Dollar Auction¶

Mechanize the escalation trap with one rule change — the second-highest bidder also pays and gets nothing — so that backward induction makes each additional bid locally rational and rational players escalate past the prize's value.

Version: v3 · 2026-09-28 · History
Domain-specific #: 288
Domain group: Formal Sciences
Origin domain: Mathematics
Subdomains: Game Theory, Auctions and Mechanism Design → Mathematics

Core Idea¶

A small game (Shubik, 1971) that mechanizes the escalation trap. A prize of fixed value is auctioned, but the second-highest bidder also pays their bid and receives nothing. Backward induction shows why this loser-pays rule spirals: whenever you are second-highest, one more increment converts a certain loss into a possible gain, so each step is locally rational and players escalate past the prize value until one exhausts resources. Its payload is the cut between structural escalation (the rule rewards continuing) and cognitive sunk-cost escalation.

How would you explain it like I'm…

The Second-Place-Pays Trap

Imagine a game where people bid for a shiny coin, but the person who comes in second has to pay too and gets nothing. Whoever is in second place thinks, "If I bid just a little more, I might win instead of losing what I paid." So both keep bidding, higher and higher, until they have paid more than the coin is worth.

The Loser-Pays Auction

In a dollar auction, a dollar goes to the highest bidder, but the second-highest bidder also pays their bid and gets nothing back. Say you bid 90 cents and someone else bids a dollar. If you stop now, you surely lose your 90 cents; if you bid a little more, you might win the dollar. Bidding again looks smart every single time, so both players end up paying more than the dollar is worth. The ways out are to pick your limit before you start, or to change the rule so only the winner pays.

Loser-Pays Escalation Trap

The dollar auction is a small game invented to show how escalation traps work. A fixed prize is auctioned, and the only unusual rule is that the runner-up also pays their bid. Whoever is currently second faces a choice between a certain loss (their bid, gone) and a chance at the prize by raising, so raising is the sensible move at each step, and bids climb past the prize's value. This is different from the sunk-cost fallacy, which is a thinking mistake about money already spent: here each raise is actually reasonable given the rules, so the trap is built into the game itself. You cannot tell the two apart just by watching people bid. The fixes are to commit to a maximum in advance, or to redesign the auction so losers pay nothing.

The Dollar Auction (Shubik, 1971) is a minimal game that turns escalation into a mechanism. A fixed-value prize goes to the highest bidder, but the second-highest bidder must also pay their bid and gets nothing. Reasoning step by step, whoever is currently in second place faces a choice between a certain loss (their bid, forfeited) and a possible gain (the prize) if they raise by one increment, so raising is locally rational every time. Players therefore escalate past the prize's value, and in classroom demonstrations total bids routinely exceed it, until someone runs out of resources or accepts the loss. The game separates two kinds of escalation: the sunk-cost fallacy, a cognitive error of weighting money already spent, and incentive-structural escalation, where the rules make continuing the rational local choice. Behavior alone cannot tell an observer which kind is happening. The remedies are pre-commitment to a maximum exposure (like a litigation budget cap or stop-loss rule) and mechanism redesign that removes the loser-pays clause so only the winner pays.

Scope of Application¶

Lives across the auction-and-mechanism-design subfields of game theory and applied settings that genuinely carry an all-pay contest with a cheap marginal step.

All-pay-auction theory — the home turf, the cleanest worked instance of the all-pay family.
Patent and pay-to-play litigation — discovery battles where each side's spend is forfeit if it loses.
Lobbying and rent-seeking contests — competitions where defeated competitors still forfeit their spend.
Conflict studies and arms races — attritional conflicts where the lagging side's investment is wasted unless it presses on.
Behavioral economics teaching — the classroom demo in which a small prize auctions for more than its value.

Clarity¶

Naming the game makes legible a distinction "sunk-cost fallacy" prose collapses: escalation driven by a cognitive error versus escalation driven by an incentive structure that genuinely rewards continuing. Behavior alone cannot separate these. The diagnostic question — reasoning failure to correct, or mechanism to dismantle? — matters because the two species call for opposite remedies.

Manages Complexity¶

A whole class of protracted escalation episodes reduces to a short structural checklist: a fixed prize, a loser-pays rule, and a cheap marginal step. Where those hold, backward induction delivers the outcome with no case-specific modeling. The two-species cut then collapses diagnosis and remedy to one discriminating question — read the species off the payoff rule, then read the fix off the branch.

Abstract Reasoning¶

The game licenses a diagnostic detecting the trap from its signature and classifying the species off the payoff rule rather than behavior, an interventionist move breaking the chain by pre-commitment or removing the loser-pays rule (and predicting debiasing fails structurally), boundary-drawing that relocates the rationality from psychology to the rule, and a prediction that clear thinking drives rather than prevents the spiral.

Knowledge Transfer¶

Within game theory the dollar auction transfers as mechanism — its backward-induction analysis and two remedies carry intact to every setting that literally has the loser-pays/small-step structure (all-pay litigation, rent-seeking, arms races), recognition of the same incentive structure, not analogy. The dividing line beyond auction theory is whether the signature really holds; where it does, the lesson is best carried by the declared parent Lock-In under an all-pay incentive (with the more general Escalation of Commitment riding alongside). Loose "quagmire" invocations that lack the payoff rule are analogy, sliding back into the cognitive sunk_cost species.

Relationships to Other Abstractions¶

Current abstraction Dollar Auction Domain-specific

Parents (1) — outgoing typed relations

Dollar Auction is a decomposition of Lock-In Prime

Stripping the discrete auction frame leaves a state in which current exposure makes the forward cost of exit exceed one more step of staying.

Hierarchy paths (7) — routes to 4 parentless roots

Dollar Auction → is a decomposition of Lock-In → is a kind of Path Dependence → presupposes Dependency

Neighborhood in Abstraction Space¶

Dollar Auction sits in a moderately populated region (42^nd percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Game-Theoretic Models & Paradoxes (38 abstractions)

Nearest neighbors

Vickrey Auction — 0.90
Ultimatum Game — 0.88
Hawk–Dove Game — 0.87
Traveler's Dilemma — 0.87
Pirate game — 0.86

Computed from structural-signature embeddings · 2026-10-25