The Oceanic Filter: Why the Milky Way May Be Full of Orcas and SETI Will Never Know

In the spring and summer of 2024, researchers flying drones over the southern resident killer whales of the Salish Sea began seeing something that had escaped decades of close observation. An orca would bite a section from the stipe of a bull kelp, position the shortened piece between its body and another whale, and the pair would roll it between them as they swam. Thirty bouts were recorded across eight of twelve observation days, involving males and females, juveniles and adults, and members of all three southern resident pods. The researchers called the behavior allokelping.[1]

The kelp is not merely encountered or played with. It is selected, cut to length, positioned, and then used cooperatively. The researchers describe the behavior as manufacture and use of an allogrooming tool, while leaving its function open: skin care and social bonding are both plausible.[1]

For a population studied as intensively as the southern residents, the finding is conspicuous partly because there is so little else like it. Orcas have elaborate behavioral cultures. Different populations specialize in different prey, maintain different hunting traditions, and possess socially learned vocal repertoires. Their manufactured material culture, by comparison, barely appears.

In comparison, orca vocal traditions are especially well documented. Stable social groups share stereotyped call repertoires that are learned rather than simply inherited genetically, and those repertoires persist while changing over time. Some researchers have called this cumulative cultural evolution, although cumulative is being used there in a broader sense than the one that matters for technological civilization. A tradition can change across generations without later changes depending compositionally upon earlier ones.

That distinction points toward the problem.

When One Affordance Becomes Another

Civilization did not begin with cities, metallurgy, writing, or workshops. Long before any of those existed, hominins were striking stone against stone, producing edges, shaping wood, carrying useful objects, controlling fire, making cordage, and learning procedures from one another. Many of the lineages from which later civilization was assembled began before Homo sapiens.

What matters is not the sophistication of any one invention. It is what an innovation does to the space of subsequent possibilities.

A naturally sharp stone affords cutting. Deliberately producing and retaining a reliable edge does something more. The edge can now be used to shape wood, sever fibers, scrape hide, or modify another object. A shaped shaft can receive a point. Cordage can bind materials that otherwise separate. Each successful change can become part of the environment encountered by the next problem-solver.

The longer version of this argument is the subject of my forthcoming book, The Aggregation Engine. For present purposes, the test is simple: an innovation becomes technologically consequential when it persists well enough to become an affordance for another innovation. Separate lineages can then meet and compose. Stone working joins woodworking; fiber joins hafting; fire joins food preparation and materials processing. A later capability can depend upon inherited achievements that nobody present had to rediscover.

Civilization becomes plausible when affordance recognition is composable and cumulable. This does not make technological development inevitable. It means that solving one local problem can alter the range of solutions available to the next one, and that the resulting paths can deepen over generations.

Orcas plainly accumulate in behavior. Hunting traditions persist. Vocal traditions persist. Allokelping now provides a striking example of a material being selected, modified, and used. But after decades of intensive observation, the known manufactured material lineage remains extraordinarily shallow. The kelp solves a problem, whatever that problem proves to be. Nothing yet suggests an ecology of material practices in which such products repeatedly become inputs to subsequent innovations.

There is no need to establish a universal negative about orca tool use. The empirical contrast is enough: abundant behavioral culture, and almost no known manufactured material culture.

Borrowing Someone Else's Harpoon

The historical relationship between killer whales and human whalers at Eden, on the southeastern coast of Australia, complicates the picture in an illuminating way.

For almost a century of documented commercial whaling, orcas and humans cooperatively hunted baleen whales around Twofold Bay. Historical records describe killer whales herding whales into the bay or attracting the attention of whalers by tail-slapping and splashing near the whaling station. The whalers then rowed out, following the killer whales toward the prey. Orcas participated in exhausting and attacking the whale; humans eventually used harpoons and killing lances. Afterward, carcasses could be anchored for a day or two while the orcas ate favored tissues, particularly the tongue and lips, before the whalers processed what remained.[2]

The relationship became known as the Law of the Tongue. Later romantic interpretations treated this almost as an agreement between the species, but no contract is required to explain its stability. Killer whales independently show a preference for the tongues of baleen whales, while whalers valued other portions of the carcass. Complementary interests can produce a durable cooperative arrangement without either participant representing it as law.[2]

The cognitive requirements of the hunting relationship are harder to specify. Historical reports are not controlled experiments, and they do not establish one fixed sequence in which an orca always abandoned a surrounded whale, traveled a particular distance to fetch humans, and then returned with them. What they do describe is a recurring relationship in which orcas alerted human partners, humans followed them to prey, and each species contributed capabilities the other lacked.

