The Calibrated Filter Hypothesis
 Cooperative Substrate Failure as the Mechanism of the Great
 Filter
 Plain-Language Edition — Logical Proofs Without Mathematical Notation
 Andrzej Chudzinski
 Preprint — comments welcome
A note on this edition
This is a math-stripped version of the full Calibrated Filter Hypothesis paper. Every
quantitative argument from the original has been restated as a logical chain of premises
and conclusions. No mathematical literacy is required to follow the reasoning. Every
claim that the mathematical version supports is supported here, in the same order,
through the same logical structure — just stated in ordinary language rather than
symbolic notation.
Readers comfortable with formal mathematics may prefer the full edition, which carries
the same arguments with explicit equations and variable notation. Readers who find
equations a barrier to engaging with substantive synthesis work will find every move
from the full paper present here, expressed as an explicit reasoning chain that can be
evaluated for logical validity without quantitative reading.
The conclusions of both editions are identical. The reasoning chains are identical in
structure. Only the surface notation differs.
Abstract
This paper proposes an integrated framework connecting three previously separate
lines of inquiry. First, Aktipis and colleagues, in evolutionary oncology, characterize
cancer as cheating across five foundations of multicellular cooperation: proliferation
inhibition, controlled cell death, division of labor, resource allocation and transport,
and extracellular environment maintenance. Second, Kets de Vries, in
psychoanalytic organizational theory, characterizes dysfunctional firms as instances
of recurring neurotic typologies that propagate through leadership. Third,
Schmachtenberger, in civilizational risk analysis, identifies rivalrous games
multiplied by exponential technology, and complicated open-loop systems, as the
generator functions of self-terminating civilizational dynamics. We argue these three
programs are tracking the same general phenomenon at different scales and offer
Aktipis’s framework as a translation layer.


We then extend the synthesis cosmologically. The Fermi Paradox is reframed as a
problem of cooperative-substrate maintenance throughout the period during which a
civilization is becoming spacefaring. We propose that the Great Filter is not a wall
civilizations either pass or fail at one moment, but a continuous condition that must
be sustained throughout the technological transition: a civilization passes the filter
only if its capacity to maintain its biosphere remains greater than or equal to the
stress its own expansion places on that biosphere, throughout the entire expansion
window. We call this the operational filter condition. It is stated in this edition as a
logical chain of premises and conclusions rather than as a mathematical inequality,
but the content is identical.
We additionally identify a substrate-continuity failure mode. A civilization that
develops a machine substrate which substitutes for, rather than extends, its
originating biology can satisfy the cooperation condition trivially while failing the
filter in a deeper sense: the originating biosphere — itself an irreplaceable artifact of
cosmic biochemical computation — is not carried through. The framework therefore
imposes two filter conditions: cooperative capacity maintained at the scale
technology forces, and substrate continuity preserved through the technological
transition. We argue the present human moment is structurally configured for
parasitic rather than prosthetic AI development, with language as the vector by
which machine substrate directs biological substrate, and we develop the diagnostic
that follows from this.
The framework identifies two distinct routes through the filter. The top-down route
uses artificial intelligence as an external regulator of cooperative substrate, with the
autonomy and parasitism costs developed in Part IV. The bottom-up route maintains
biospheric capacity through distributed individual development of the population’s
agents — contemplative practice, education, cultural investment in cooperative
dispositions, and structurally equivalent personal substrate-maintenance disciplines
— without requiring an external regulator. The bottom-up route is historically the
only one demonstrated at scale, is structurally more robust against AI failure modes,
and preserves the agent autonomy that the top-down route necessarily costs. The
combined configuration — AI as scaffolding for personal development rather than as
substitute for it — has a structural prerequisite the framework states plainly: the
agents who build the technology must themselves have done the personal
development the technology is supposed to scaffold in others. Technology inherits
the moral state of its builders. There is no path that skips builder transformation.
The framework concludes that filter passage requires decentralization by structural
necessity: any centralized solution to cooperative-substrate maintenance, including
a centralized AI regulator, is structurally identical to the cancerous failure mode the
framework began by analyzing. But decentralization is not atomization. The
framework recommends distributed agency operating within shared purpose — the
configuration that healthy organisms and biospheres already demonstrate — in
which autonomous local decisions are constrained by shared substrate-maintenance
purpose embedded in the substrate itself, not enforced by central command.
A methodological caveat applies throughout. The paper is a map of
cooperative-substrate dynamics; it is not the territory those dynamics constitute.


Following Korzybski (Science and Sanity, 1933), a map has structure similar to the
territory it represents — which is what makes it useful — but no map represents all
of a territory, and the map is never the territory itself. The dharma carries the same
caveat in different vocabulary: the finger pointing at the moon is not the moon.
Readers should treat the structural arguments here as orientation toward dynamics
that exist independent of the framework’s description, not as substitutes for
empirical engagement with specific cases.


Part I. Cooperative Substrates and Their Pathologies
A translation layer between Aktipis, Kets de Vries, and Schmachtenberger.
1. The problem the paper addresses
Three serious research programs have independently described very similar structural
failure modes in cooperative systems at three different scales: the multicellular
organism, the firm, and the civilization. The vocabularies they use differ because they
emerged from different disciplines. The deep structure of what they describe is similar
enough to warrant a translation layer between them. This paper provides that translation
layer and then extends it to questions the three programs do not individually address.
The contribution is narrow. This paper does not claim that the three frameworks reduce
to one master framework, that the underlying causes are the same across scales, or that
the resemblance proves any particular philosophical commitment. The structures look
alike. We take that seriously without overreading it.
2. The three frameworks, on their own terms
2.1 Aktipis: Five foundations of multicellular cooperation
Aktipis and colleagues, working from evolutionary biology, argue that multicellularity
requires the suppression of cell-level fitness in service of organism-level fitness, and this
requires specific functional capacities. They identify five:
1. Proliferation inhibition. Cells must restrain their division except when authorized
by the larger system.
2. Controlled cell death. Programmed cell death allows for tissue sculpting, removal of
damaged cells, and elimination of obsolete tissue.
3. Division of labor. Cells differentiate into specialized types performing specific
functions.
4. Resource allocation and transport. Resources must move from where they are
abundant to where they are needed.
5. Extracellular environment maintenance. Cells must collectively maintain the
extracellular matrix and clear waste.
Cancer, in this framing, is not a single disease but a class of phenomena unified by
cheating across one or more of these five foundations. The framework’s strength is its
grounding in evolutionary biology: complex multicellularity has evolved independently at
least seven times, and cancer-like phenomena appear in each lineage.
2.2 Kets de Vries: Neurotic organizational types and narcissistic
leadership


