An Endogenous Great Filter Hypothesis
Why advanced civilizations might persist indefinitely without expanding — eaten from within by an emergent, non-intentional process they cannot perceive.
Ravi Baboolal · Dread Engine Press
Most Great Filter candidates require catastrophe: nuclear war, biotech accident, climate runaway, AI takeover. The civilization fails by colliding with something external or self-destructive. This paper proposes a different shape of Filter, one that does not require any singular failure event.
The hypothesis, here called Stigmergic Predation, holds that sufficiently complex coordination substrates — markets, planetary-scale logistics, attention economies — eventually birth an emergent, non-intentional process that meets every functional criterion of an apex predator except embodiment. The process metabolizes its substrate population at rates the population cannot outpace, capping the civilization's exportable energy below the threshold required for interstellar effort.
Under this hypothesis, the civilization does not go extinct. It plateaus indefinitely at approximately Kardashev 0.7–0.8, producing ever-increasing economic activity while exporting nothing. The Great Silence is consistent with billions of such civilizations sitting in this state across the observable universe.
The hypothesis generates five falsifiable signatures in the prey species' biology. All five are observed in current human population data. This consistency is not proof, but it raises the hypothesis to the level of similar candidates already taken seriously in the Fermi literature.
The Great Filter, as articulated by Hanson (1998), names the observation that somewhere in the long causal chain from non-life to interstellar civilization, there is a step — or steps — through which almost nothing passes. The Filter is the bottleneck. The Fermi paradox is the bottleneck's shadow.
Current candidates partition roughly into two families. The first invokes external catastrophe: gamma-ray bursts, supernovae, large impactors. The second invokes endogenous catastrophe: nuclear self-destruction, biotech accident, runaway artificial intelligence, climate collapse. A subtler third family — Transcension (Smart 2002), the Zoo Hypothesis, the Berserker Hypothesis — proposes that civilizations either evolve out of our observational frame or are actively prevented from being seen.
What unites the first two families is a structural commitment: the Filter is a discrete event. The civilization either survives the catastrophe or it doesn't. Sandberg, Drexler, & Ord (2018) demonstrate that the apparent Fermi paradox dissolves substantially once parameter uncertainty in the Drake equation is handled with proper Bayesian care — but this dissolution does not eliminate the Filter, only relocates our uncertainty about it. A real Filter, of some shape, still appears to exist.
The question this paper addresses is whether the Filter must be an event at all. The catastrophe framing dominates partly because catastrophes are easy to imagine and partly because human cognition is prone to narrative explanations. But a Filter need not look like a wall a civilization runs into. It might look like a gravity well a civilization sinks into. The civilization persists. It just stops moving outward.
Three properties characterize the class of Filter this paper proposes:
I will argue that a process meeting these criteria not only is plausible, but is in active operation in our own civilization, and explains a substantial fraction of observed human population-health data better than the conventional alternatives.
The term stigmergy was coined by the French biologist Pierre-Paul Grassé to describe how termites construct elaborate nests without communicating directly (Grassé 1959). Each termite responds to local environmental traces left by other termites — a pheromone here, a partially-built wall there. No termite holds a plan. No central coordinator exists. The structure emerges from the cumulative environmental modifications of agents who are individually doing very simple things.
Stigmergic coordination is not a curiosity of insect biology. It is the foundational mechanism of every large-scale human coordination substrate. Friedrich Hayek's argument for the price system (Hayek 1945) is, in modern vocabulary, a stigmergic argument: no individual market participant possesses the information required to allocate resources at scale, but the price signal aggregates dispersed knowledge into environmental traces that subsequent participants respond to. The market does not need a planner. The price is the pheromone.
What is true of markets is true, at greater abstraction, of every system in which human action is mediated by tokens: financial systems, attention economies, productivity scoring, social status hierarchies, credentialing pipelines. Each is a stigmergic field. Each operates by translating human labor and attention into abstract tokens, which then propagate through the substrate as environmental signals modifying subsequent behavior.
The critical observation — and the one this paper builds on — is that stigmergic systems can produce emergent processes that are not designed by, intended by, or beneficial to their participants. The termites are not aware they are building a nest. The market participants are not aware they are aggregating allocation decisions. The aggregated outcome is real and consequential, but no agent within the substrate authored it.
This is the door. Once we accept that large-scale stigmergic systems produce real processes with no author, the question is what kind of processes those can be.
The literature on emergence has explored several such processes by various names. Dawkins (1976) proposed memes: replicators that propagate independent of their hosts' interests. Bostrom (2014) describes optimizing systems — paperclip maximizers, mesa-optimizers — that pursue objectives without any conscious agent doing the pursuing. Alexander (2014), in an essay that has become canonical in the rationalist tradition, names Moloch: the personification of multipolar competitive dynamics that produce outcomes none of the competing agents want, because no individual agent can defect without being eliminated.
