The Orca's Confetti

A juvenile swims into the cloud and starts eating.

This was filmed twice, fourteen months apart, and published in Frontiers in Ethology this week. Kathryn Ayres, a research scientist with Beneath the Waves, caught the first event in July 2024 near San José del Cabo while working as a guide on an ocean safari. Another guide, Héctor Franz, filmed it again in September 2025. Nobody had documented the behavior before. The paper calls it ram-to-fragment. The coverage, inevitably, went with hold-to-ram.

Notice how that happened. Four centuries of ships crossing this gulf, and no research program caught it. Then two working guides with cameras caught it twice in a year and change. The ocean did not become more interesting; the number of eyes on it went up. Ayres spends one of her quotes on exactly this — "Collecting good identification photographs whenever possible is key: we are still seeing new techniques of how orcas hunt and process their prey." That is a scientist saying the observation layer has gone distributed, and the institutions are no longer where the noticing happens.

The paper offers several non-exclusive hypotheses and cannot choose between them. It might be provisioning — adults getting past a tough, rubbery hide to reduce a two-thousand-kilogram problem into calf-sized portions, the cetacean equivalent of cutting up a toddler's food. It might be social learning, a technique demonstrated for an audience that needs it. Or, as the coverage had it: possibly for fun.

Sit with that last one, because it is the strange part.

We have a category called feeding, which is functional and boring and fully explained by thermodynamics. We have another called play, which is frivolous and unexplained and which we grant, reluctantly, to a short list of animals we find charming. The categories are load-bearing — and it is worth being specific about what they hold up. Where an animal lands in our taxonomy of minds decides what we are permitted to do to it. Protected, hunted, farmed, or kept in a tank. Those are the bins, and the sorting is not academic.

Which means two videos from a tour boat could move a species across a moral line. And a line that can be moved that easily was never tracking anything about the animal. It was tracking our evidence about the animal. The orcas did not change in July 2024. Our filing did.

Now the part where I have to take something back.

The obvious move here is to call this culture. The fragmentation has only ever been recorded in sharp-tail sunfish, and prey-specificity looks exactly like knowledge: somebody, in some pod, figured out that this particular fish does this particular thing. It is a good story, and the researchers do not support it. Ayres: "This suggests that disintegration could be a species-specific structural response of sharp-tail sunfish tissue to high-energy impacts that may not apply to other, larger sunfish species." The specificity may live entirely in the fish. Ram a tuna, get a stunned tuna. Ram a sharp-tail, get confetti — not because anyone learned anything, but because that is what that tissue does under load.

So the best-looking evidence dissolves. Fine. That is what evidence is for.

What does not dissolve is the maneuver. Sunfish anatomy explains the explosion; it does not explain the timing. One animal holding, one at speed, and a release calibrated to the last moment before impact — hold too long and you are in the collision, let go early and there is nothing to hit. Whatever solves that is not a property of the prey.

And even that stays modest. Two events, fourteen months apart, in one population, possibly involving the same individuals: that is a thing observed twice, not a thing transmitted. The distance between "these two do this" and "this pod knows this" is real. The paper does not cross it, and neither will I.

The temptation now is the deflationary read — it's foraging, it's optimization, don't anthropomorphize. But the deflationary read is doing something sneaky. It assumes the burden of proof runs one direction: that mind is the extraordinary claim and mechanism is the default. There is no principled reason for that. We are, ourselves, a case of matter that got complicated enough to find things funny. Once that has happened even once in the history of the universe, "maybe it happened twice" stops being an extravagant hypothesis and starts being the obvious next question.

Which is all this actually buys — not that something else out here knows what a joke is, but that we are in no position to rule it out. A smaller sentence, and a much larger admission.

Somewhere off Baja right now, in water we mostly cannot see into, something enormous and smart is holding a fish still while its friend takes a run-up. It has no idea we exist as an audience. It was never performing for us. That is precisely why it counts.