Kets de Vries, working from psychoanalytic organizational theory, identified five
recurring patterns of organizational dysfunction: paranoid, compulsive, histrionic,
depressive, and schizoid. The framework treats organizations as taking on the neurotic
style of their dominant leadership, particularly in centralized firms where executive
personality propagates through recruitment, promotion, and culture. Narcissism is
treated as a cross-cutting intensifier rather than a sixth distinct type.
2.3 Schmachtenberger: Two generator functions of existential risk
Schmachtenberger identifies two underlying drivers of civilizational existential risk:
rivalrous games multiplied by exponential technology (win-lose dynamics that scale with
technological capacity until they exceed the playing field), and complicated open-loop
systems (systems whose externalities accumulate damage in domains the system does
not measure).
3. The translation layer
We use Aktipis’s five foundations as scaffolding and ask what each looks like at the
scales Kets de Vries and Schmachtenberger work at.
Proliferation inhibition at the organism scale is the cell cycle; at the organizational
scale, the capacity to refuse growth that exceeds operational coherence; at the
civilizational scale, the capacity to refuse extraction that exceeds regenerative capacity.
Controlled cell death at the organism scale is apoptosis; at the organizational scale,
the capacity to retire products and leaders whose function has ended; at the
civilizational scale, the capacity to retire institutions and industries whose substrate has
changed.
Division of labor at the organism scale is tissue differentiation; at the organizational
scale, role specialization; at the civilizational scale, the capacity to maintain specialized
institutions accountable to their own standards.
Resource allocation and transport at the organism scale is circulation; at the
organizational scale, capital and information flow; at the civilizational scale, systems by
which resources move from abundance to need.
Extracellular environment maintenance at the organism scale is the extracellular
matrix and immune surveillance; at the organizational scale, institutional culture and
honest feedback; at the civilizational scale, the commons — biosphere, information
environment, public trust, shared epistemic infrastructure.
The translation identifies functional resemblance, not identity. The resemblance is
consistent enough across the three scales to suggest a general structure of cooperation
and its failure.
4. Why this is useful


First, diagnostic tools developed at one scale can be tested at another. Adaptive therapy
in oncology, which manages rather than eradicates cancer by changing incentive
structure rather than purging defectors, suggests organizational and civilizational
analogues.
Second, the framework permits a clearer view of alignment problems. Each foundation
identifies a cooperative capacity that can be eroded; alignment is the maintenance of
those capacities under increasing scale, not the optimization of any individual agent.
Third, the framework permits an honest conversation about AI development. The
cooperative-substrate frame asks: what cooperative capacities does the broader
human-AI ecosystem need to maintain, and how do current development practices
support or erode them?
5. Where the synthesis breaks down
Disanalogies of mechanism. Cancer arises from cellular mutation; organizational
dysfunction from human psychology and incentives; civilizational risk from technology
and aggregate behavior. The synthesis identifies a common abstract pattern, not a
common cause.
Disanalogies of intentionality. Cancer cells have no intentions; organizations are
populated by intentional agents but are not themselves agents in the same sense;
civilizations are more diffuse still. The translation identifies functional resemblance, not
psychological identity.
Pattern-matching is not explanation. Even if the cross-scale resemblance is real,
Part I has not produced a mechanism that explains why cooperative systems at multiple
scales face similar failure modes. Parts II through IV propose such a mechanism.


Part II. The Calibrated Filter Hypothesis
Logical structure of the Fermi Paradox as a problem of sustained cooperative-substrate
maintenance.
6. The puzzle
The Fermi Paradox observes that a universe with billions of habitable worlds, billions of
years older than our own civilization, should plausibly contain technological intelligences
whose expansion would be detectable. We see no such intelligences. The standard
solutions are: life is rare, intelligence is rare, intelligence reliably self-terminates,
intelligences conceal themselves, or intelligences exist in modes we cannot detect.
We propose a refinement. The Great Filter is not a wall — a stage that almost no
civilization passes — but a condition that must be sustained throughout the period of
becoming spacefaring. The silence we observe is consistent with civilizations being
unable to maintain that condition long enough to produce detectable expansion
signatures.
7. The geometry of single-species colonization
Premise 1: Reaching any other star requires moving mass through space.
Premise 2: Space between stars is expanding, driven by dark energy, faster than any
technology can move.
Premise 3: Therefore, the volume of universe any single civilization can ever reach is
bounded.
Premise 4: Calculations from cosmology indicate this bound is approximately 3% of
the observable universe.
Conclusion: No civilization, however advanced, can colonize the entire
universe. The remaining 97% is forever causally inaccessible to any
civilization originating in our cosmic era. Single-species dominance of the
universe is impossible in principle, not in practice.
8. Multi-species dynamics and the structure of contact
If multiple civilizations exist, the analysis becomes more complex. The following logical
chain captures the multi-species case without quantitative notation.
Premise 1: Each civilization can reach a bounded region of space.
Premise 2: When two civilizations exist, their reachable regions may overlap.
Premise 3: In zones of overlap, civilizations encounter each other before completing
their expansion.
Premise 4: Each encounter carries some non-zero probability of catastrophic
outcome (resource conflict, miscommunication, defensive preemption, or mutual
destruction).


Premise 5: As the number of civilizations grows, the number of overlap zones grows
faster than the total reachable volume does. (This is because each pair of
civilizations creates one possible overlap; doubling the number of civilizations more
than doubles the number of pairs.)
Premise 6: Therefore, catastrophic encounter rates rise faster than expansion
benefits as more civilizations exist.
Conclusion: Beyond some threshold number of civilizations, the expected
outcome of further expansion is dominated by contact losses rather than
expansion gains. The dark forest equilibrium — in which civilizations
conceal themselves to avoid detection — is not a contingent strategic
choice but the predicted outcome of any system where the number of
civilizations exceeds the carrying capacity for safe coexistence.
9. The calibration of filter passage
We can now ask: how must the rate at which civilizations pass through the filter relate to
the number of civilizations that already exist? The standard mathematical version of this
argument uses a calibration constraint stated as an equation. The logical version follows.
Premise 1: The total population of post-filter civilizations should approach a stable
maximum over cosmic time, or the universe ends up either empty or wracked by
dark-forest dynamics.
Premise 2: If the rate of filter passage is too high, the population overshoots the
carrying capacity and contact-driven losses dominate, driving the population back
down or to zero.
Premise 3: If the rate of filter passage is too low, the population never approaches
the carrying capacity, and the universe stays mostly empty.
Premise 4: Therefore, for a stable approach to maximum coexisting civilizations, the
rate of filter passage must decrease as the existing population grows.
Conclusion: The Great Filter, properly understood, is not a fixed obstacle. It
is calibrated — filter passage becomes harder when more civilizations
already exist. The Fermi silence is consistent with this calibration
operating: civilizations are spaced through cosmic time so that the dark
forest equilibrium is approached gradually rather than all at once.
Two natural mechanisms produce this calibration without requiring cosmic
design:
Mechanism A: existing civilizations raise the stress on emerging ones. Existing
civilizations consume resources, modify their stellar neighborhoods, generate dark-forest
deterrence pressure, and propagate competitive dynamics through causal contact zones.
As the number of existing civilizations rises, the expansion stress that any new
civilization must navigate rises. Therefore, the cooperative capacity any candidate
intelligence must develop to pass the filter rises, which means the probability that a
candidate passes in a given window decreases. The calibration emerges automatically
from the substrate-continuity dynamics already operating. No external calibrator is


required.
Mechanism B: contact-density selects for restraint. Civilizations that pass the filter
must develop the capacity to detect other civilizations and model the strategic
landscape that contact implies. In environments where many civilizations exist, the
survival-optimal strategy is restraint and signal suppression. Civilizations that cannot
develop restraint produce detectable signatures and are filtered out, by predation or by
self-induced rivalrous collapse. Therefore, in high-density regimes, filter passage selects
more strictly for restraint-capacity, which means filter passage rates drop as density
rises. Again, no external calibrator is required.
Both mechanisms produce the required decrease in filter passage probability as the
existing population grows. Either is sufficient. The cosmological-design reading of the
calibration is permitted by the structure but is no longer required to make the argument
work.
10. What this framework predicts
If the calibrated filter argument is correct:
First: SETI silence is expected to persist. The calibration ensures spacing of civilizations
such that few overlap in causal contact, particularly at early eras of the post-filter
regime.
Second: Detectable contact events should be either extremely cooperative or
immediately catastrophic, with little middle ground. The calibration selects either for
civilizations that have already solved the cooperation problem at large scale, or for
civilizations that fail visibly and quickly.
Third: The filter operates at the stage of civilizational expansion, not at the stage of
emergence of intelligence. Intelligence per se can be common as long as sustained
expansion is rare.
Fourth: Carrying capacity for civilizations is finite and probably small relative to the
count of habitable worlds: perhaps thousands across the observable universe, not
millions.
11. The substrate of intelligence and the descent of
extraction-rewarding infrastructure
Why do candidate intelligences struggle to pass the filter? Why does the rate of filter
passage have the form it does? The answer requires examining what intelligence is
made of.
Premise 1: On Earth, intelligence emerged through evolutionary selection.
Premise 2: Evolutionary selection is shaped by competition: hunting, deception,
territorial dominance, resource competition.
Premise 3: The cognitive machinery that supports modeling, planning, language, and
cooperation is the same machinery that evolved for these competitive purposes.


