I got thinking about black SUVs recently - why a city full of them makes every individual one harder to read - and ended up in the biology of mimicry. Which has a cleaner taxonomy than you’d expect.

In Batesian mimicry, a harmless species copies a dangerous one. Hoverflies mimicking wasps, king snakes copying coral snakes. The mimic benefits. The model is harmed, because every fake wasp that doesn’t sting teaches predators to stop respecting the signal. Too many mimics and the whole system collapses.

In Müllerian mimicry, multiple genuinely dangerous species converge on the same warning signal. Several toxic butterfly species all evolving similar wing patterns. Both benefit: predators learn the pattern faster, each species shares the cost of educating them.

These are the textbook cases. They cover most of what you see in nature, and almost all of what gets taught.

But there’s a third dynamic that fits neither. The mimic is harmless, the model is dangerous - classic Batesian setup - except the model actually benefits from the mimics’ existence. Not through signal reinforcement like Müllerian mimicry. Through cover.

The Black Suburban

A black Chevrolet Suburban with tinted windows on a DC street. Secret Service? FBI? Tech executive’s driver? School pickup in Bethesda? Uber Black?

You can’t tell. You register power but not whose.

This is a mimicry ecology. Federal agencies and private security are the models - genuine authority and threat. The wealthy civilian in a blacked-out Escalade is the Batesian mimic, borrowing the threat display without the backing.

Standard Batesian logic says this should harm the model. Every fake government SUV that turns out to be a soccer mom should erode the signal, train observers to stop treating black SUVs with caution.

The opposite happens. The government wants a city full of black SUVs. Their actual surveillance and enforcement vehicles disappear into a sea of identical-looking civilian ones. The mimics don’t dilute the signal. They provide camouflage.

In normal Batesian mimicry, the model would prefer fewer mimics. Here, the model would prefer more.

The Suit Doesn’t Work This Way

You might think suits are the same deal. Banker, lawyer, politician, con artist - all wearing the same thing, all benefiting from the ambiguity.

But suits are standard Batesian. Legitimate professionals are harmed when con artists wear suits. The signal degrades. Trust in “well-dressed man” erodes every time one turns out to be running a scam. Bankers and lawyers would prefer that con artists couldn’t dress like them.

Military uniforms sit differently again. You can buy and wear military surplus freely — full camo, boots, the lot. What’s restricted isn’t the aesthetic but fraudulent representation: claiming to be active military in order to deceive. The state doesn’t care if you wear the look. It cares if you use it to impersonate.

And crucially, the state doesn’t benefit from civilians wearing camo. It’s indifferent. A city full of people in surplus gear doesn’t help actual soldiers or intelligence operatives do their jobs.

Three cases, three dynamics:

  • Suit: Adversarial. Mimics degrade the model’s signal. Classical Batesian.
  • Military uniform: Neutral. Aesthetic adoption is unrestricted and doesn’t affect the model either way.
  • Black SUV: Mutualistic through camouflage. Mimics improve the model’s position by providing cover.

The black SUV is the one that doesn’t have a category.

What Biology Says (and Doesn’t Say)

The biological literature circles around this but never lands on it.

The closest named concept is quasi-Batesian mimicry, described by Michael Speed in the early ’90s - interactions where the mimic is mildly unpalatable rather than completely harmless, creating dynamics that shift with relative frequency. It’s a spectrum between Batesian and Müllerian. Close, but it’s still about signal reliability, not camouflage.

The target dilution effect is mechanically closer: more bodies in a group means lower per-capita predation risk. More civilian black SUVs means any given government vehicle is harder to pick out. And positive frequency-dependent selection on warning signals means the more ubiquitous the aesthetic becomes, the more protective it is for everyone using it - including the model.

The evidence is there. A model that benefits from mimics not through signal reinforcement but through concealment via ubiquity - documented, empirically supported, and unnamed.

The most precise description you’ll find is something like “frequency-dependent predator confusion in systems where abundant mimics provide incidental protection to models through signal dilution.” Which is why nobody uses it.

Every Dissident Needs a Million Civilians

Once I had the framing, I started seeing the same structure wherever strategic actors need to hide in plain sight.

Encryption. When only dissidents use encrypted communication, using encryption is the signal. “This person has something to hide.” When everyone uses Signal or WhatsApp end-to-end, dissidents disappear into the crowd. People with genuine operational security needs benefit directly from mass civilian adoption. The mimics provide cover.

Cash. When only criminals use cash, cash use becomes suspicious. When everyone uses cash, criminal transactions are invisible. The model benefits from the mimics’ ubiquity.

Tor. The cleanest example. The US Naval Research Laboratory developed the underlying technology for Tor, and its designers recognised from the outset that a network used only by intelligence operatives isn’t anonymous — it’s a list. The whole system depends on civilian “mimics” flooding the network to make strategic traffic unidentifiable. The mutualism isn’t incidental. It’s architectural.

In each case, the model would be harmed by being the only one using the signal. The mimics aren’t degrading it - they’re making it invisible through sheer volume.

The Gap in the Taxonomy

The mimicry taxonomy has expanded well beyond Batesian and Müllerian. Emsleyan mimicry covers a deadly species mimicking a moderately dangerous one. Aggressive mimicry covers predators disguised as harmless species. Named categories exist for edge cases most people will never encounter.

But this specific dynamic - harmless mimics that benefit the model by providing concealment rather than signal reinforcement - sits in a gap. Described piecemeal. Never unified. Nobody’s bothered to name it.

Named concepts get taught, cited, used outside the original field. Unnamed ones stay scattered across subdisciplines, rediscovered independently by people looking at black SUVs and encryption adoption and cash economies without realising they’re looking at the same thing.

Maybe “cover mimicry” works. But the interesting thing isn’t the name - it’s what the gap reveals: signalling theory is overwhelmingly focused on signals that work. It has less to say about signals that succeed by becoming indistinguishable from noise.