually believe in — which feels like laziness, and worse, feels like a claim someone might test.
So the honest answer is always no, probably not, and the honest answer commits you to another draft.
And the trap tightens with each round, because each round is evidence. Lord improved it three times. Every improvement proves the previous yes would have been false, which retroactively justifies the question and makes the next one harder to refuse.
The exit is not analytical. Lord doesn't reason his way out — he snaps. "Damn it, yes." The technique ends when the target's patience fails, not when they satisfy it, and Kissinger's timing is exact: the moment the exasperation is genuine is the moment the effort really is exhausted.
He was not measuring the document. He was measuring whether Lord had stopped having more to give, and the reliable signal for that is irritation rather than argument.
Worth holding both, because the case supports each and Greene supplies neither.
As a management technique it is efficient and it works. The report Kissinger eventually read was, by construction, the best version its author could produce. No amount of line-editing by a senior person achieves that; it can only be produced by the author, and only if they are not permitted to stop early.
As a use of a person it is unattractive. Lord spent days on three drafts that were never read, in exchange for a superior who declined to say what he wanted. The efficiency is entirely on Kissinger's side of the ledger — he outsourced the whole cost of quality assurance to someone who was not told that was happening.
And the asymmetry has a further cost the anecdote doesn't reach: Lord learned nothing transferable. He now knows that Kissinger will push back three times. He does not know what makes a report good, because that was never communicated. The technique extracts maximum effort and teaches nothing, which means it must be repeated on every document forever.
A note on evidence, because it stands out in this corpus.
Greene's marginal quotations are typically undated, unattributed, or credited to an anthology. This one is credited to Walter Isaacson, Kissinger, 19921 — a named biographer, a named subject, a dated book, and a named third party (Winston Lord) who was a real State Department official and could in principle be asked.
That is a checkable citation, and in a corpus carrying a standing 🚩 SINGLE SOURCE flag it deserves saying out loud. It does not make the anecdote true — Isaacson calls it "oft-told," which is a hedge — but it is the difference between a claim with a provenance and a claim without one.
Someone hands you work and you can tell it's a first pass.
The Kissinger move is available and you should understand what it costs before using it.
If you send it back with is this the best you can do?, you will get a better document and a worse relationship, and you will get the better document every time forever, because you have taught nothing.
The version that keeps the effect and drops the cost is one extra sentence. Not feedback on the content — that defeats the purpose, because supplying the standard means they meet your standard rather than finding their own ceiling. One sentence on the dimension:
Take another pass — I think the argument is there and the evidence isn't yet.
That still refuses to close the loop. It still requires them to find the improvement themselves. But it points, and pointing is what converts an extraction into training.
And from the receiving end, if this is being run on you: the honest response is not to keep drafting. It is to ask the question that forces the loop closed. What would a version you'd approve look like? If no answer comes back, you have learned that the reviewer has no standard in mind and is measuring your stamina — and at that point "yes, this is the best I can do" is both true and the correct move, delivered without the snap.
Lord got there in three rounds. He could have got there in one.
Strongest support. A named source, a named participant, a dated biography — the best-sourced marginal quotation encountered in Cluster B — and a mechanism that is fully legible from the exchange itself without needing Greene's interpretation.
Tension — it doesn't illustrate Law 4. Law 4 is Always Say Less Than Necessary, and its argument is that brevity makes you appear powerful and mysterious. Kissinger's six words are not concealing anything or generating an aura; they are a management instrument for extracting effort. Greene places it in the margin of Law 4 without comment, and the fit is superficial — both involve saying little, for entirely different reasons.
Tension — Isaacson's own hedge. "One oft-told tale" signals a story in circulation rather than a documented event. Greene reproduces it without the hedge's implications.
Tension — Kissinger appears three times in this cluster running three different techniques. Recruiting his would-be kidnappers at Law 2, boring negotiators into submission at Law 3, and this at Law 4. He is one of the book's recurring figures, and the recurrence is not remarked on. Logged for the recurrence structure the PRD is tracking.
🚩 SECONDARY WITHOUT PRIMARY does not apply — the source is named and datable. 🚩 SINGLE SOURCE applies weakly. [POPULAR SOURCE].
Open questions.
Greene's Discreet Versus Overt Flattery — built at Law 1 in this cluster — records Greene's instruction to "commit harmless mistakes that will not hurt you in the long run but will give you the chance to ask for his help," because masters "adore such requests."
Kissinger is the same relationship read from the other chair, and the two together produce something neither states.
Law 1 says the subordinate should manufacture opportunities for the superior to bestow expertise. This case shows a superior who declines to bestow anything at all — and gets more work out of the subordinate precisely by withholding it.
So the same relationship supports two opposite strategies, and both are presented as effective. What resolves it is what each party is optimising for. The Law 1 subordinate wants the superior's investment, because an invested superior protects you. The Kissinger superior wants the subordinate's output, and investment costs him time.
Which means the request-for-help technique works only on a superior who wants to be needed. Against one who wants to be delivered to, it reads as incapacity — and Greene gives no way to tell them apart before you've committed.
→ business · The Single-Piece Feedback Rule
The business page holds that development comes from specific, repeated correction on one thing at a time, and that how someone absorbs a single concrete correction is the real signal about their trainability.
Kissinger's method is the exact photographic negative of that, and the contrast is the useful part.
The single-piece rule supplies a specific correction and observes the response. Kissinger supplies no correction and observes the persistence. One produces improvement in a named dimension; the other produces maximum exertion in an unnamed one.
And they optimise different things. The feedback rule optimises for a person who gets better over time. Kissinger's optimises for a document that is as good as this person can currently make it — which is the right objective when you need the artefact and the wrong one when you need the person.
The insight neither page yields alone: effort and skill are separately extractable, and the techniques for extracting them are mutually exclusive. You cannot simultaneously withhold the standard (which maximises effort by preventing early stopping) and supply it (which builds capability by naming the dimension). Kissinger chose the artefact three times running, and the cost — invisible in the anecdote — is that Lord's fourth report would have needed the same three rounds.
→ creative-practice · The Doctor–Patient Feedback Protocol
The craft page holds that useful feedback requires the recipient to name what they were attempting before the reader responds — you diagnose against the author's stated intent rather than against your own preference.
Kissinger's exchange has no such step, and running the protocol against it locates precisely what makes it corrosive rather than merely demanding.
Under the protocol, is this the best you can do? is not feedback at all, because it references no intent. It cannot be, since Kissinger hadn't read the drafts. It is a response to the author in the absence of any engagement with the work.
And that's the distinction worth having, because high standards and this are easily confused. A demanding reader who has read your work and says not yet is applying a standard. A reader who has not read it and says not yet is applying pressure, and the two feel identical from the receiving end and produce entirely different things.
The insight neither field states alone: you can tell them apart by a single test — ask what the reviewer thinks you were trying to do. A demanding reader can answer. A pressure-applier cannot, and the inability is diagnostic rather than rude to surface. Lord spent three drafts without ever running that test, and the anecdote is told as a lesson about excellence when it is at least as good a lesson about verifying that anyone is home.
Sharpest implication. Kissinger had not read a word of any draft, which means the question was never about the report. It was a stamina measurement wearing the costume of a quality standard, and the two are indistinguishable from inside — which is why Lord ran three rounds before the exasperation that ended it. The available defence is not working harder or arguing; it is a single question that reveals whether the reviewer has engaged with the work at all.
Generative questions.