The Clumsy Forest

Or, How Asimov Might Have Saved Kardashev a Lot of Time

Isaac Asimov gave his Galactic Empire an emperor, a capital, and a plan, but not a single shared mind. Hari Seldon's psychohistory worked by treating the movements of enormous populations statistically, not by pretending that an empire wanted something in the way a person does. When the Mule appeared—an exceptional actor the plan had not forseen—the future stopped behaving as scheduled. This was fiction, not evidence, but fiction can expose a category error before a spreadsheet does.

Detectability and concealment are coordination problems, and civilizations are not coordinators by default. A civilization may contain governments, firms, laboratories, machines, settlements, factions, and descendants that share an origin without agreeing on much. I call this the Clumsy Forest Hypothesis. Clumsy does not mean stupid or incapable. It means many capable parts without a permanently sovereign whole.

Nikolai Kardashev's famous scale classifies hypothetical civilizations by their rate of energy use: planetary, stellar, and galactic. For questions about detectability, this is a useful simplification. More power can produce stronger transmitters, more engineering, and more waste heat. But a wattage is not an agent. Access to stellar-scale energy does not automatically produce a stellar-scale will. ([1])

Kardashev gave us a spherical cow. For estimating how much power a distant observer might detect, the approximation has value. It becomes less useful when asked whether the cow is well governed or capable of agreeing not to moo.

The Dark Forest gives the cow motives. In Cixin Liu's formulation, civilizations seek survival, expand into finite resources, remain separated by chains of suspicion, and cannot know when another civilization will undergo a technological explosion. Under those conditions, silence or preemptive attack can appear rational. The uncertainty is real inside the model. The question is who, exactly, sustains the resulting policy. ([2])

The strongest answer does not require a planetary emperor. If exposure is fatal, every capable actor might infer the danger independently. Silence could emerge from convergent incentives rather than central enforcement. But that explanation still requires all actors capable of making the civilization visible to continue reaching substantially the same conclusion across changes in evidence, values, technology, and appetite for risk.

Some Fermi models state this requirement openly. The aestivation hypothesis proposes that advanced civilizations may expand, secure large resource domains, and then wait for a colder universe in which computation becomes more efficient. Its authors list among their assumptions that these civilizations have solved their coordination problems, and they explicitly treat them as Bostrom-style singletons: systems with a highest level of agency capable of resolving internal conflict. That honesty is useful. It puts the real question on the table. Should solved civilizational agency be our starting assumption? ([3])

Two Sides of a Threshold

Popular accounts often treat technological intelligence as though astronomical visibility follows automatically. It does not. A society can possess industry, computation, astronomy, and global communications without producing anything we could recognize from thousands of light-years away. A civilization can be extremely noisy to itself and silent to us.

Sofia Sheikh and her colleagues asked how far an exact technological counterpart of present-day Earth could detect our current technosignatures. Under their assumptions, an intermittent planetary-radar transmission could in principle be detected from roughly 12,000 light-years away if its narrow beam crossed the receiver at the right time. A persistent Deep Space Network uplink reached about 65 light-years, combined LTE leakage about four, and a Voyager-like transmission less than one. “We use radio” and “the Galaxy can hear us” are not remotely equivalent. ([4])

Below such thresholds, fragmentation may help explain why many technological societies remain obscure. Large and persistent signatures can require resources to be assembled, projects to survive political turnover, institutions to preserve their purpose, and machinery to operate long enough to matter astronomically. A civilization divided among short-lived governments, local priorities, and abandoned projects may accomplish remarkable things without sustaining the particular things that make it conspicuous from another star.

That is a tendency, not a law. Competition can produce arms races, prestige projects, reckless experiments, and waste on a heroic scale. Fragmentation can prevent one civilization-wide project while giving six rival factions excellent reasons to build their own. Its most defensible effect is not silence but variance: more stalled projects, more local projects, and occasionally more spectacular ones. Even then, what is spectacular nearby may remain negligible across interstellar distance.

This is the first Clumsy Forest regime. Many civilizations may remain locally capable because nothing ever binds enough capability, time, and purpose into a durable astronomical project. Many is not all, and telescopes do not necessarily show us the average.

When Ancestry Stops Being Government

Now suppose some faction crosses the threshold. It builds powerful transmitters, large propulsion systems, persistent industrial artifacts, or settlements beyond its home system. The same plurality that made conspicuous activity difficult begins multiplying the number of actors whose behavior must be controlled if the larger group is to hide.

