The Revelation Principle

Imagine you’re negotiating the sale of a car in a country whose language you don’t speak, so you hire a translator. Here’s the twist: this translator is a shark. You tell them, in your own tongue, exactly how much the car is worth to you and exactly how you’d love to play the negotiation — and the translator does all the lying, bluffing, and strategic footwork on your behalf, in the other language, better than you ever could. You never have to bluff. You never have to shade the truth. You just tell your translator the honest facts and your honest wishes, and they produce whatever cunning performance the situation calls for.

Now ask yourself: sitting across from your own translator, do you have any reason to lie to them? None. They’re on your side, they already do all the strategising, and lying to your own agent only sabotages the very person working for you. So you’re perfectly honest with the translator — even though the translator, out in the world, is as slippery as the game demands.

That translator is the entire idea of this lesson wearing a disguise. It is the intuition behind one of the deepest results in mechanism design — a theorem that takes an impossible-looking search and quietly collapses it down to something a mathematician can actually finish. It’s called the revelation principle, and by the end of this page you’ll see why theorists treat it as the master key to the whole field.

The search problem: infinitely many rulebooks

Back in lesson 1 we defined a mechanism as a set of rules: who moves, what they’re allowed to say or do, and how those choices map to an outcome. So far so tidy. The horror only starts when you ask a designer to find the best mechanism — because the space of possible mechanisms is monstrously, unmanageably large.

Think about how many different rulebooks you could write just to sell one painting. Sealed bids in envelopes. Open outcry where people shout over each other. An English auction where a price ticks upward and bidders drop out. A Dutch auction where a price ticks downward until someone snaps. Multi-round formats. Formats where you submit a whole strategy rather than a number. Formats with side payments, entry fees, reserve prices, weird tie-breaking rituals, and message spaces where bidders send haikus. Each of these is a legitimate mechanism, and there are infinitely many more — the messages players can send are unbounded, so the set of conceivable games has no edge to it.

This is genuinely paralysing. You can’t check infinitely many games one at a time, and you can’t even neatly describe the space you’d be searching. If mechanism design stopped here, it would be a field of clever one-off tricks with no way to prove that any trick was the best available. What the designer desperately needs is a way to shrink the search — some argument that says “you can safely ignore almost all of these rulebooks.” That argument exists. But to state it we first need to isolate one very special, very boring-looking family of mechanisms.

Direct mechanisms and truthful mechanisms

Among all those baroque rulebooks, one family is stripped down to the bone.

A direct mechanism is a mechanism whose only move is: each player reports their private information directly. No paddles, no rounds, no bluffing rituals. If your private information is how much the painting is worth to you, a direct mechanism just asks, “How much is it worth to you?” — you write a number, everyone submits at once, and a fixed rule maps the reported numbers to an outcome (who wins, who pays what). Whatever secret a player holds — their value, their preference ranking, their priority, their type — a direct mechanism simply asks them to state it.

Now, asking people to state their secret is not the same as expecting them to state it honestly. That’s a separate, precious property:

Contrast this with an indirect mechanism, where players never state their type at all — they reveal it only through their actions. In an English auction you don’t announce “the painting is worth $900 to me”; you just keep raising your paddle until the price passes your limit, and your willingness to bid one more time silently leaks information about your value. On eBay, you don’t declare your true maximum out loud in a way anyone can read directly off the screen — the bidding war reveals it move by move. Indirect mechanisms are the flashy, familiar ones. Direct truthful mechanisms are the plain ones where everybody just tells the truth and a formula does the rest.

Why isolate this dull-looking family? Because it’s about to become the entire search space. Hold onto one nagging question: those indirect mechanisms produce real outcomes in the world every day. If we throw them all away and look only at direct truthful mechanisms, aren’t we giving something up? The answer — and it’s the whole point of the lesson — is no.

The revelation principle

Here is the theorem, due to Roger Myerson (and it’s a big part of why he shares a Nobel Prize).

Read that slowly, because it’s astonishing. It says every outcome that any rulebook can achieve in equilibrium — every clever multi-round, paddle-raising, envelope-shuffling contraption — can also be achieved by a plain mechanism where players simply tell the truth. Nothing an indirect mechanism can do is beyond the reach of an honest direct one.

