Game Theory · GT-37
Instead of analyzing a given game, design the rules of the game itself so that self-interested players' best responses produce the outcome you actually want.
Sometimes called 'reverse game theory': rather than taking a game's rules as fixed and solving for the equilibrium, mechanism design starts from a desired social outcome and works backward to construct the rules (the mechanism) under which self-interested, strategic players will voluntarily choose actions that produce that outcome as the equilibrium — even though the designer can't directly observe each player's private information.
Founded by Leonid Hurwicz starting in the 1960s, significantly extended by Eric Maskin and Roger Myerson — the three shared the 2007 Nobel Memorial Prize in Economic Sciences specifically for laying the foundations of mechanism design theory.
The Mechanism
Designing backward from the outcome you want to the rules that produce it
The direction of reasoning is inverted from standard game theory: instead of 'here are the rules, what will rational players do,' mechanism design asks 'here's what I want to happen, what rules would make that the rational choice for self-interested players who each hold private information I can't directly observe?'
01 · 'INCENTIVE COMPATIBILITY' IS THE CENTRAL DESIGN CONSTRAINT
A mechanism must make honesty (or the desired behavior) the actual best response
The core mathematical requirement in mechanism design is that the mechanism must be incentive-compatible — each participant, given their own private information, must find that truthfully participating (rather than lying, gaming, or free-riding) genuinely maximizes their own payoff. A mechanism that requires participants to act against their own self-interest for the good outcome to emerge will fail in practice, no matter how well-intentioned its design.
02 · IT UNDERPINS AUCTION DESIGN, VOTING SYSTEMS, AND MARKET DESIGN BROADLY
A field with enormous practical reach
Vickrey's second-price auction (see Auction Theory) is itself a landmark early mechanism-design result — it's specifically constructed so that truthful bidding is each bidder's dominant strategy. The same underlying logic extends to matching markets (kidney-exchange algorithms, school-choice assignment systems, designed substantially by Alvin Roth, who shared the 2012 Nobel Memorial Prize for this work), and to voting-rule design more broadly.
03 · THE REVELATION PRINCIPLE SIMPLIFIES THE ENTIRE DESIGN PROBLEM
A powerful theoretical shortcut
A foundational result (the Revelation Principle) shows that, for any mechanism achieving a given outcome, there exists an equivalent 'direct' mechanism where every participant's dominant strategy is simply to honestly reveal their private information — this dramatically simplifies mechanism design theory, since designers can restrict their search to truthful, direct mechanisms without losing any achievable outcomes.
Where It Fails / Inversion
Where it fails / inversion
Even a theoretically well-designed, incentive-compatible mechanism can fail in practice if participants don't understand or trust it, if implementation details introduce unanticipated loopholes, or if the assumption of purely self-interested rational players doesn't hold (real participants sometimes behave altruistically, spitefully, or with bounded rationality in ways theoretical mechanism design doesn't capture) — elegant mechanism design on paper still requires careful real-world testing and iteration.
How To Use It
Worked example · kidney exchange matching systems
Alvin Roth and collaborators designed matching mechanisms (starting in the early 2000s) that allow incompatible kidney donor-patient pairs to be matched into exchange chains with other incompatible pairs, producing viable transplants that wouldn't otherwise occur — critically, the mechanism is designed so that each hospital and patient has an incentive to participate honestly (reporting their true compatibility information) rather than trying to game the system for their own patient's advantage, which was essential to getting real hospitals to actually adopt and trust the system at scale.
How to use it
When you're trying to get a group of self-interested people to collectively produce an outcome you want (honest reporting, efficient allocation, fair division), don't just ask them to cooperate or trust you — design the actual rules of the interaction so that behaving the way you want is each person's own individually rational best response, given their private information. That's the entire discipline of mechanism design, and it works far more reliably than appeals to goodwill.
See Also