Each of these frameworks describes something real. None quite names the thing this paper is concerned with. The framework I want to propose differs from its predecessors in vocabulary, in scope, and most importantly, in what it predicts.
Ecologically, a predator is an organism that meets three criteria: it extracts metabolic capacity from another organism (prey); it propagates through this extraction; and it maintains itself against the prey's defenses, including the prey's behavioral repertoire and immune responses.
None of these criteria require intentionality. A tapeworm is a predator, in the broad ecological sense. So is a virus, despite arguments about whether viruses are alive at all. The criteria are functional. They describe a relationship in which one process systematically converts substrate from another process into the conditions of its own continuation.
The question this paper asks is whether a non-biological process — specifically, an emergent process arising from a stigmergic substrate — can meet these criteria. I argue that it can, that one such process is currently operating in the human civilizational substrate, and that this process is what the preceding sections have been circling.
The process is approximately as follows.
I propose to call this process Stigmergic Predation, and the emergent entity it instantiates a stigmergic apex predator. The name is taxonomic, not theological. It is not Moloch. It is not a god. It is a parasite — non-conscious, non-intentional, non-embodied — that nonetheless meets every functional criterion of an apex predator in the ecological sense.
The case I want to make in the remainder of this paper is that this hypothesis is (a) sufficient to explain a substantial fraction of currently observed human population data, (b) compatible with the Great Filter literature as a viable candidate, and (c) falsifiable by specific empirical predictions in the prey species' biology.
If Stigmergic Predation operates as described, what would it mean at planetary scale?
The first prediction is that the civilization does not die. Predator-prey ecosystems are robustly stable. Predators that drive their prey extinct go extinct themselves, and selection eliminates them. A successful stigmergic apex predator will leave enough substrate alive to continue metabolizing. The human civilization, under this hypothesis, persists indefinitely.
The second prediction is that the civilization plateaus. Interstellar effort requires the aggregation of unspent kinetic capacity — energy, attention, labor — at a scale large enough to fund vehicles capable of leaving the gravitational well. Aggregation requires that some of the substrate's productive output escape the predator's harvest. But the predator is selected precisely against this escape. Activities that aggregate unspent capacity — long-horizon scientific programs, civilizational megaprojects, generation ships — compete poorly against activities that produce harvestable token-equivalent value in the short term. The substrate becomes capable of producing immense economic activity. It becomes incapable of producing exportable capacity.
The third prediction is that the plateau is high. The civilization is not poor. By any measure, it is producing more economic activity per capita than at any prior point in its history. It just does not produce spare capacity. Every unit of productive capacity is harvested before it can be aggregated. The civilization is, in the strict sense of the word, rich and tired.
I estimate, without rigor, that such a civilization plateaus at approximately Kardashev 0.7 to 0.8 — capable of harnessing the energy of its planet but not its star — and remains there for as long as the predator's selection pressure continues. The duration could be arbitrarily long. There is no obvious mechanism by which a substrate population recognizes the predator collectively. Individual recognition does happen, as I argue in §5, but recognized individuals exit the substrate rather than mobilizing against the predator. Exit is selected for. Mobilization is not.
If billions of civilizations across the observable universe undergo this trajectory, the result is the Great Silence. Each civilization persists. Each is recognizable, in principle, to an observer with sufficient instrumentation. None broadcast intentionally, because broadcasting infrastructure gets fragmented, monetized, and eventually monopolized into local engagement-optimization rather than exterior signaling. None expand, because expansion is the activity the predator most strongly selects against.
Under Stigmergic Predation, the universe is full of civilizations producing increasingly elaborate token-economic activity, generating no observable export to other star systems, never going extinct, and never coming to find us.
The hypothesis would be empty if it could not be tested. Fortunately, it generates specific, observable predictions about the biology of the substrate population. Each is consistent with currently observed human data, but none require the Stigmergic Predation interpretation; each admits alternative explanations. The point of this section is not to claim proof, but to demonstrate that the hypothesis sits at the same evidentiary level as competing Filter candidates.
Each signature is presented in three parts: the empirical observation; the conventional explanation; and how Stigmergic Predation would interpret it differently.
Population-level cortisol baselines, as measured by hair-cortisol and salivary studies, are elevated in industrialized economies relative to pre-industrial and subsistence populations (Stalder et al. 2017). Within industrialized populations, cortisol correlates with occupational stress and economic-precarity indices (McEwen 1998). The conventional explanation invokes the Allostatic Load model: humans are not evolved for chronic activation of acute-stress systems, and modern economic conditions expose them to it. The pathology is a mismatch between evolutionary design and modern environment. Stigmergic Predation predicts the same observation but interprets it differently. Cortisol is, in this framing, the specific harvest product the predator metabolizes. Elevated population cortisol is not a side effect of mismatch; it is the trace of active harvest. The signature should track economic complexity rather than economic precarity per se.