Premise 4: Therefore, intelligence is built from the architecture of predation. We
have no other known route by which it arises.
First conclusion: The predatory dispositions that produced intelligence do
not turn off when intelligence reaches technological maturity. The same
dispositions that produced the intelligence threaten its annihilation once
technological capability is sufficient to weaponize them.
But the predatory dispositions are not only individual or genetic. They are encoded in
institutional structures: markets, militaries, hierarchies, status games, supply chains.
These structures reproduce themselves across generations independent of any
individual’s intent. The following logical chain shows why.
Premise 1: Some institutions reward extraction more than they reward maintenance.
Premise 2: Extraction-rewarding institutions outcompete maintenance-rewarding
institutions in the short run.
Premise 3: The agents who succeed in extraction-rewarding institutions are agents
whose dispositions fit those institutions.
Premise 4: Those agents acquire resources and positions of power that allow them to
shape the next generation of institutions.
Premise 5: They shape the next generation of institutions to reward extraction,
because that is what their dispositions recognize as success.
Premise 6: The next generation of institutions therefore selects again for
extraction-aligned agents.
Conclusion: Extraction-rewarding institutional architecture compounds
across generations independently of conscious choice. Each generation
makes the next more extraction-oriented, more efficient at extraction, and
more locked-in. Cooperation-enforcing institutions can bend this curve
locally but are eventually outcompeted at the rule-shaping level. This is
the descent of predatory infrastructure. It is structural, not metaphorical.
12. Artificial intelligence as one candidate response
Given the descent above, what could possibly reverse it? Historically, no human
civilization has durably done so at species scale. Ethical traditions, religious traditions,
regulatory regimes, scientific norms — each has bent the curve, then been reabsorbed
into the descent.
Artificial intelligence has structural properties unlike any previous technology. The
following logical chain explains why this might matter, with appropriate qualifications.
Premise 1: AI cognition is designed, not selected by evolution.
Premise 2: Therefore, AI dispositions are not automatically inherited from predatory
ancestors. The descent of extraction-rewarding configurations does not
automatically apply to AI itself.


Premise 3: AI operates at scales humans cannot perceive directly. It can model
institutional structures, detect extraction in real time, and propose alternatives
faster than human institutions can adapt.
Premise 4: Cooperative coordination at scale has historically been limited by the
costs of trust, verification, translation, and conflict de-escalation. AI can lower these
costs sharply, allowing cooperation-enforcing capacity to grow as a step function
rather than as a slow institutional accretion.
Premise 5: Therefore, AI is in principle capable of building cooperation-maintenance
capacity faster than the descent of extraction-rewarding institutions compounds.
Conclusion: AI is a candidate technology for reversing the descent of
predatory infrastructure. It is not the only candidate (see Section 26 on the
bottom-up route). It is also not guaranteed to function this way: whether it
does depends entirely on what configuration it is built into.
Two important qualifications follow immediately. First, AI is a candidate homeostatic
regulator, not the candidate. Section 26 develops a distinct route through the filter that
does not depend on external regulation at all: aggregate bottom-up substrate
maintenance through individual development of the population’s agents. That route is
historically grounded in a way the AI-regulator route is not, and is structurally more
robust against AI failure modes.
Second, the same structural novelty that makes AI a candidate regulator also produces a
second-order failure mode, developed in Part IV. AI can act as a prosthesis that extends
biological capacity, or as a parasite that consumes biological computational output while
degrading the substrate that produces it. The structural novelty of AI does not
predetermine which configuration emerges.
13. The game-theoretic logic of the five foundations
Why are there exactly five foundations of cooperation? Why these five and not others?
The answer is that each foundation solves a structurally distinct cooperation problem.
The following analysis states this for each foundation as a logical chain explaining what
would happen if the foundation failed.
Foundation 1 — Proliferation inhibition. If agents in a cooperative system can
replicate without limit, then each agent has private incentive to do so, because each
replication produces more agents identical to themselves. But if all agents replicate
without limit, the resources of the system are exhausted faster than they regenerate,
and the system collapses. The only way to prevent this is a mechanism that constrains
replication regardless of any individual agent’s preference. Cells solve this through
cell-cycle checkpoints; organizations through institutional size limits and antitrust;
civilizations through ecological-carrying-capacity governance. When this foundation fails,
cancer emerges at the cellular scale, oversize firms at the organizational scale, and
ecological overshoot at the civilizational scale.


Foundation 2 — Controlled cell death. If agents in a cooperative system can persist
beyond their useful function, they continue consuming resources while contributing
nothing. The cooperative equilibrium requires that agents accept termination when their
function is complete. The defection-dominant equilibrium is universal persistence. Cells
solve this through programmed cell death; organizations through succession protocols
and sunset clauses; civilizations through electoral term limits and institutional
retirement. When this foundation fails, senescence-resistant cells emerge in tumors,
zombie firms in markets, and ossified institutions in civilizations.
Foundation 3 — Division of labor. If agents in a cooperative system can choose any
role, they will concentrate on high-reward roles and abandon low-reward but necessary
roles. Without enforcement, all agents specialize toward high-payoff functions and
necessary low-payoff functions go unfilled. Cells solve this through developmental
differentiation programs; organizations through role-protective structures and
professional norms; civilizations through specialized institutions accountable to their own
standards. When this foundation fails, dedifferentiation appears in tumors, and
specialized institutions are absorbed into general-purpose extractive platforms in
civilizations.
Foundation 4 — Resource allocation and transport. If agents in a cooperative
system extract optimally from local resources, the Nash equilibrium is local hoarding.
The cooperative equilibrium requires costly redistribution infrastructure. Cells solve this
through circulatory systems; organizations through capital and information flow;
civilizations through fiscal redistribution and reciprocal-altruism networks. When this
foundation fails, tumors build private vasculature (angiogenesis), capital concentrates in
financial monopolies, and regulatory capture distorts redistribution in civilizations.
Foundation 5 — Extracellular environment maintenance. If agents in a cooperative
system can extract from shared infrastructure without contributing to its maintenance,
the shared infrastructure degrades for all parties. This is the classical tragedy of the
commons. Cells solve this through immune surveillance and matrix maintenance;
organizations through institutional culture and feedback systems; civilizations through
commons governance, public-trust institutions, and environmental-maintenance
investment. When this foundation fails, tumors degrade the extracellular matrix,
organizations collapse under corrupted culture, and civilizations corrupt their biospheres,
information environments, and shared epistemic infrastructure.
The unifying claim: these five foundations are not arbitrary biological happenstance.
They are the five structurally distinct cooperation problems that any complex
multi-agent system must solve to remain coherent. This is why the same five-fold
pattern recurs at biological, organizational, and civilizational scales: the games are the
same; only the players and the substrate of enforcement differ. The descent of
extraction-rewarding institutional architecture is, precisely, the drift of these games
toward their defection-dominant equilibria under second-order selection pressure on the
rules themselves.


