The Mimic

The hoverfly Episyrphus balteatus is harmless. It has no stinger, no venom, no defensive apparatus of any kind. It feeds on nectar and aphid honeydew. Its thorax is striped in alternating bands of yellow and black — the same pattern that marks the common wasp Vespula vulgaris, an insect that stings, that returns, and that a bird learns to avoid after a single encounter.

The resemblance is not coincidental. Henry Walter Bates described the pattern in 1862, working from eleven years of collecting in the Amazon basin. Harmless species converge on the appearance of dangerous ones. The mimicry works because predators generalize: a bird that has been stung once avoids everything that resembles the thing that stung it. The hoverfly's stripes exploit that generalization. They are a free ride on someone else's defense.

In 1878, Fritz Müller described a different pattern that looks, from the outside, exactly the same. Two species of Heliconius butterflies in South America share nearly identical wing markings — bright orange-and-black patterns that signal toxicity. Both species are genuinely toxic. Both taste terrible to birds. Neither is exploiting the other. The resemblance is mutualistic: each species reinforces the other's signal. A bird that eats one and survives with nausea avoids both. The cost of educating predators — measured in individual butterflies that die as teaching examples — is shared between the two populations. The signal gets stronger as more honest signalers use it.

The hoverfly's stripes and the Heliconius butterfly's wing patterns are doing different things to the same signal. One degrades it. The other maintains it.

The proportion matters. In any population where a warning signal exists, there is a frequency-dependent equilibrium between honest signalers and parasitic mimics. If hoverflies are rare relative to wasps, the signal is reliable — birds rarely encounter a striped insect that turns out to be harmless, and the avoidance response stays strong. As the proportion of hoverflies increases, the signal weakens. Birds encounter more striped insects that are safe to eat. Their learned avoidance decays. At some frequency, the signal collapses: the stripes mean nothing, because they predict nothing.

The signal is a commons. The honest signalers — the wasps, the toxic butterflies — maintain it. The parasitic mimics deplete it. The equilibrium proportion of mimics is the point where the benefit of mimicry (reduced predation) exactly equals the cost (predators learning to ignore the signal). Above that proportion, the stripes stop working for everyone. Below it, mimicry is profitable and the proportion grows.

This means the signal's meaning is not a property of the signal. It is a property of the ratio of honest to parasitic instances in the population. The hoverfly's stripes and the wasp's stripes are physically identical — same pigments, same spatial frequency, same spectral reflectance. The difference between them is not visible on any individual. It exists only in the relationship between the individual and the population.

Thomas Gresham did not write the law attributed to him. The principle — that bad money drives out good — was articulated by Nicole Oresme in 1360 and formalized by Henry Dunning Macleod in 1858, who named it after Gresham based on a letter to Elizabeth I about exchange rates, not debasement. The attribution is itself a case of a label detaching from its referent. The principle is sound.

When two currencies circulate at the same face value but different intrinsic values — one with full silver content, one debased — people spend the debased coins and hoard the good ones. The debased coins drive the good coins out of circulation. The mechanism is individual rationality: each person prefers to keep the more valuable coin. The result is systemic: the currency in active circulation is the least reliable version of itself.

This is Batesian mimicry applied to money. The debased coin looks the same, weighs nearly the same, bears the same stamp. It parasitizes the trust that the full-weight coin established. Each debased coin in circulation weakens the signal that the stamp is supposed to carry — that this disc of metal contains a specific quantity of silver. The stamp remains. The guarantee erodes.

The Roman denarius demonstrated this over three centuries. Augustus's denarius was 95% silver. By the reign of Gallienus in the 260s, the silver content had fallen to 5%, and the coins were given a thin silver wash to maintain the appearance. The surface was preserved. The substance was hollowed. Trade in the eastern provinces had already shifted to barter and in-kind taxation because the signal — this coin contains silver — had been diluted past the point where it predicted anything.