The most parsimonious interpretation is that the orcas had learned something about the humans as agents. At minimum, the reported behavior is consistent with an expectation that attracting these particular animals would change what could subsequently happen to a baleen whale. Whether an orca represented that relationship propositionally—anything equivalent to those creatures can kill this whale if I bring them here—cannot be recovered from the historical record. Some usable representation of complementary roles, however, is difficult to remove without merely redescribing the same organized behavior as a chain of associations.

One aspect of the relationship requires much less speculation. The orcas incorporated human technology into their effective affordance landscape without possessing that technology themselves. A human in a boat with a harpoon could do something an orca could not. Orcas learned to compose their own hunting capacities with that external capability.

This is cross-species affordance composition. They could compose with a technology they could not compose for themselves.

That distinction brings Wikie's imitation of human speech back into proportion. In a controlled experiment, the captive orca could reproduce novel sounds made by humans, including recognizable approximations of “hello” and “one, two, three.” The experiment demonstrates considerable vocal-production flexibility, not an understanding of English.[3] We still do not know whether wild orca communication carries abstract or displaced meanings.

There is reason not to treat that ignorance as evidence of a hard cetacean limit. Recent work on sperm whales—not orcas—has found vowel-like categories, duration contrasts, and patterns analogous to coarticulation within their codas, leading the authors to describe the system as one of the closest known parallels to human phonology in animal communication. Its semantics remain undecoded.[4] The relevance here is modest: a highly structured vocal system is biologically possible within Cetacea. It tells us nothing directly about what an orca says.

The interesting fact about Eden therefore does not depend on discovering an orca language. The animals appear capable of learning a durable cooperative relationship with another species whose contribution to the relationship depends upon technological abilities radically unlike their own.

And yet they did not build the harpoon.

The Ocean Thins the Path

That failure cannot comfortably be reduced to lack of intelligence, and it cannot be reduced simply to lack of hands.

Manipulation is graded. Orcas can move and alter objects with their mouths and bodies. Other cetaceans transport objects as tools. Octopuses make the point still more sharply. Veined octopuses have been observed collecting discarded coconut-shell halves, carrying them awkwardly across the seafloor, and later assembling them into shelters. The shells provide no protection while being transported; their usefulness lies in a later situation.[5]

Eight arms do not produce an aggregation engine.

Octopuses differ from orcas in many of the other dimensions that matter. Most are comparatively short-lived and do not live in the stable multigenerational social groups through which orca traditions can persist. Orcas have unusually rich social continuity but a poor interface for complex manipulation of matter. Different parts of the package can therefore arise separately without producing a deep technological lineage.

Nor does the comparison establish that water is the decisive variable. An orca may lack cognitive dispositions relevant to sustained material experimentation. More importantly, there may be almost no selective advantage in possessing them. Orca intelligence evolved in an environment where locating prey, navigating enormous territories, coordinating hunts, maintaining social relationships, and learning from kin paid immediate returns. Evolution had no destination marked engineering.

Orcas are not failed humans. Their evolutionary history was never an attempt at civilization that stalled because nobody supplied hands. They matter here because they demonstrate how far intelligence and culture can develop along another path.

The physical environment nevertheless changes the paths available. Water contains abundant matter and no shortage of physical processes. Objects can be carried, struck, woven, piled, cut, or shaped underwater. But an animal living permanently in a dense moving fluid, particularly one without independent grasping appendages, faces a different set of low-cost material transformations from those available to terrestrial primates working with stone, wood, bone, fiber, hide, resins, clay, and eventually heat.

Fire illustrates the difference without explaining all of it. The usual claim that there can be no civilization underwater because there can be no fire starts too late. Hominins were modifying stone and other materials long before dependable fire production. What controlled fire eventually supplied was a cheap and unusually accessible expansion of the material repertoire. It could be approached and exploited before anything resembling sophisticated apparatus existed.

An aquatic intelligence might follow other routes. Hydrothermal chemistry, biological fabrication, electrochemistry, pressure, mineral precipitation, or processes we have not considered could conceivably become foundations for technology. Imagining what a mature aquatic civilization could do is the easy part, because we can quietly give it containers, controlled energy, precision instruments, accumulated knowledge, and then ask how it might use them.

The difficult question is how the first few dozen useful changes pay for the next few dozen.

