Someone tells the boss, in a meeting, that the new strategy is brilliant.
Watch what happens to the room. Two people glance at each other. Somebody's mouth does something. The boss says thanks and moves on, and the compliment has cost the person who paid it more than it bought.
Now compare: a week later someone else says, "Can I ask how you got to the third option? I keep landing on the second and I think I'm missing a step."
Nobody discounts that one. The boss talks for four minutes. He enjoys it. And the person who asked has just been flattered at by the person they were flattering — he explained, so now he's invested.
That gap is what this page is about. Greene's claim is that praise has an audience problem, and the fix is to stop delivering praise at all.
"Overt flattery can be effective but has its limits; it is too direct and obvious, and looks bad to other courtiers. Discreet flattery is much more powerful."1
Two separate objections, and the second is the one people miss.
Too direct and obvious — the target can see the mechanism. Praise that announces itself as praise invites the question what does this person want, and once that question is live the compliment is doing the opposite of its job.
Looks bad to other courtiers — and this is the real ceiling. Flattery is delivered in public, which means it has two audiences with opposite reactions. The target may enjoy it. The peers who watched now have something on you, because visible currying is the one court behaviour everyone is licensed to punish.
That's the Preface's two-audience problem, arriving as a concrete technical constraint. Anything you do to please the master is scored a second time by people who lose when you succeed.
Greene gives four moves, and they share a structure: each one implies superiority instead of asserting it.
Seem less intelligent than you are. "If you are more intelligent than your master, for example, seem the opposite: Make him appear more intelligent than you. Act naive. Make it seem that you need his expertise."1
Commit harmless mistakes. "Commit harmless mistakes that will not hurt you in the long run but will give you the chance to ask for his help. Masters adore such requests."1
Tone down the wit. "If you surpass your master in wit, it is okay to play the role of the court jester, but do not make him appear cold and surly by comparison."1 And more generally: don't be "the cloud that blocks his radiance from others."1
Note what none of these is. None of them says a false thing. You genuinely don't understand step three; the idea genuinely did build on something he said; you genuinely are asking. Greene's flattery is not lying — it's selective display of your own limitations, which is why it survives scrutiny that a compliment wouldn't.
The highest-leverage move gets one sentence:
"If your ideas are more creative than your master's, ascribe them to him, in as public a manner as possible. Make it clear that your advice is merely an echo of his advice."1
In as public a manner as possible is the operative phrase, and it inverts the usual logic of flattery. Every other technique here works by being invisible. This one works by being maximally visible — because the thing being transferred is credit, and credit only exists in front of witnesses.
It also costs more than the others, and the cost is cumulative. Tone down your wit for a year and you've lost nothing permanent. Give away authorship for a year and there is now a documented record, in other people's memories, of whose ideas those were. You cannot later reclaim them without looking like you're stealing.
Which sets up the only real constraint on the technique: it works best exactly when you least need it — early, when you have no reputation to spend — and becomes expensive precisely as your ideas start to be worth attributing.
The second technique is the strangest and worth isolating, because it's the only one that requires you to actively degrade your own output.
"Commit harmless mistakes… Masters adore such requests."1
The reasoning is that a master who never gets to help you has no way to discharge the role, and Greene adds the sting: "A master who cannot bestow on you the gifts of his experience may direct rancor and ill will at you instead."1
So the model is that being helped is a need the superior has, not a favour they do you. Flawless work starves them of it. The subordinate who never asks anything is not admirably self-sufficient; they're withholding.
That's a genuinely useful reframe, and it's also where the technique turns. There is a real version — bring the problem you actually have, early, before you've solved it — and a manufactured version, which is engineering an error so someone can correct it. The first is just working with people. The second requires you to submit work you know is worse, which costs the organisation something real, and Greene doesn't price that at all.
Compare the two moves from the target's side, because that's where the asymmetry lives.
Praise gives the master a completed judgement. He is the object. His only available responses are to accept it (slightly awkward), deflect it (also awkward), or enjoy it visibly (costly in front of others). And he has done nothing — the transaction was entirely yours, which is why it reads as a gift, and gifts from below invite the question of what's being bought.
A request for expertise gives him a task. He becomes the subject. He talks, he demonstrates, he chooses what to emphasise. And critically he now has effort in the exchange — he has spent four minutes of his own thinking on your problem, which makes your problem partly his.
The second is more powerful for a reason Greene names but doesn't explain: the master "adores such requests"1 because they let him perform superiority rather than merely receive an assertion of it. Performed superiority is self-authenticating. Asserted superiority depends on the assertor, and can therefore be doubted.
There's also the audience difference. Praise is legible to the room as currying. A question is legible as work. The peers who would punish the first have nothing to punish in the second — you asked a colleague for help, which is the most defensible thing anyone does in an office.
So the technique's real advantage isn't subtlety. It's that it converts a transaction the room can score into one the room cannot see.
You need your director invested in something before Thursday.
Kill the instinct to open with how good their framing was. That's the overt version, everyone will hear it, and it buys you a nod.
Instead find the genuine gap — there will be one, because there always is. Not a fake one. The place where you actually can't tell which way to go, or where their context beats yours. Then take that, and only that, to them.
"I've got two ways to structure this and I keep going round in circles. You dealt with the regional thing last year — does that change which one I should pick?"
Watch what happens. They will not give you a one-line answer. They'll give you the history, the thing that went wrong last time, the person to avoid. Somewhere in minute three it stops being your project.