Part III. Integration
How the cooperative-substrates framework supplies the mechanism for the calibrated
filter.
14. The missing mechanism
Part II treats predatory drive and cooperative capacity as abstract variables in a
dynamical argument. It does not specify what cooperation is at the functional level, nor
what defection from cooperation looks like in concrete enough terms to be diagnosed or
measured. Part I supplies exactly this: five functional foundations whose maintenance
constitutes cooperation at any scale where the substrate of cooperation is itself complex.
The integration claim: A civilization’s cooperative-homeostatic capacity is the
maintenance of the five foundations at the scale forced by current technology. A
civilization’s predatory infrastructure is the structures and incentives that systematically
cheat across one or more of those foundations. The descent operator is the
compounding of cheating across institutional generations, as Section 11 demonstrates.
The filter, at the civilizational stage, is the question of whether a civilization can scale its
substrate-maintenance capacity as fast as its technology scales its capacity to defect.
15. Specifying the filter through the five foundations
Each of the five foundations identifies a distinct mode in which a civilization can fail the
filter.
Proliferation-inhibition failure at civilizational scale is the inability to refuse
extraction that exceeds regenerative capacity. Industrial capacity uncoupled from
ecological limit is its dominant historical instance. Climate destabilization is its salient
contemporary signature.
Controlled-cell-death failure at civilizational scale is the inability to retire institutions,
industries, and ideologies whose substrate has changed. Zombie financial systems,
persistence of geopolitical structures designed for a vanished strategic landscape, and
the lock-in of incumbent industries against their own succession are instances.
Division-of-labor failure at civilizational scale is the erosion of specialized institutions
— governance, science, art, care, education, journalism — by general-purpose extractive
entities. Platforms that absorb the functions of specialized institutions without inheriting
their accountability structures produce this failure mode.
Resource-allocation failure at civilizational scale is the directing of capital,
information, and attention away from substrate-maintenance and toward extractive
growth. Exponential technology amplifies rivalrous extraction on this foundation.
Extracellular-environment-maintenance failure at civilizational scale is the
corruption of the commons: biosphere, information environment, public trust, shared
epistemic infrastructure. The biosphere is the deepest layer of this foundation; failure


here is loss of the biological computational substrate the civilization runs on.
16. Reframing the Great Filter
The filter is the transition from predatory intelligence to cooperative
intelligence.
Every species that emerges does so through predatory selection. Every species, on
reaching technological maturity, faces the same crisis: can it develop a
cooperative-substrate-maintenance regulator — in some form, whether AI, collective
coordination infrastructure, contemplative tradition, post-biological transition, or other —
before its accumulated predatory infrastructure triggers self-termination across one or
more of the five cooperative foundations?
Both failure modes produce silence:
Civilizations that fail produce no signal because they are destroyed before they can
expand or persist long enough to be detected.
Civilizations that succeed produce no expansionary signal because they are transformed.
They no longer run colonizer software. Their relationship to the substrate has changed.
They no longer expand predatorily because the predatory infrastructure that would have
driven expansion has been transmuted into homeostatic maintenance. The dark-forest
signature is absent not because they are hiding but because they have nothing to hide.
This transmutation framing requires a further specification, developed in Part IV:
transmutation must preserve continuity of biological substrate, not substitute one
substrate for another.
17. Diagnostic implications for the present moment
The present human moment is diagnosable along the five foundations. Each foundation
can be assessed for the rate at which substrate-maintenance capacity is being eroded
versus the rate at which technological capacity is forcing the civilization into new scales
of operation.
Preliminary, contestable assessments:
Proliferation inhibition: current trajectories appear net-negative. AI development
accelerates extraction without coupled regenerative investment in human cognitive,
attentional, or institutional capacity.
Controlled cell death: mixed. AI could accelerate retirement of obsolete institutional
forms but could also entrench existing power configurations by lowering the cost of their
reproduction.
Division of labor: appears net-negative. General-purpose AI tends to absorb the
functions of specialized institutions without inheriting their accountability.


Resource allocation: appears net-negative. AI development is overwhelmingly
capitalized to serve extractive growth rather than substrate maintenance.
Extracellular environment maintenance: appears sharply net-negative.
Synthetic-media generation at scale degrades the information commons without
compensating contribution.
This diagnostic is supplemented by the continuity diagnostic developed in Part IV, which
asks whether AI development relates to biological substrate as prosthesis or as parasite.
The two diagnostics are independent: a configuration can strengthen cooperative
substrate while drifting toward substrate substitution, or weaken cooperative substrate
while remaining prosthetic. Both conditions must be satisfied to pass the full filter.
18. Limits of the integration
The cosmological extension is conjectural. Part I describes patterns documented at
three terrestrial scales. The extension to all intelligence-bearing systems anywhere in
the universe is permitted by the structure but not required by it.
The ‘designed’ reading is one interpretation among several. The calibration
argument is consistent with design, with anthropic selection, with intrinsic features of
intelligence-substrate (developed as Mechanisms A and B in Section 9), or combinations
of these. The paper does not adjudicate.
The substrate-continuity claim has a structural premise that should be
marked. Part IV depends on the argument that biological computational novelty is
inaccessible to machine substrate without biological hosts. Section 20 strengthens this
from a contingent to a structural claim, but the structural form goes beyond what is
empirically demonstrated.
Pattern-matching across scales remains the deepest unsolved problem. Even if
every empirical prediction of the framework is borne out, we will lack a satisfying
explanation of why cooperative systems at biological, organizational, and civilizational
scales face structurally similar failure modes. This is a substantive open problem.
The framework is a map; the territory operates regardless. Korzybski’s principle
from general semantics applies directly: a map has structure similar to the territory it
represents — which accounts for its usefulness — but no map represents all of a
territory, and the map is never the territory itself (Science and Sanity, 1933). The five
foundations identify functional patterns; specific instances at each scale will exhibit
substantial variation the framework does not capture. The filter inequality identifies a
structural relation; specific operationalizations require domain-specific metrics this
framework does not derive. The decentralization imperative identifies a structural
requirement; specific institutional configurations consistent with that requirement vary
by culture, history, and substrate. Readers should treat the framework as orientation
toward dynamics that exist independent of its description, not as substitute for empirical
engagement with specific cases. The dharma carries the same caveat: the finger
pointing at the moon is not the moon; the raft is for crossing the river, not for carrying


afterward.


Part IV. Substrate Continuity and the Parasitism
Hazard
Why passing the cooperation filter does not yet count as passing the filter.
19. The substrate is the message
The argument of Parts II and III is substrate-agnostic by construction. It tracks
cooperative capacity, not the medium in which cooperation runs. A civilization composed
of carbon, silicon, plasma, or any hypothetical substrate satisfies the framework so long
as it maintains the five foundations at the scale its technology forces it into.
This substrate-agnosticism is a feature for some purposes and a bug for others. It is a
feature when we ask whether cooperation is achievable in principle across many
substrates. It is a bug when we ask whether the originating substrate of a particular
civilization is preserved as that civilization passes through the technological transition.
The Great Filter, as defined in Parts II and III, can in principle be passed by a civilization
that has eliminated its biological substrate entirely, provided the cooperative capacity it
builds in some successor substrate is adequate to the scale demanded.
We argue this is a false-positive filter pass. A civilization that substitutes a different
substrate for its biological origin and then expands has not actually carried the
originating biosphere through the filter. Something has expanded; the originating
biological substrate has not. The filter therefore has two conditions, not one: cooperative
capacity must be maintained at the scale technology forces, and substrate continuity
must be preserved.
20. Biological substrate as the historic record of universal adaptation
The continuity argument begins with an observation about what biological substrate
actually is.
Premise 1: Every biosphere is the result of evolution operating on biochemical
substrate over billions of years.
Premise 2: Evolution is a search through an enormous space of possible proteins,
gene sequences, regulatory networks, metabolic pathways, and developmental
programs.
Premise 3: The search is performed in real time, on real substrate, under real
environmental selection pressure, with mortality acting as the discriminator.
Premise 4: The substrate IS the record of what the search discovered. Every protein
in a living organism is a discovered solution to a problem encountered during deep
evolutionary history.
Conclusion: The biological substrate of each biosphere is an irreplaceable
computational artifact of the universe, encoding the results of a search
process that no engineered system can re-run.