Jason Wright and his colleagues made a closely related argument in SETI language in 2014. Once light-travel time between settlements becomes comparable to or longer than the timescale on which cultures change, they argue that the result should not be described as one civilization with one culture. It is a supercivilization: many civilizations whose shared characteristics come from common origin rather than current social coordination. Elsewhere in the same paper, they call the still broader habit of extending one culture's sociological explanation to every culture of every species the monocultural fallacy. The supercivilization point has a narrower consequence for this argument: a preference held by one culture cannot simply be treated as permanent policy for all its descendants. ([5])

Models of galactic settlement do not require a uniform empire steadily painting the Milky Way one color. Jonathan Carroll-Nellenback and colleagues found that settlement can spread under conservative travel assumptions while producing a patchwork of inhabited and uninhabited systems, particularly when settlements have finite lifetimes. These are models, not observations of alien behavior, but they break the easy equivalence between expansion and permanent political integration. ([6])

A settlement behind another star may share an origin without sharing a government. Decisions that cannot wait must be made locally, and every light-year adds a year to one-way communication. One branch may regard concealment as sacred, another as obsolete, and a third may never have received the most recent version of the rule. None needs to be stupid or suicidal. They need only possess different information, circumstances, or objectives.

Human concealment provides a limited analogy. Viet Cong tunnel networks hid positions and infrastructure from an active adversary. Japanese holdouts preserved orders and an internalized threat model after their active chains of command disappeared. These cases show that severe danger can sustain unusual silence without fresh instructions every morning. They do not show that concealment will survive unknown sensors, unknown opponents, and millennial timescales.

The more relevant pattern is partial consensus with leakage. The Manhattan Project concealed the purpose and details of an enormous program from most of the world, yet Soviet intelligence obtained consequential information through participants including Klaus Fuchs. Strong institutions can produce substantial secrecy. They do not thereby turn every participant into the same agent. ([7])

This is the second Clumsy Forest regime. Once conspicuous capability is distributed, perfect concealment requires either enforcement or durable alignment across every actor who can create a detectable exception. One sufficiently strong and intercepted transmission defeats perfect concealment for its source. It does not necessarily reveal every settlement in the lineage, but an informed dissident may expose connected sites, and ordinary movement or conflict may reveal relationships no one intended to publish.

The Hunter Is Part of the System

Concealment is not solely a property of the hidden civilization. If silence is inferred from the abstract logic of the Dark Forest, the convergence problem remains: different actors must continue deriving the same strategy. If silence is learned from actual predation, the incentive becomes easier to explain, but the conflict itself becomes harder to hide.

Search systems, weapons, propulsion, logistics, destroyed infrastructure, and altered environments can all create signatures. Civilization One may control its own transmitters without controlling the energy Civilization Two spends trying to find it. A disciplined prey can be revealed by a careless hunter. A disciplined hunter can be revealed by the shot.

This difficulty already appears in attempts to formalize Dark Forest reasoning. Chao Yu argues that attacking another civilization may expose the attacker to still more powerful observers and therefore shifts the safest strategy toward concealment. That may select for quieter survivors, but it adds requirements: attacks must eliminate visible targets, their physical consequences must not become easier to detect than the targets, and the surviving lineages must continue preserving the strategy. The relevant unit of analysis is no longer one civilization. It is the adversarial system. ([8])

The Exception Cuts Both Ways

This is where the Mule returns. An exceptional faction can break a strategy of universal concealment. Unfortunately for the Clumsy Forest, an exceptional faction can also break a population-level expectation of clumsiness.

A society need not preserve a policy for a million years if it can preserve it long enough to build something that does. A self-replicating probe, autonomous industrial seed, or sufficiently independent settlement can continue acting after its originating polity fragments, changes its mind, or disappears. Forgan takes as a starting premise the familiar implication: one civilization producing one successful self-replicating interstellar probe could generate a fleet that spreads through the Galaxy on timescales far shorter than the age of Earth. His model then asks whether predator-prey dynamics among such probes could keep them hidden and concludes that they would not. It does not establish that such a machine can be built or remain stable. The narrower inference here is mine: once agency has been transferred to an artifact, political discontinuity is no longer a sufficient answer to expansion. ([9])

The same problem applies to rare coordinated civilizations. A hive mind, long-lived AI, or society that physically centralizes all consequential technology might genuinely act as one agent for a very long time. These systems are not minor exceptions if detectability selects for them. One durable expansionist can affect more of the sky than a million societies that never leave their planets.

Robin Hanson and his colleagues build this selection into their model of “quiet” and “loud” aliens. Loud civilizations expand, visibly alter the volumes they occupy, and prevent comparable civilizations from emerging inside them. Their numerical estimates depend on contestable assumptions and need not be accepted here. Their further conclusion—that quiet civilizations must also be rare if loud aliens explain humanity's early arrival—is in genuine tension with a universe in which clumsy civilizations are common. The Clumsy Forest need not adopt their demographic model to inherit its observational warning: prevalence and observability are not the same distribution. Clumsy civilizations can be common and still fail to explain the sky if rare non-clumsy lineages dominate what a telescope could see. ([10])

The one-exception argument therefore cannot be used only against hiding. One dissident threatens universal silence. One singleton or replicator threatens universal clumsiness. Asimov's Mule was dangerous not because psychohistory had nothing to say about populations, but because a statistical plan was being asked to absorb the actor it could least afford to miss.