And the proof is exactly our shark translator. Suppose some elaborate indirect mechanism has an equilibrium: in it, a player of type “the painting is worth $900 to me” follows some optimal strategy — maybe that strategy is “raise my paddle until the price hits $740, then stop,” i.e. a strategic shading of their true value. Now build a new, direct mechanism with a trusted intermediary, an honest broker, bolted on:

1. Each player quietly tells the broker their true type — the honest $900.
2. The broker, who knows the original game inside out, plays the original equilibrium strategy on that player’s behalf — it raises the imaginary paddle to $740, executes exactly the bluff the player would have executed themselves.
3. The mechanism produces whatever the original game would have produced.

That folding-in is the revelation principle in a sentence: any outcome reachable by any mechanism is reachable by a truthful direct mechanism, because you can always absorb the players’ strategising into the mechanism itself. And it licenses one glorious simplification:

Without loss of generality, a designer can restrict attention to truthful direct mechanisms.

That phrase — without loss of generality — is the treasure. It means the infinite, unsearchable space of all conceivable rulebooks can be replaced, for the purpose of asking “what outcomes are achievable?”, by the small, well-behaved space of mechanisms where everyone just reports honestly. You lose nothing. Every outcome still on the table.

Why it’s a power tool

The revelation principle is not a design recommendation you hand to a policymaker — it’s the crowbar theorists use to pry open the whole subject. Two jobs it does:

It makes possibility and impossibility results provable. Suppose you want to know whether any mechanism can achieve some goal — say, “always sell to the highest-value bidder while never running a deficit.” Checking every possible rulebook is hopeless. But thanks to the revelation principle, you only have to ask: is there a truthful direct mechanism that does it? If yes, one exists. If you can prove that no truthful direct mechanism can do it, then — because a truthful direct version could always be built from any successful mechanism — no mechanism of any kind can do it either. That’s how the famous impossibility theorems get proved: you show the honest version can’t exist, and the principle extends the verdict to every rulebook at once. (We’ll meet exactly those “you can’t have it all” results in lesson 6.)

It turns the auction zoo into two examples of one animal. Here’s the payoff for everything you learned in lesson 3.

Once you’ve seen the second-price auction as the truthful twin of the first-price auction, the abstract theorem stops feeling abstract. Every messy indirect mechanism out there has a plain honest doppelgänger sitting quietly beside it, reaching the same place — and it’s that honest twin the theorist studies.

Honest caveats: what the principle does NOT promise

This is a subtle theorem, and it’s easy to over-claim. Three things it emphatically does not say:

• It is analytical, not a design recommendation. The revelation principle tells you a truthful direct mechanism exists that matches any given outcome. It does not say that mechanism is a good one to actually deploy. The guaranteed honest version can be fiendishly complex, opaque, or otherwise undesirable — sometimes the messy indirect auction is far nicer to run in a real room. The principle says nothing is lost in principle by studying the truthful versions; it says nothing about which mechanism you should build.
• “Truthful” doesn’t mean “better.” Reproducing an outcome is not improving it. The honest twin reaches the same result, not a superior one — no one gets richer, the pie doesn’t grow. The principle is about the reach of what’s achievable, never about ranking mechanisms by how good their outcomes are.
• It leans on assumptions you must not forget. The construction quietly requires that the designer can commit to the rules (the broker won’t secretly change the mapping after hearing your type), that players actually play the equilibrium the mechanism is built around, and that the mechanism computes the outcome correctly. Break commitment — let players suspect the broker will exploit their honest report — and the reason to tell the truth evaporates. These are the same load-bearing assumptions we’ll interrogate in lesson 6.

When to use it

Reach for the revelation principle whenever you catch yourself asking “could any set of rules possibly achieve X?” — a question that otherwise drowns you in infinitely many rulebooks. The principle hands you a lifeline: instead of searching every conceivable game, search only the truthful direct ones. If an honest mechanism can do X, then X is achievable; if you can prove no honest mechanism can do X, then nothing can. It’s the standard first move in any proof about the limits of what mechanisms can accomplish — the tool that converts “search all games” into “check the honest ones.”

Just keep the caveats bolted on: it tells you what’s possible, never what’s best to build, and it leans on commitment, equilibrium play, and correct computation. Use it to map the frontier of the achievable — then use judgment to pick which real mechanism to actually run.

Next up: lesson 5, Mechanisms Everywhere — we leave the theory bench and go hunting in the wild, where the same design logic quietly runs pollution taxes, deposit-refund schemes, cap-and-trade, and the matching markets that place kidneys with patients and students with schools, often without any price changing hands at all.