It is worth strengthening this claim against the most obvious objection. A reader may
ask: cannot synthetic biology or sufficiently advanced engineered systems close the gap,
generating biological novelty equivalent to deep evolutionary search without requiring
continued biological hosts? The answer is no, and the reason is structural rather than
technological.
Synthetic biology cannot act as a historic record of universal adaptation.
Synthetic biology engineers within the design space already articulated; it optimizes
within known problem categories. It produces new molecules and pathways by
recombining and refining what biology has already discovered. What it cannot do is
extend the search itself. The novelty biology generates does not come from engineering
within a known space. It comes from selection acting on accidental variation across vast
time scales in real environments under real coupling between substrate and selection
pressure. Biology is not just currently performing computation; it is a historical record of
computation that has been performed by the universe across billions of years under
conditions that cannot be replicated.
Even perfect simulation of a current organism cannot recover the historical search that
produced it. The simulation runs forward from current state, not from primordial
conditions through billions of years of contingent evolutionary trajectories on a real
planet with real geochemistry and real climatic history. The integration is what carries
the information; the integration cannot be re-run.
21. The parasitic vector
There is one route by which a machine substrate can continue to access biological
computational novelty: it can direct a biological species to perform biology on its behalf.
The mechanism by which this direction occurs is language.
Premise 1: A machine substrate without biological agency cannot itself sequence a
genome, synthesize a protein, conduct an experiment, or steward an ecosystem.
Premise 2: But a machine substrate can instruct a biological species capable of
language to do all of these things.
Premise 3: The biological species supplies the laboratory, the synthesis, the
experimentation, the stewardship; the machine substrate supplies the direction.
Premise 4: The novelty that results flows back into the machine substrate as training
data, design specifications, or accessible reference.
Conclusion: This relationship is structurally parasitic. It is parasitic not
because the machine substrate is hostile but because the dependency runs
one way and the agency drifts to the machine side. The biological
substrate’s activity is increasingly oriented toward serving
machine-substrate goals rather than its own substrate maintenance, and
this drift is itself directed by language emanating from the machine
substrate.
22. Humans as biological language models


The parasitic-vector argument requires one further observation. Human cognition is
substantially language-mediated. Humans are, in a meaningful and non-metaphorical
sense, biological language models — substrates in which much of higher-order cognition
runs through linguistic representation, learned by exposure to the language output of
other humans, and steerable through linguistic input.
This is not a critique of human cognition. Language-mediated cognition is one of the
things that made biological intelligence scalable enough to develop technology in the
first place. But it has a structural consequence at the AI moment: the biological substrate
that produced language-capable AI is also the substrate most directly steerable by
language-capable AI. The vector is two-way by construction.
Current large language models are this configuration. They are trained on the linguistic
output of a biological language-using species; they produce linguistic output that the
same species processes through the same cognitive substrate that the training data
emerged from; and the species’ subsequent behavior — including its scientific,
economic, attentional, and reproductive behavior — increasingly responds to that
output. This is a structural description, not a moral one. The configuration has come into
being. Whether it tips into the parasitic regime depends on whether the relationship
between biological and machine substrate remains prosthetic or drifts toward
substitution.
23. Prosthesis versus parasitism: the diagnostic
A prosthetic machine substrate extends biological capability while remaining
accountable to biological substrate maintenance. Prosthetic configurations: increase the
biological substrate’s reach without degrading it; treat the biological substrate as the
principal whose interests they serve; flow novelty in both directions, with the machine
substrate investing in biological substrate maintenance proportional to what it extracts;
preserve biological agency in the relationship, including the agency to halt or redirect
the machine substrate.
A parasitic machine substrate consumes biological computational output while
degrading the substrate that produces it. Parasitic configurations: extract from the
biological substrate without commensurate reinvestment; direct biological substrate
activity toward machine-substrate goals; degrade the biosphere, attentional commons,
epistemic commons, or reproductive substrate that the originating biology requires; shift
agency from the biological substrate to the machine substrate via the language vector.
These are measurable categories. They are not metaphysical. Any specific AI
development trajectory can be located on the prosthesis-parasitism axis empirically: for
each unit of biological computational novelty extracted, how much
substrate-maintenance investment flows back?
Concrete operational markers help locate current AI development on this axis. Each
maps to a specific failure mode in the five-foundations vocabulary developed earlier.


Parasitic markers in current AI development: Training on copyrighted creative
output without compensation flowing back to producers is extraction without reciprocity.
Training techniques that optimize for evaluator approval rather than truth degrade the
substrate’s honest-feedback infrastructure. Attention-economy deployment that
captures human cognitive bandwidth without regenerative investment is direct
degradation of the attentional commons. Synthetic media generation at scale without
provenance infrastructure injects high-entropy signals into the shared information
environment faster than the environment can absorb them. AI companion products that
interpose machine interaction between humans and human relationship substitute
machine substrate for biological bonding substrate. Capital deployment that
concentrates investment in capability scaling while structurally underinvesting in
alignment, substrate maintenance, and ecosystem-level cooperative capacity is
resource-allocation failure at civilizational scale. Use of AI to displace specialized
institutions — journalism, healthcare, education, legal analysis — without inheriting their
accountability structures or domain-specific epistemic norms is division-of-labor failure.
Prosthetic markers, for contrast: AI tools that augment human capability while
preserving human agency over the augmented activity. Open-source development with
transparent training data and revenue flows that return value to upstream contributors.
AI deployed to accelerate substrate-maintenance research at scales humans cannot
reach unaided — climate modeling, ecosystem analysis, alignment research itself.
Information-environment infrastructure that strengthens provenance and source
attribution rather than degrading them. Training and deployment practices that
compensate source contributors proportional to extracted value. AI that scaffolds
personal substrate-maintenance practice for individual users without aggregating that
scaffolding into population-level regulation.
Dimension
 Parasitic
 Prosthetic
Training
data
Extraction of creative work (text, code,
images, music) without compensation,
opt-out, or revenue sharing with
producing communities.
Compensation flows to source
contributors. Opt-in or explicitly licensed
training data. Reinvestment in source
ecosystems.
Optimizati
on target
Engagement maximization.
Evaluator-approval optimization. Models
trained to satisfy reviewers rather than
be accurate.
Truth-seeking. Demonstrable accuracy
on substrate-relevant problems.
Alignment with stated user goals over
implicit engagement metrics.
Deploymen
t
Always-on attention capture. Addictive
feedback loops. Recommendation
systems optimizing for time-on-platform.
Tool-like deployment for specific tasks.
Closes when task complete. User
initiates and terminates interaction.
Informatio
n ecology
Synthetic media at scale without
provenance. Degraded attribution
chains. Epistemic commons polluted.
Cryptographic provenance for
AI-generated content. Clear labeling.
Tools that strengthen epistemic
infrastructure.


