I asked whether intelligent life on another planet would follow anything like our evolutionary path, or whether humans are a one-off.

Everybody who answers this question reaches for the same argument, and I reached for it too before I checked it. It is a good argument. It is also, in the specific form everyone uses, wrong.

The argument everybody makes

Evolution repeats itself. Give it the same problem twice and it finds the same answer twice, so the features that keep reappearing on Earth are the features we should expect anywhere.

The star witness is the eye. In 1977, Luitfried von Salvini-Plawen and Ernst Mayr surveyed photoreceptors across the animal kingdom and concluded that they “have been evolved independently in at least 40, if not 65 or even more separate phyletic lines.”

Forty times. Sixty-five, maybe. Evolution keeps finding vision, over and over, in creatures that share no eye-bearing ancestor. If it is that easy here, surely it is that easy elsewhere. Therefore aliens have eyes.

I believed this for about twenty years. It has a problem.

Those forty origins are not independent

Start with the protein. Every animal eye that has ever been examined detects light using an opsin, a receptor protein that sits in a membrane and changes shape when a photon hits the pigment bound inside it.

Not opsins in the sense of “a similar kind of molecule.” Opsins in the sense of one family, descended from one ancestral protein. Roberto Feuda and colleagues resolved the phylogeny in 2012 and found that all known opsins across cnidarians, ctenophores and bilaterians fall into the same groups and share a common ancestor with the melatonin receptors. Even Placozoa, which are flat animals with no eyes, no neurons and roughly four cell types, are carrying opsins around.

So the light-detecting molecule was invented once, before the ancestors of jellyfish and humans went their separate ways, and everything since has been reuse.

Then there is the switch. In 1995 Georg Halder, Patrick Callaerts and Walter Gehring took the fruit fly gene eyeless and turned it on in the wrong places. They grew eyes on the wings, the legs, and the antennae of flies. Not eye-shaped blobs: “the ectopic eyes appeared morphologically normal and consisted of groups of fully differentiated ommatidia with a complete set of photoreceptor cells.” Eyeless is a fly version of Pax6, the same gene whose disruption causes aniridia in humans. Their conclusion was that it “may function as a master control gene throughout the metazoa.”

One protein for catching photons. One genetic switch for saying “build the eye here.” The forty independent origins are forty deployments of a kit that was already in the box.

Biologists call this deep homology, and Shubin, Tabin and Carroll gave it that name in Nature in 2009. The bodies are unrelated. The instructions are cousins.

Which quietly ruins the argument for aliens

Here is the part that took me a while to see, and which I have not found stated plainly anywhere.

The convergence argument needs eyes to be easy. What deep homology shows is that eyes are easy if you already have opsins and Pax6. Forty origins from a shared starting kit tells you almost nothing about how hard it is to invent the kit.

And an alien does not have our kit. It has no opsins, because opsins are a particular arrangement of amino acids that appeared once, in one lineage, on one planet. Whatever it uses to detect light, it had to invent from nothing.

So the single strongest piece of evidence for convergent evolution is close to the weakest possible evidence about aliens. It is a measurement of how often life reuses a solution, offered as though it measured how often life finds one.

What does transfer is physics, not genes

None of which means the whole enterprise is hopeless. It means we have been citing the wrong constraint.

Genes do not cross between biospheres. Physics does. And a great deal of what an eye is has nothing to do with opsins.

If you want to know where something is, rather than merely that light exists, you need to form an image, and forming an image in a transparent medium means an aperture, a way of bending light, and a surface to project onto. That is optics, and optics is the same in Andromeda. The trade-off between a wide aperture that gathers photons and a narrow one that resolves detail is not a fact about Earth. It falls out of the wave nature of light.

The same goes for moving through fluid. Ichthyosaurs, sharks and dolphins converged on the same body because the Navier-Stokes equations do not care what your ancestors were. This is the same argument that makes a photosynthetic human impossible: physics sets the shape of the possible, and biology fills in what it can.

So the confident prediction is not “aliens will have eyes.” It is: any organism that hunts, in a transparent medium, lit by a star, will face the identical optical problem, and there are not many solutions to it. Whether it solved that problem, or ever needed to, is a different question.

Crabs, five times, and then not

My favourite piece of evidence points both ways at once.

Decapod crustaceans keep turning into crabs. Wide, flat, with the abdomen tucked underneath. It has happened, in the words of a 2021 review by Wolfe, Luque and Bracken-Grissom, in “at least five cases of convergent evolution of the crab-like body plan,” a phenomenon named carcinisation over 140 years ago.

Five separate lineages, one shape. This is the convergence argument at its best, because there is no shared “crab gene” being switched on. It is a body plan that keeps winning.

Except the same review names the opposite process, and it is just as real. Decarcinisation. Lineages become crabs, and then lineages stop being crabs, repeatedly.

If a shape is so favoured that evolution finds it five times, and so unfavoured that evolution abandons it about as often, then “favoured” is doing less work than the story needs. Constraint is real. Constraint is not destiny.

The number that ends the conversation

Eyes: many times. Powered flight: four times, give or take. Crab: at least five, minus the times it gave up.

Technological intelligence, the thing everybody actually means when they ask this question: once.

One event. And it is an event we can only observe from the inside, because a planet where it did not happen has nobody standing on it wondering about the rate. There is no way to derive a probability from a sample of one that you were required to be part of in order to take the sample.

Stephen Jay Gould said that if you replayed the tape of life the result would be unrecognisable. Simon Conway Morris said the tape would converge on something rather like us. They were arguing about the same fossils. That argument has not been settled, and I do not think it can be settled from one planet, because the entire dataset is a single run of an experiment with no control.

So what would they look like

If they exist, if they live in a transparent medium, and if they need to know where things are, I will bet on something eye-shaped, because the optics permit little else. I will bet on streamlining for anything that swims fast. I will bet against five fingers, against two arms, and against anything you could put in a rubber suit.

And on the question of whether there is anybody there to have a shape at all, I have exactly one data point, and it is me.

Which is not nothing. It is just not a rate.

Sources