Then, in the Thursday meeting, the sentence that does the work: "This follows the approach Priya laid out for the regional rollout." True, specific, and said in front of people.
Now the honesty check, because this technique is one step from something worse. Ask: would I have asked this question if I didn't need them invested? If yes, you've just done good collaborative work and the political benefit is a side effect. If no — if you manufactured the gap, or asked about something you'd already settled — you've spent their afternoon on theatre, and you should notice that you did.
The tell that you've drifted: you stop listening to the answers. If you can't remember what they said by Friday, you weren't asking.
The Preface argues that deception runs continuously from tact to the con, and this law's techniques land at different points on it.
Ascribing your idea to the boss when it genuinely built on his remark is near the tact end — a true framing, chosen. Asking for help you actually need is not deception at all.
Manufacturing a mistake so he can correct it is a different thing. Nobody is lied to in words, but the work has been deliberately made worse to produce a feeling, and the target would not endorse it if he could see it. That's the consent test the Preface never applies, and this technique fails it.
Greene runs all four together as one method, which is the same move the Preface makes with the word "deception" — legitimacy earned at the harmless end, spent at the other.
Strongest evidence. The two-audience constraint is real and load-bearing: praise is public, and the peers who watch have a licence to punish visible currying that they don't have for a question. That alone explains the technique's advantage without needing any claim about the master's psychology.
Tension 1 — the manufactured mistake has an unpriced cost. Greene treats deliberately submitting worse work as a free move. It isn't; someone downstream absorbs it.
Tension 2 — the techniques aren't morally equivalent and are presented as if they were. Asking for real help and engineering a fake error differ on consent, and Greene runs them as one list.
Tension 3 — it's unfalsifiable from inside. If you dim yourself and things go well, that reads as the technique working. There's no observation available to the practitioner that shows the dimming was unnecessary.
🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY — no case, no citation; this is Greene's assertion. All claims [POPULAR SOURCE].
Open questions. Does the discreet version work on someone who has also read this book — or does shared knowledge of the technique collapse it back into the overt kind? And is there any way to distinguish, from inside, a career of skilful deference from one of unnecessary self-diminishment?
Greene teaches the same operation twice with the ethical framing inverted.
Here it's craft. The subordinate manages a superior's insecurity, and the tone is admiring — Galileo did it beautifully, Fouquet failed to.
In the Laws of Human Nature material in this vault on flatterers and on the group's courts, the identical behaviour appears as a type to detect. The flatterer is one of the recognisable characters in any court, and the reader is taught to spot them and discount what they say.
So the corpus contains both the instruction manual and the detection guide for the same technique, and Greene never notices that his own later book is written from the position of the target of his earlier one. If the detection guide works, the technique degrades — which is the unexamined problem in teaching influence to a mass audience.
Handshake 1 — psychology: Manipulation Technique — Flattery and Charm
The same behaviour appears in this vault twice, once as courtiership and once as a coercive-control technique, and the pair is uncomfortable in a useful way.
In the coercive-control literature, flattery and charm are catalogued as grooming — selective attention and admiration used to build a dependency the target doesn't perceive as one, precisely because it feels like being valued rather than being worked on.
The mechanics are not merely similar; they're identical. Give the target a feeling of being uniquely appreciated. Make the attention feel unearned and therefore precious. Keep the machinery invisible.
What separates them isn't the act. It's whether the target can leave. A director being asked for advice by an ambitious report retains every option they had before — the exchange is asymmetric in power but not in exit. A person being groomed is having their exits closed by the same warmth that feels like care.
The insight neither domain produces alone: flattery's ethics are entirely a function of the target's alternatives, not of the flatterer's sincerity. Greene evaluates by skill — discreet good, overt clumsy. The control literature evaluates by consequence — did this narrow the other person's world? That gives a test Greene doesn't have and badly needs, because his own framework has no way to distinguish managing a boss from working on someone who can't get away. Both are just "the technique" to him.
Handshake 2 — business: Five Listening Tips — Presence, Curiosity, Silence, Understanding
Greene's best technique and the sales corpus's core discipline are the same behaviour, arrived at from opposite motives, and comparing them isolates the one variable that separates them.
Greene: act naive, seem to need his expertise, ask for help. The listening material: presence, genuine curiosity, silence held long enough for the other person to fill it, and understanding as the goal rather than the pose.
Both produce the identical experience in the target — being asked, being listened to, being the one who knows. And both work for the same reason: people invest in what they have explained.
The difference is what happens to the answer. In the listening discipline the answer changes what you do — that's the entire point, and it's why the curiosity has to be real, because a fake question produces an answer you weren't going to use and the target eventually notices which of their advice ever landed.
Greene's version has no such requirement. The question is a device for producing a feeling; whether the reply is useful is incidental.
What the two produce together is a diagnostic sharper than sincerity, which nobody can audit in themselves: track whether the advice you solicit ever visibly changes your work. If it does, the flattery is a by-product of collaboration and it's durable. If a year of asked-for guidance has altered nothing you did, you've been performing, and the target has a year of evidence accumulating whether or not they've articulated it. Greene's technique is safe exactly to the extent that it isn't only a technique.
Sharpest implication. The reason a question beats a compliment isn't subtlety — it's that praise is a transaction the room can score and a request for help isn't. Flattery's ceiling is set by its witnesses, not by its target. Which means the whole technique is an answer to the two-audience problem, and any move that gets the master's regard without giving the peers something to punish will work as well.
Generative questions.
social-approval-research stub.