# Signaling Game **The model:** A strategic interaction where one party (the sender) has private information and takes a costly action to credibly communicate that information to another party (the receiver). The signal must be *costly* to be credible — cheap signals can be faked by anyone, so they convey no information. Costly signals work precisely because they're expensive: only senders who genuinely possess the claimed quality can afford to send the signal. Spence's canonical example: a college degree. The degree may or may not increase a worker's actual productivity, but it signals to employers that the worker has the intelligence, discipline, and conformity to complete a difficult, costly process. High-ability workers can afford the degree (low marginal cost of studying); low-ability workers can't (high marginal cost). The employer reads the signal and hires accordingly. The degree works as a signal even if it adds zero actual skill. ## The Mechanism **The separating condition.** A signal works only if the cost of sending it differs across types. If both high-quality and low-quality senders can easily send the signal, it conveys no information (a *pooling equilibrium* where the signal is noise). If only high-quality senders can afford to send it, the signal separates the types (a *separating equilibrium* where the signal is informative). The cost differential is the engine of credibility. **Cheap talk is not credible.** Any message that is costless to send and bears no consequence if false conveys no information in equilibrium. "I'm a hard worker" is cheap talk. Working 80-hour weeks for years at a prestigious firm is a costly signal. Campaign promises are cheap talk. Voting records are costly signals (they constrain future flexibility). **Costly signals can be wasteful.** The peacock's tail, the college degree, the luxury brand — all involve real resources spent on signaling rather than production. From a social-welfare perspective, this expenditure is pure waste (it doesn't increase the total pie, just redistributes information). Signaling arms races — where each sender must signal more expensively to stay ahead — can consume enormous resources for purely distributive purposes. **Counter-signaling.** Sufficiently high-quality senders sometimes *refrain* from signaling — confident that their quality is evident without it, or deliberately distinguishing themselves from try-hard signalers. [[Warren Buffett]] drives a modest car; Silicon Valley founders wear hoodies. Counter-signaling works only from an already-established position of credibility. ## KB Applications ### MAGA Loyalty Signaling The MAGA movement has developed an elaborate signaling economy. Loyalty signals are *costly* by design — they require saying things that damage your reputation outside the movement, defending positions that are empirically indefensible, or attacking former allies. This cost is the point: it's what makes the signal credible. Anyone can say "I support Trump"; only a genuine loyalist will publicly defend the indefensible (January 6 revisionism, election-denial claims they know to be false, tariff policies they know will hurt constituents). [[J.D. Vance]]'s trajectory illustrates the pattern. His 2016 anti-Trump statements made his 2022 pro-Trump conversion a more costly signal — precisely because it required repudiating his own public record. "I will do what [Trump] asks" is not a statement of policy preference; it's a costly signal of unconditional loyalty. The Senate confirmation process has become a signaling arena: nominees signal loyalty by accepting positions they'd previously criticized. ### Virtue Signaling (the Term's Game-Theoretic Content) The right-wing accusation of "virtue signaling" has actual game-theoretic content, even though it's deployed as a dismissal. The claim is that progressive public statements (pronouncing pronouns, displaying flags, making solidarity declarations) are *cheap talk* — costless signals that don't separate genuine commitment from performative display. The game-theoretic response: some of these signals *are* costly (putting your career at risk, alienating family members, spending time and money on activism) and therefore credible. Others are genuinely cheap (a social-media bio line, a bumper sticker) and convey less information. The right applies "virtue signaling" indiscriminately to both, which is analytically sloppy — but the underlying game-theoretic insight (cheap signals don't separate types) is valid. The irony: MAGA's own signaling economy operates on exactly the same principles, with the same mixture of genuinely costly signals and cheap performative display. The red hat is cheap talk; quitting your law firm is a costly signal. ### Political Endorsements as Signals Endorsements are signals. Their informational value depends on their cost. A safe endorsement (backing the front-runner after they've already clinched the nomination) is cheap and uninformative. An early endorsement of a risky candidate is costly (reputation risk, opportunity cost of backing the wrong horse) and therefore credible. [[Chris Christie]]'s early Trump endorsement (2016) was a genuinely costly signal — it told Trump's campaign that Christie was willing to burn bridges for the alliance. ### Credential Display and Authority The display of credentials (degrees, titles, institutional affiliations) is signaling. The game-theoretic question is whether the signal's cost structure still separates types. The right's anti-elite populism exploits a real concern: if credential acquisition has been corrupted (grade inflation, prestige capture, institutional homogeneity), the signal's separating power diminishes, and credential holders may be type-pooling rather than type-separating. The response — "don't trust experts" — overcorrects by rejecting *all* costly signals of expertise, which is game-theoretically worse than accepting imperfect signals. The optimal response is to improve the signal (restore credential quality) or develop supplementary signals (prediction markets, track-record databases, skin-in-the-game requirements). ### Information Warfare and Signal Flooding One Russian information-warfare strategy is *signal flooding* — generating so many signals (true, false, half-true, misleading) that the receiver's ability to distinguish costly from cheap signals collapses. When the information environment is saturated with noise, even genuinely costly signals lose their separating power because the receiver can't process them. This is the game-theoretic foundation of Bannon's "flood the zone with shit" — it's an attack on the signaling equilibrium itself. ## See Also - [[Adverse Selection]] — what happens when signaling fails and low-quality types drive out high-quality ones - [[Commitment Device]] — a related mechanism; commitment is a costly signal of future behavior - [[Moral Hazard]] — the post-signal problem: behavior changes after the information asymmetry is resolved - [[Game Theory Hub]] — hub page - [[George Akerlof]] — information economics; the "Market for Lemons"