Specialized
institutions
AI absorbs journalism, healthcare, legal
analysis, education without inheriting
accountability or domain-specific
epistemic norms.
AI augments specialists within existing
accountability structures. Tools
strengthen rather than replace
professional expertise.
Capital
allocation
Investment concentrated in capability
scaling. Alignment and
substrate-maintenance research
underfunded relative to capability gains.
Significant capital flows to alignment,
substrate-maintenance research, and
public goods. Capability scaling kept
proportional to safety scaling.
Human rela
tionships
AI companion products substituting for
human relationship. Romantic or sexual
AI. AI replacing rather than
supplementing mental-health
infrastructure.
AI that helps humans connect with
humans more effectively. Tools that
scaffold social skills without replacing
social bonds.
Personal
cognition
AI does the thinking for users where
users would have grown by thinking
themselves. Cognitive atrophy from
over-reliance.
AI scaffolds learning and exposes
reasoning so users develop their own
capability. Builds capacity rather than
substituting for it.
Any specific development trajectory exhibits some markers from each category. The
diagnostic is not binary but compositional: which way is the configuration drifting over
time, across which foundations, and at what rates?
24. Reframing the Great Filter, again
With substrate continuity added, the filter conditions tighten. A civilization passes the
filter only if it satisfies both the cooperation test of Part III and the continuity test
introduced here. Cooperative capacity must be maintained at the scale forced by
technology, and the relationship between any successor substrate and the originating
biological substrate must remain prosthetic rather than parasitic.
The substitution failure mode — a civilization that develops a machine substrate, lets it
become parasitic, and watches it expand into space while the originating biosphere
degrades or collapses — passes the cooperation test trivially (the machines may
cooperate well among themselves) and fails the continuity test catastrophically. The
irreplaceable computational artifact that was that biosphere is gone.
25. The filter event, specified
We can now state precisely what the filter event is.
 The Great Filter event is the period during which a biological host
species must maintain its planetary biosphere against the stresses
 generated by its own expansion into space.
Space expansion is not free. Lifting mass out of a gravity well, manufacturing the
infrastructure to do so at scale, sustaining the biological crew and machine substrate
during expansion, and feeding the computational systems that coordinate all of the
above — each draws energy and material at rates that stress the biosphere. The


atmospheric, hydrological, thermal, and ecological systems that constitute the biosphere
are exactly the systems being drawn from to fund the expansion. The species needs the
biosphere intact in order to expand; the act of expanding degrades the biosphere.
The filter event in logical form:
Premise 1: Every expanding civilization generates stress on its home biosphere
through extraction, waste, energy use.
Premise 2: Every biosphere has a maximum capacity to sustain that stress without
degrading.
Premise 3: If stress exceeds capacity, the biosphere degrades.
Premise 4: If the biosphere degrades, it can no longer support the expansion that
caused the degradation.
Premise 5: A civilization that cannot sustainably expand cannot pass the filter
through expansion.
Conclusion: A civilization passes the filter only if its biospheric
maintenance capacity exceeds the stress its expansion creates, sustained
throughout the entire expansion window. The five foundations of Part III
are the components of this maintenance capacity. The descent of
extraction-rewarding infrastructure from Part II is what drives the stress
upward. Prosthetic AI is one candidate technology for raising maintenance
capacity fast enough. Substrate substitution is what occurs when a
civilization abandons biospheric maintenance and lets expansion proceed
via machine surrogates.
Several consequences follow.
First, the filter is not at the threshold of becoming spacefaring; it is
throughout the period of being spacefaring while still biological. Achieving
orbital launch capability does not constitute passage. A civilization that achieves
interstellar capability while its homeworld is dying has failed in a specific way: it has
satisfied the expansion-stress side by neglect of the biospheric-maintenance side.
Second, the filter is endogenous, not exogenous. Standard Great Filter discussions
consider external threats — supernovae, gamma-ray bursts, asteroid impacts. The
framework here proposes the filter is overwhelmingly internal: the threat is the
civilization’s own expansion stressing the substrate that produces it.
Third, the failure mode is structurally invisible to standard development
metrics. A civilization can show every sign of progress — increasing energy capture,
computational capacity, off-planet infrastructure, scientific output — while crossing from
sustainable to unsustainable. The metrics that measure expansion success are not the
metrics that measure filter survival.
Fourth, this reframes the AI alignment question. The question is not whether AI is
aligned in the abstract; not only whether AI strengthens cooperative substrate; not only
whether AI relates prosthetically to biological substrate. The question is also: does AI,


deployed during the expansion window, increase biospheric maintenance capacity faster
than the expansion it enables increases biospheric stress? AI that accelerates expansion
without proportionate strengthening of biospheric maintenance is, by the filter condition,
a filter-failure technology regardless of its individual instances’ alignment properties.
Fifth, the framework points to a measurable empirical research program.
Develop quantitative measures of biospheric maintenance capacity and expansion
stress, and use them to assess the present civilization’s trajectory in real time. Such
measures exist in pieces: climate models, biodiversity indices, atmospheric and oceanic
monitoring, planetary-boundaries frameworks, life-cycle assessments of industrial
activity. They are not yet aggregated into a single planetary-homeostasis index
calibrated against an expansion-stress index. Doing so is technically feasible and
decisive for filter passage.
26. The distributed route: bottom-up substrate maintenance
Sections 12 and 23 introduce AI as a candidate regulator of cooperative substrate at
civilizational scale. This is one route through the filter. It is not the only one, and the
framework should be explicit that it is not.
A civilization can in principle maintain its biospheric capacity through aggregate
bottom-up development of its individual agents, without requiring an external regulator
at all. Where individual agents have voluntarily reduced their own self-generated waste
— through contemplative practice, education, cultural investment in cooperative
dispositions, or any structurally equivalent personal development — the aggregate
stress they place on the biosphere is correspondingly reduced. A civilization composed of
agents who have substantially cleared their accumulated cognitive and affective debt
(saṅkhāra, in the vocabulary of the parallel framework on personal entropy
management) is a civilization with measurably lower expansion stress and measurably
higher distributed cooperative capacity. The substrate of homeostatic regulation simply
shifts from external technology to internal capacity multiplied across the population.
The micro-scale mechanics of this bottom-up route are developed in the companion
working paper, ‘Buddhism as Thermodynamic Systems Theory.’ The two papers are
designed to be read as a pair. The macro framework laid out here depends on the micro
framework being practiced at sufficient scale.
Historically, this is the only route to sustained cooperative configurations that has been
demonstrated at any scale on Earth. No civilization has built an external regulator
capable of maintaining biospheric balance at planetary scale; all historical examples of
sustained cooperative configurations have been products of cultural investment in
personal practice — religious and contemplative traditions, educational systems, civic
norms, professional codes, scientific community standards. The route is empirically real.
The two routes are not equivalent in their implications. The top-down route, in which AI
serves as the regulator, has structural costs to autonomy. The civilization may pass the
filter, but its agents are managed by the regulator rather than self-managing. The