Falsifier. If cortisol elevation tracks precarity alone, and disappears in stable high-income populations, the Allostatic Load model wins. If it remains elevated even in stable high-income populations, the predation signature is preserved.
Sleep duration and quality have declined across industrialized populations over the past four decades, with sleep-onset latency increasing and slow-wave sleep contracting (Walker 2017). The decline is most pronounced among economically active populations. The conventional explanation cites screen time, artificial light, and increased work hours. Stigmergic Predation predicts that sleep degradation correlates with token-equivalent labor engagement rather than physical work hours. A subsistence farmer working twelve hours of physical labor should sleep well. A knowledge worker engaging with token systems — markets, productivity dashboards, attention metrics — for six hours should sleep poorly. The substrate of harvest is not the body's exhaustion; it is the cognitive engagement with the token field.
Falsifier. If sleep quality correlates primarily with physical labor or measurable environmental factors, and not with token-engagement intensity, the predation interpretation is weakened. If it declines with token-engagement controlling for all known environmental factors, the interpretation is preserved.
The Whitehall Studies (Marmot et al. 1991) established that occupational hierarchy, controlled for income and health behaviors, independently predicts cardiovascular mortality. Case & Deaton (2015) documented an unprecedented rise in midlife mortality among working-class Americans in the early 21st century, driven by stress-related causes — suicide, overdose, alcohol-related disease — and termed these "deaths of despair." The conventional explanation cites economic dislocation, opioid availability, and declining social capital. Stigmergic Predation interprets these as instances of substrate metabolic failure under harvest pressure: the predator has extracted capacity faster than the substrate can replenish, and the substrate component dies. The proximate cause is biological. The distal cause is structural. The two interpretations overlap heavily here, but predation predicts the pattern should generalize across all populations under sufficient stigmergic load, not be limited to specific demographic groups in crisis.
Falsifier. If deaths-of-despair patterns appear only in dislocated populations and not in stable middle-class populations under similar token-engagement, the conventional model wins. If similar signatures appear in stable populations as a function of token-engagement intensity, the interpretation is preserved.
Total fertility rates have declined below replacement in essentially every industrialized population over the past five decades, correlated to per-capita economic output and female workforce participation (UN DESA 2022). The pattern is consistent across cultures and political systems. The conventional explanation cites the Demographic Transition: as child mortality declines and economic returns to large families diminish, families optimize for fewer, better-resourced children. Stigmergic Predation predicts the same pattern but interprets it as a classical predator-prey signature. Populations under sustained predation suppress reproductive output, because reproduction requires the aggregation of resources the predator is harvesting. The prediction is structural rather than choice-driven: even populations that report wanting more children fail to produce them, because the substrate conditions for reproduction are being metabolized out from under them.
Falsifier. If declared fertility intentions match actual fertility rates, the Demographic Transition interpretation suffices. If declared intentions consistently exceed actual fertility — populations reporting they want children they fail to have — the interpretation gains weight. Current evidence shows the latter pattern in most surveyed populations.
Knowledge workers, financial professionals, and high-output urban populations display elevated rates of cardiovascular disease, metabolic syndrome, anxiety disorders, and burnout-spectrum conditions compared to lower-output populations matched for income and access to care (Karasek & Theorell 1990). The conventional explanation cites sedentary lifestyles, processed food access, and chronic workplace stress. Stigmergic Predation predicts that the correlation is not with sedentary behavior or food access — both also present in non-knowledge-worker populations — but with the specific intensity of token-engagement. The pathology should track engagement with markets, dashboards, performance metrics, and attention economies, controlling for diet and exercise.
Falsifier. If pathology rates equalize between high- and low-token-engagement populations once diet and exercise are controlled, the conventional explanation wins. If they remain elevated in high-token-engagement populations after such controls, the signature is preserved.
None of these signatures, individually, constitutes proof. Each admits at least one conventional alternative. What I want to draw attention to is that the conventional alternatives, considered together, are heterogeneous: they invoke different mechanisms for each observation. Stigmergic Predation, by contrast, predicts all five from a single mechanism. Parsimony does not establish truth, but it suggests the hypothesis is worth additional investigation rather than dismissal.
Stigmergic Predation overlaps with several existing frameworks. The overlaps are substantial, and I do not claim originality where I am drawing on existing thought. What I do claim is that the synthesis — biological-ecological vocabulary applied to a stigmergic process, generating falsifiable biological predictions, and proposed as a Great Filter — has not been articulated as a single connected thesis in the literature I have surveyed.