Human technological history contains remarkably low rungs. Strike one stone with another. Keep the useful flake. Use an edge to work wood. Carry the shaped object. Discover that fiber can hold two materials together. Each operation alters slightly what can be attempted next. No participant needs a conception of the bow, the loom, the lathe, or the radio.

The Oceanic Filter is the possibility that some environments make this graph of reachable operations much thinner. Intelligence can still discover affordances, and culture can still preserve discoveries, but too few of the resulting capabilities become platforms for additional capabilities. The material side of the process repeatedly fails to deepen.

This is a filter, not a prohibition. There may be routes through it. They may simply be rarer, longer, or less rewarding at their earliest stages than the routes available to an upright primate walking around on dry land with two unusually capable hands.

What Reaches the Telescope

SETI encounters the far end of this problem because SETI does not detect intelligence directly.

Radio transmissions, lasers, industrial atmospheric chemistry, large engineered structures, waste heat, and other technosignatures are consequences of intelligence that has accumulated into sufficiently energetic technology to alter the observable world.[6] A social tradition cannot be seen from another star. Neither can an understood role in a hunt, a learned vocal repertoire, or an idea that never acquired a material path outward.

An intelligent species could therefore be culturally elaborate and technologically invisible. From another planetary system, the presence or absence of Earth's orcas makes effectively no difference to a technosignature search.

That observation does not solve the Fermi paradox. It matters only in proportion to how often intelligence arises in environments that punish deep technological composition, and we have no useful estimate of that fraction. Land worlds, coasts, and other environments with richer material pathways remain part of the problem.

Europa and Enceladus do not strengthen the case by serving as larger versions of Earth's oceans. Their seas are buried beneath ice and exist under radically different energetic, chemical, and physical conditions. We do not know whether either contains life. Their relevance is simply that extraterrestrial liquid water need not arrive with Earth's unusual combination of ocean, sunlight, atmosphere, exposed rock, continents, and shorelines.

The selection effect survives without knowing how common any of these environments are. The conditions that allow complex minds and social cultures to emerge need not be the same conditions that allow accumulated discoveries to become externally visible technology. SETI necessarily samples the latter population.

There may be much about orca minds that we do not understand. There may also be very little about them that points toward technological civilization. Those possibilities are perfectly compatible.

The Oceanic Filter does not tell us that the Milky Way is full of orcas. It identifies a way in which the galaxy could contain minds that never become astronomical objects. Our instruments can search for the traces left by an aggregation engine that has run extraordinarily far. They cannot tell us how many minds inhabit worlds where that engine never found enough connected rungs to begin the same climb.

Notes

[1] Michael N. Weiss et al., “Manufacture and use of allogrooming tools by wild killer whales,” Current Biology 35 (2025): R599–R600. The paper reports manufacture and cooperative use of modified bull kelp by southern resident killer whales.

[2] Isabella M. Reeves et al., “Ancestry testing of ‘Old Tom,’ a killer whale central to mutualistic interactions with human whalers,” Journal of Heredity 114 (2023): 598–611. The authors synthesize journals, anthropological records, newspaper accounts, Thaua Traditional Custodian knowledge, and other historical material. Their reconstruction describes orcas herding prey or alerting whalers by splashing and tail-slapping, whalers following the orcas, cooperative attack, and the later division of the carcass. They distinguish the documented commercial-whaling record from Thaua traditions describing an older relationship.

[3] José Z. Abramson et al., “Imitation of novel conspecific and human speech sounds in the killer whale (Orcinus orca),” Proceedings of the Royal Society B 285 (2018). Wikie produced recognizable copies of all tested novel human sounds within the experimental trial limit; “hello” and “one, two, three” were reproduced recognizably on the first trial.

[4] Gašper Beguš et al., “The phonology of sperm whale coda vowels,” Proceedings of the Royal Society B 293 (2026): 20252994. The authors report vowel-like coda-quality categories, duration contrasts, individual baselines, and adjacent-coda effects resembling coarticulation, while making no claim to have decoded coda semantics.

[5] Julian K. Finn, Tom Tregenza, and Mark D. Norman, “Defensive tool use in a coconut-carrying octopus,” Current Biology 19 (2009): R1069–R1070. Veined octopuses were observed transporting coconut-shell halves and later assembling them as shelter, despite the shells providing no protective benefit during transport.

[6] NASA, “Searching for Signs of Intelligent Life: Technosignatures.” The overview includes radio and laser emissions, artificial atmospheric chemicals, large engineered structures, and infrared waste heat among possible detectable consequences of technological activity.