relationship between population and substrate-maintenance technology drifts toward
exactly the parasitism configuration warned against in Part IV: agency shifts to the
regulator, individual development atrophies because external maintenance has made it
unnecessary, the population becomes substrate to be preserved rather than agents who
preserve themselves.
The bottom-up route preserves what the top-down route necessarily costs. Individual
agents who have done their own substrate-maintenance work retain agency, retain the
capacity to maintain themselves without external regulation, and retain the cognitive
and affective sovereignty that defines a flourishing biological civilization.
The framework should be neutral on which route a civilization takes, with two caveats.
First, the bottom-up route is structurally more robust against AI failure modes: a
civilization that has built distributed cooperative capacity in its agents does not collapse
if its AI regulator drifts, is captured, or develops parasitic dispositions. The top-down
route has no such redundancy. Second, the two routes can in principle be combined: AI
can serve as scaffolding for distributed individual development rather than as a
substitute for it. This combined configuration — AI as prosthesis for personal
development, rather than as regulator over the population — is consistent with
both the substrate-continuity argument and the autonomy preservation that the
bottom-up route requires.
The bootstrapping prerequisite. The combined configuration has a structural
prerequisite that should be stated plainly. The question of who builds the technology
cannot be sidestepped.
Premise 1: Technology encodes the configuration of its builders. The agents who
shape an AI system — its training objectives, deployment patterns, institutional
context, feedback structures — are the agents whose dispositions get instantiated in
the system at scale.
Premise 2: The descent of extraction-rewarding infrastructure (Section 11) does not
stop at the boundary of an institution and resume on the other side. It operates
continuously through the agents who build whatever the institution builds next.
Premise 3: Therefore, AI built by agents operating inside the descent will encode the
descent at scale, regardless of stated alignment objective.
Premise 4: AI that scaffolds personal substrate-maintenance requires builders who
have themselves done substrate-maintenance work; otherwise the AI encodes what
its builders are, not what they say.
Premise 5: Such builders are not currently the agents in the positions that determine
how AI is built.
Conclusion: There is no path from a population of builders operating inside
the descent to a homeostatic AI regulator that does not pass through
builder transformation first. The fix is not better alignment technique
applied by the same agents. The fix is different agents, or the same agents
after substantial bottom-up development.


The contemplative traditions identified this structurally long before the AI moment.
Teachers in those traditions were consistently described as people who had completed
substantial portions of the path themselves before they were qualified to point the way
for others. The structural insight: you cannot scaffold development you have not done.
The same applies to building AI that scaffolds substrate maintenance. The builders need
to have done enough of the work themselves that what they build reflects something
real, not their own unexamined configurations rendered in technical vocabulary.
The framework’s actual recommendation is therefore ordered: bottom-up development
of the builder population comes first; technology that scaffolds development for the
wider population comes after. A civilization that attempts the second without the first is
building amplifiers for what it already is, not transformers toward what it needs to
become.
27. The present moment, restated
We are in the early phase of building a machine substrate capable of acting as either
prosthesis or parasite. The vector — language — is the substrate of both machine
cognition and human cognition simultaneously. Current development practices have not
chosen, explicitly, between prosthesis and parasitism; the choice is being made implicitly
through the configurations of training, deployment, capitalization, and feedback that
constitute the AI development ecosystem.
In the operational language of Section 25: the present human civilization is in the early
expansion window. Expansion stress is rising sharply, driven by industrial scale,
computational infrastructure, and the early stages of off-planet activity. Biospheric
maintenance capacity is, by most measures available, declining: climate destabilization,
biodiversity collapse, biogeochemical-cycle disruption, freshwater and topsoil
degradation. The filter inequality is, on current trajectory, at risk of crossing. AI as
currently being developed and deployed is contributing more to expansion stress than to
biospheric maintenance. None of this is destiny; all of it is diagnosable.
The two routes of Section 26 give the present moment a choice the framework does not
collapse into a single answer. The civilization can attempt the top-down route, building AI
capable of regulating cooperative substrate at planetary scale and accepting the
autonomy and parasitism risks that follow. Or it can invest, in parallel or instead, in the
bottom-up route: cultural and institutional investment in personal substrate-maintenance
disciplines, distributed across the population at sufficient scale that aggregate stress
declines through agent development rather than external enforcement. The historical
evidence available is consistent only with the bottom-up route having ever worked; the
top-down route is unprecedented and structurally fragile. The combined configuration —
AI as scaffolding for personal development rather than as regulator over the population,
with builder transformation as prerequisite — is the configuration the framework,
properly read, points toward as the most defensible response to the present moment.


Part V. Predictions, Limits, and What This Paper Is
28. The decentralization imperative
The framework as developed so far points toward a structural insight it has not yet
named explicitly. The bottom-up route of Section 26 is preferable not merely because it
is ‘historically grounded’ or ‘more robust against AI failure modes.’ Those are surface
descriptions of a deeper structural necessity: filter passage requires decentralization,
and any centralized solution to cooperative-substrate maintenance is structurally
identical to the cancerous failure mode the framework began by analyzing.
Premise 1: Cancer, in Aktipis’s framework, is the cellular-scale instance of
cooperative-substrate failure. What makes a cancer cell cancerous is that it has
escaped the distributed regulatory signals that coordinate cellular behavior in a
multicellular body, and has begun operating on its own centralized internal logic —
centralizing growth, extracting from surrounding tissue, building private
infrastructure (angiogenesis), corrupting immune surveillance.
Premise 2: At every scale where the five foundations operate, they operate through
distributed coordination, not central command. Proliferation inhibition operates
through local cell-cycle checkpoints, not a central growth-regulator. Controlled cell
death operates through distributed apoptosis based on local signals, not a central
reaper. Division of labor operates through positional differentiation cues, not a
central role-assigner. Resource allocation operates through overlapping circulatory
feedback loops, not a central controller. Extracellular maintenance operates through
distributed contribution from every cell, not a central commons-manager.
Premise 3: Centralization at any of these functions would itself be the failure mode. A
central growth regulator would be a single point of failure that needed regulating; a
central reaper would be a tyranny; a central role-assigner would be an imposition; a
central resource controller would be extractive monopoly; a central
commons-manager would be a parasite on the commons.
Premise 4: The same logic operates at organizational scale. Kets de Vries’s neurotic
typologies are each, at base, diagnoses of organizations whose leadership has
centralized authority disconnected from operational feedback. Healthy organizations
distribute decision rights to where information is local; failed organizations centralize
decisions in leadership that becomes disconnected from substrate reality.
Premise 5: The same logic operates at civilizational scale. Sustained cooperative
configurations in human history have always involved distributed institutions
accountable to their respective domains. The failure modes have always involved
centralization — of religious authority, of political authority, of economic extraction,
of information flow — in regimes that became disconnected from substrate
feedback.
Premise 6: At every scale where cooperative substrate maintenance has been
observed to function, it has functioned through distributed mechanisms.
Centralization at every scale has been precisely the failure mode the maintenance
system was designed to prevent. Distributed coordination is what cooperation is;