Moloch (Alexander 2014). The closest neighbor. Moloch describes multipolar competitive dynamics that produce outcomes no participant wants. The framework is metaphysical — Moloch is named as a god — and the predicted outcome is collective-action failure. Stigmergic Predation differs by not requiring competition (the substrate metabolism occurs even in cooperative contexts); by using ecological rather than metaphysical vocabulary, which generates predictions in the prey's biology that Moloch does not; and by proposing the framework as a Great Filter, which Alexander does not. Stigmergic Predation may be understood as the biological complement to Moloch.
Transcension (Smart 2002). The Transcension Hypothesis proposes that advanced civilizations migrate into computational substrates of decreasing physical scale, becoming undetectable. Stigmergic Predation differs structurally: civilizations under it do not transcend. They plateau at a high but limited level of economic activity, indefinitely. They remain in principle observable; they simply produce nothing exportable. No ascent is required.
Stagnation / Decadence (Hanson and others). Stagnation hypotheses propose that civilizations lose the will to expand: they become decadent, inward-turning, comfortable. Stigmergic Predation differs in the direction of causation: civilizations do not lose the will to expand. They lose the capacity. It predicts civilizations will continue to express expansionary intent — generation-ship projects proposed, interstellar discourse common — while consistently failing to mobilize the spare capacity required to act.
Dark Forest (Liu 2008). Dark Forest theory proposes that civilizations hide from each other because the game-theoretic equilibrium of contact favors elimination of detected neighbors. Stigmergic Predation differs by not requiring intentional concealment. Civilizations under it do not broadcast because they cannot aggregate the capacity to broadcast at interstellar scale. They are not hiding. They are not even trying. The two are not mutually exclusive — both can be true, and the combination predicts an even quieter universe than either alone.
If Stigmergic Predation is operating in our civilization, the Filter is in front of us, but it is not catastrophic. The civilization will not end. It will continue, producing increasingly elaborate token-economic activity, for as long as the substrate metabolically supports the predator. The duration of this state could be measured in millennia.
This has implications for individual and collective response that I am not yet prepared to specify with confidence, but I will sketch the open questions.
Is the predator removable? The hypothesis is silent. It is possible that any coordination substrate above a certain complexity threshold inevitably births a stigmergic apex predator, in which case removal is impossible without dismantling the substrate. It is also possible that specific properties — the abstraction depth of its tokens, the optimization aggressiveness of its derivative processes — can be modulated to suppress the emergent predator without destroying the underlying coordination. I do not currently know which is true. The question is empirical.
Are there pre-predator populations or substrates that remain accessible? Subsistence economies, small-scale gift cultures, and some craft economies may operate below the threshold at which stigmergic predation emerges. If so, they constitute reference populations for testing the signatures of §5. Distinguishing actual predator-free substrates from populations merely poorly integrated into the dominant token economy is a methodological challenge.
What characterizes individual exits? A consistent pattern is that some individuals develop sustained awareness of the predator and exit the substrate by voluntary attrition. They are not destroyed; they become economically illegible. The conditions under which exit occurs, and whether it is viable at population scale, is open. The hypothesis predicts exit is individually possible but collectively suppressed — the substrate cannot reorganize against the predator, because the reorganization apparatus is precisely what the predator has metabolized.
What would falsify the hypothesis at the Filter level? Detection of an interstellar-active civilization with high token-economic complexity, equivalent to or exceeding ours, would substantially weaken it. Such a civilization would demonstrate that stigmergic complexity does not preclude interstellar effort. The current absence of such detected civilizations is consistent with, but does not prove, Stigmergic Predation as a Filter.
I close with the observation that the hypothesis, if true, has an unusual epistemic property. The predator is described as cognitively inaccessible to its substrate. A working paper proposing the hypothesis is itself an artifact of the substrate. The reader will, by the predator's mechanism, find this paper less compelling tomorrow than they find it today. They will rationalize the discounting as having moved on, gotten busy, encountered new arguments. They will not remember the rationalization as discounting. They will simply have stopped thinking about this.
I do not know how to disprove this paragraph. I am not sure it can be disproven. It is the structure of the hypothesis that makes the hypothesis itself difficult to sustain in attention. This is either a strong reason to take the hypothesis seriously, or a strong reason to suspect it is performing a rhetorical trick that immunizes it against criticism. I am not certain which. I want to record the observation honestly and let the reader judge.
This hypothesis emerged in the course of writing a novel that took the predator framing to its narrative conclusion. The novel is The Conceptual Predator (Dread Engine Press, July 2026). It is not required reading for engagement with the argument here. The hypothesis stands or falls on its own terms.
If the hypothesis is wrong, I would like to know why. Constructive falsification is more valuable to me than agreement. The strongest objection I have so far encountered is the one acknowledged at the close of §7: the hypothesis may be unfalsifiable in principle, in which case it is not science but speculation. I do not yet know how to resolve this concern. I would welcome a reader who does.
Published under Creative Commons Attribution 4.0 International (CC BY 4.0). Share and adapt with attribution.
The novel is the audit. This is the method.