These are not peripheral caveats. They are why the Clumsy Forest is a correction to the unit of analysis rather than a complete solution to the Fermi Paradox. The symmetry is logical before it is observational: one exception defeats a universal claim about concealment or clumsiness, but it does not follow that humanity has detected the exception. Detection depends on what signature it creates, where it appears, how long it persists, and whether any search crosses it. Some signatures, moreover, do not require an exceptional actor at all.

Signatures Without Speakers

The word signature hides several different relationships between agency and evidence. An intentional transmission requires someone to choose, power, and aim it. A megaproject requires considerable organization before it exists, although it may later become autonomous or distributed. Waste heat is different. Once an industrial ecology uses enough energy, the signature follows from the physics whether anyone wants to make an announcement or not.

This is where Kardashev's spherical cow remains most useful. Wright and his colleagues argue that searches for very large energy supplies can rely more directly on conservation of energy and thermodynamics than searches for communication. They also note that a Dyson swarm need not be centrally planned; it might emerge from the accumulated energy demands of independent subcultures. Fragmentation could produce the very signature it had seemed likely to suppress.

Clumsiness can still matter before the fact. A society may never build a stellar-scale industrial system, or it may operate at much lower total throughput than the familiar Kardashev story assumes. But once a civilization is processing a significant fraction of a star's output, disagreement among its institutions does not make the waste heat disappear.

Nor does nondetection yet establish that no such systems exist. Project Hephaistos used Gaia and WISE observations to constrain modeled partial Dyson spheres radiating between 100 and 1000 kelvin. Its limits weaken for less complete swarms, greater distances, and temperatures outside the searched range. Communication SETI occupies another thin region of possible sky positions, times, frequencies, powers, and duty cycles. We have meaningful null results for particular signatures, not a completed census of the Galaxy. ([11]; [12])

These searches are therefore testing a different proposition from communication SETI. They do not ask whether a civilization chose to speak; they ask whether sustained energy use left consequences large enough to measure. The Clumsy Forest changes the expected organization behind the signature, not the physics of the signature itself.

What the Hypothesis Is For

Biology has repeatedly built cooperation, hierarchy, and higher-order individuals from previously independent parts. Nothing in the Clumsy Forest says civilization-scale agency is impossible. It says agency requires mechanisms, boundaries, enforcement, and persistence; it does not arrive automatically when a population acquires a collective noun.

The Fermi Paradox is the case through which the problem becomes visible; civilizational agency is the subject. The Clumsy Forest is best used as a discipline of explanation. “A civilization did X” is not an explanation until the coordinating mechanism is named and its persistence accounted for.

It also changes what we should expect. Below conspicuousness, fragmentation should produce variance in completed, long-lived projects rather than a default of either silence or spectacle. Above it, policy should drift as settlements separate, cultures change, and messages spend years in transit; ancestry is not government. Across both regimes, intentional beacons, autonomous replicators, and thermodynamic leftovers must be analyzed separately because they place agency in different locations—or, in the last case, may require no civilization-wide agent at all.

A clumsy forest is not an empty one. It is a forest in which most activity may remain local, occasional concentrations of power fail to become durable, and no lineage can assume that every descendant, rival, or pursuer will preserve its policy. The silence may tell us less about universal fear than about how difficult it is for complexity to become a single voice. Before a sentence tells us what “a civilization” chose, it must show us how the civilization became the kind of thing that could choose it.

Notes

[1] Kardashev, “Transmission of Information by Extraterrestrial Civilizations”

[2] Official excerpt from Cixin Liu's The Dark Forest

[3] Sandberg, Armstrong, and Ćirković, “That is not dead which can eternal lie,” 2017

[4] Sheikh et al., “Earth Detecting Earth,” 2025

[5] Wright et al., “The Ĝ Infrared Search for Extraterrestrial Civilizations with Large Energy Supplies. I,” 2014

[6] Carroll-Nellenback et al., “The Fermi Paradox and the Aurora Effect,” 2019

[7] U.S. Department of Energy, “Espionage and the Manhattan Project, 1940–1945”

[8] Yu, “The Dark Forest Rule,” 2015

[9] Forgan, “Predator-Prey Behaviour in Self-Replicating Interstellar Probes,” 2019

[10] Hanson et al., “If Loud Aliens Explain Human Earliness, Quiet Aliens Are Also Rare,” 2021

[11] Suazo et al., “Project Hephaistos I,” 2022

[12] Wright, Kanodia, and Lubar, “How Much SETI Has Been Done?”, 2018