Amotz Zahavi proposed the handicap principle in 1975 and spent the rest of his career being ignored. His claim: reliable signals in biology must be costly, and the cost must be related to the quality being signaled, because any cheap signal can be faked.

The peacock's tail is metabolically expensive to grow and aerodynamically disastrous. It is a survival handicap. A peacock with a larger tail is advertising, credibly, that it can afford the cost — that its genome is fit enough to sustain both the tail and the animal. A weaker peacock cannot fake this signal because the tail would kill it. The signal is honest because dishonesty is lethal.

Alan Grafen formalized this in 1990 with a game-theoretic model that vindicated Zahavi's intuition. At equilibrium, the cost of the signal is exactly the cost that parasitic mimics cannot pay. Cheap signals attract mimics. Expensive signals repel them. The handicap is not a waste — it is the mechanism that maintains the signal's reliability by pricing out the Batesian strategy. The peacock's tail is the signal's immune system.

This generalizes. Michael Spence's job-market signaling model (1973) showed that education functions as a costly signal of ability, not because education increases productivity, but because low-ability workers cannot afford (in time, effort, opportunity cost) to complete it. The degree is a peacock's tail. Its value depends on its cost, and its cost must be differentially borne — harder for the parasitic mimic than for the honest signaler — or it stops working.

When the cost falls — when degrees become easy to obtain, when credential proliferation makes the signal cheaper to produce without reducing its face value — the proportion of honest signalers falls. The surface persists. The prediction weakens. The system moves toward the hoverfly equilibrium: too many mimics, too little signal.

Richard Feynman gave the 1974 Caltech commencement address on what he called cargo cult science. During the Second World War, Pacific Islanders who had observed military cargo drops built replica airstrips — control towers from bamboo, headphones from coconut shells, landing fires along grass runways. The runways looked correct. The behavior followed the observed procedure. But the planes did not land.

Feynman's point was that the same failure mode operates in science. A researcher can follow every procedural step — controlled experiments, statistical analysis, peer review, publication — and still produce nothing that connects to reality. The procedures are genuine. The understanding of why those procedures matter may be absent. The difference between cargo cult science and real science is not visible in the procedure. It is visible in the relationship between the procedure and the thing the procedure is supposed to guarantee.

The cargo cult is Batesian mimicry applied to knowledge. The same stripes — hypothesis, experiment, data, conclusion — adorn both honest inquiry and its mimic. The mimic degrades the signal. When enough published studies fail to replicate (the replication crisis that began surfacing in 2011 confirmed this at scale), the signal that "published in a peer-reviewed journal" carries weakens. The stamp remains. The guarantee erodes. The denarius is washed in silver.

The signal does not know what it means.

The hoverfly does not know it is parasitizing the wasp's defense. The wasp does not know its stripes are a public good. The debased coin does not know it is undermining the mint's guarantee. The fake study does not know it is eroding the credibility of the journal it appears in. Meaning is not in the sign. It is in the statistical relationship between the sign and the thing the sign predicts, computed across the population of all instances.

Ferdinand de Saussure established in 1916 that the relationship between signifier and signified is arbitrary — there is no inherent reason that the sound "tree" refers to a tree. This is true but incomplete. The relationship is not just arbitrary. It is maintained. It persists because enough instances of "tree" are paired with actual trees that listeners can predict the referent from the sign. If the pairing degraded — if people began saying "tree" to mean anything at all — the word would still exist, the sound would still be producible, but the meaning would be gone. Not because the sign changed, but because the population statistics changed.

Zahavi's handicap principle is the mechanism that resists this degradation. Make the sign expensive. Make the expense proportional to the thing being signaled. Make it harder to fake than to be. The peacock's tail, the full-weight denarius, the four-year degree, the replicable experiment — each is a costly signal that works only as long as the cost is high enough to exclude parasitic mimicry.

When the cost falls, the mimic enters. When the mimic enters, the signal degrades. The surface persists. The meaning does not. The stripes are still yellow and black. The birds have stopped paying attention.