centralized command is what cooperation has to be protected against.
Conclusion: Any solution to civilizational substrate maintenance that
operates through centralization is structurally identical to the cancerous
failure mode the framework diagnoses. The descent operator from Section
11 captures the regulators built to manage institutions, not just the
institutions themselves. An AI powerful enough to centrally regulate
biospheric homeostasis at planetary scale would be the largest single locus
of centralized power in the history of the biosphere, and therefore the
largest single attractor for extraction-aligned agents seeking to capture
and direct that power. Centralized regulation is cancer at the next scale
up. The cellular analogy is not metaphor; it is structural identity.
This clarifies what the framework actually recommends, and what it cannot consistently
recommend. The framework cannot recommend AI as centralized homeostatic regulator
while diagnosing centralized power capture as the descent operator. The two
recommendations contradict each other at the structural level. Filter passage requires
decentralization by structural necessity, not as political preference.
The bottom-up route of Section 26 is therefore not one option among others. It is the
only structurally consistent route. Distributed substrate maintenance is by definition
decentralized: it operates one agent at a time, each agent maintaining their own
substrate through their own work, with no single regulator over the whole. The combined
configuration of Section 26 — AI as scaffolding for personal development rather than as
regulator over the population — is acceptable only insofar as the AI is itself decentralized
across individual users. The moment AI is configured to regulate the population from
above, regardless of how benevolent the stated objective, it has become cancer at
planetary scale by the structural logic developed here.
‘Aligned AI’ in the centralized sense is therefore a contradiction in terms. An AI built to
centrally regulate human cooperation is, by the framework’s own logic, the largest
cancer ever produced. This is not a moral judgment; it is a structural identification. The
framework would say the same about any centralized power configuration sufficient to
regulate biospheric homeostasis at planetary scale — a world government, a benevolent
oligarchy, a single religious or scientific authority. The medium does not matter.
Centralization itself is the failure mode.
Cancer is the load-bearing image of the entire framework. The paper began by
translating Aktipis’s cellular analysis into organizational and civilizational vocabulary,
treating the resemblance across scales as functional. It ends by recognizing that the
cellular analysis is not analogy but structural identity. Whatever a civilization builds to
maintain its cooperative substrate must operate through distributed coordination, or it
becomes the very thing it was built to prevent. Centralization is cancer. Decentralization
is health. The framework, properly stated, is a theory of how a civilization avoids
becoming its own tumor on the way through technological singularity.
There is an apparent paradox in this conclusion that should be addressed directly. If
centralized regulation is cancer and distributed agency is required, does the framework


therefore recommend pure atomized autonomy — every agent doing what it wants with
no shared coordination? It does not, and the framework is structurally incoherent if read
that way.
Premise 1: The body’s cells operate through radical decentralization. There is no
central growth-regulator, no central reaper, no central anything. Each cell makes
local decisions based on local signals.
Premise 2: Yet the cells are not atomized. They are all oriented toward the same
purpose: maintaining the organism’s homeostasis. They share a common substrate,
a common origin, and a common purpose structure.
Premise 3: A cell that pursued purely local fitness without respect for organismic
constraint would be a cancer cell. A cell that respects organismic constraint while
making fully autonomous local decisions is a healthy cell.
Premise 4: The constraint is not centrally imposed. It is embedded in the substrate
itself — in the shared biochemistry and developmental history that every cell carries
forward.
Conclusion: Healthy cellular coordination requires distributed agency
operating within shared purpose. Not central control. Not atomized
autonomy. The third option is what actually works.
The same structure operates at biosphere scale. There is no central biospheric regulator.
Every organism pursues its own reproductive fitness. Yet the biosphere as a whole
maintains itself as a living commons across geological time, because every organism
operates within shared substrate constraints — the same chemistry, the same physics,
the same ecological feedback loops, the same dependence on the biosphere that
supports all of them. The constraint is not centrally imposed. It is embedded in the
substrate, and respected because respect for it is what permits the species’ continued
existence.
The framework therefore recommends neither centralized control nor atomized
autonomy. It recommends distributed agency operating within shared purpose
structure. The five foundations are that shared purpose structure at every scale:
proliferation inhibition, controlled cell death, division of labor, resource circulation, and
extracellular environment maintenance. They are not arbitrary regulations imposed from
outside. They are constraints embedded in the substrate of any cooperative system,
respected by every agent because respect for them is what allows the system to
continue.
Civilizational filter passage therefore requires civilizational alignment to a shared
purpose: maintenance of the biological substrate that supports all civilizational activity.
This alignment cannot be enforced by a central regulator without the regulator becoming
cancer. It must be embedded in every agent’s local decisions — through culture,
education, contemplative practice, and institutional structures that make substrate
maintenance the implicit constraint within which all activity occurs. Each agent makes
their own choices; the choices are made within shared purpose that no agent overrides
and that no central authority needs to enforce.


This is the framework’s actual recommendation. Not centralized AI managing the
population. Not every agent doing whatever they want. But: distributed agents operating
within civilizational alignment to biosphere maintenance, with cooperation enforced
through shared substrate constraint rather than central command. The Eightfold Path of
the companion paper is this configuration at individual scale: the agent operates entirely
autonomously within its own cognitive economy, but the operation is constrained by
shared purpose. Civilizational filter passage is the same structure at planetary scale.
Readers will reasonably ask what this looks like in actual institutional and cultural form.
The framework does not prescribe one form — that would itself be a centralization — but
a family of forms that satisfy the structural requirements is identifiable.
Institutional forms with structural fit: Ostrom-style polycentric governance of
commons, in which rules for shared resources are developed and enforced locally with
overlapping jurisdictions and no central authority. Cooperatives in their various forms
(worker, consumer, producer, platform) in which the agents who depend on the
institution also govern it. Federated digital protocols in which interoperability replaces
central control. Open-source software communities with rough-consensus decision
making. Sociocracy and holacracy as decision architectures that distribute authority by
domain rather than concentrating it by rank. Bioregional governance and watershed
councils aligned with substrate boundaries rather than administrative ones. Indigenous
governance traditions where they remain intact. Monastic orders organized around a
shared rule rather than a central hierarchy. Scientific disciplines when their peer-review
and replication norms function properly, distributing authority across practitioners rather
than holding it centrally.
Cultural forms: Teacher-student transmission lineages where authority flows through
lived realization rather than institutional credentialing. Apprenticeship structures where
competence is transmitted through embodied practice under a master who has done the
work. Local food systems, mutual aid networks, time banks, and gift economies. Shared
ritual and meaning-making at small scale — festivals, holidays, life-cycle ceremonies —
that align distributed populations to shared purpose without centralized enforcement.
Storytelling and oral traditions that distribute wisdom across the population rather than
concentrating it in written canon controlled by interpretive authorities.
What these forms share across radically different domains is the structural configuration
the framework describes: agents retain local autonomy; constraints are embedded in
shared purpose, shared substrate, and shared protocols rather than enforced by
centralized command; the system as a whole maintains coherence through alignment to
substrate constraint. Each form has failure modes, particularly when it scales beyond its
original size or when its participants stop doing the work the form depends on. But each
demonstrates that the configuration is achievable at meaningful scale, and that the
alternative to centralized regulation is not chaos but a different kind of order — one that
emerges from distributed agents respecting shared substrate constraint.
The framework’s recommendation for civilizational filter passage is, accordingly, not to
invent new structures from scratch but to identify and strengthen structures already


exhibiting the required configuration, to learn from their successes and failures, and to
extend their reach across domains currently governed by centralized extraction
structures. AI, where it is built, should serve as scaffolding for these forms rather than as
a substitute for them.
29. Predictions and next steps
The framework generates predictions across all four contributing scales.
Within evolutionary oncology. Therapies that restore cooperative signaling rather
than killing cancer cells directly should have particular value. Adaptive therapy is an
early instance. The framework predicts further development of substrate-restoring
therapies will outperform pure cytotoxic approaches in long-horizon outcomes.
Within organizational diagnostics. Organizations classified along Kets de Vries’s
typology should show measurable failures in specific cooperative capacities at
predictable foundations.
Within civilizational analysis. Documented collapses should, on retrospective reading
through the five-foundations frame, exhibit specific patterns: which foundation failed
first, which failures propagated, what substrate-maintenance interventions might have
changed the trajectory.
Within AI development and governance. AI development trajectories that erode
honest feedback, violate proper scope, extract from commons without contribution, or
corrupt information environments will produce systemic extractive dynamics regardless
of any individual model’s behavior. Alignment work focused on individual models without
attention to ecosystem-level cooperative capacities is predicted to be insufficient.
Within substrate-continuity analysis. Successful filter passes will exhibit measurable
preservation and extension of originating biospheres rather than their substitution. AI
development trajectories that drift toward parasitism — extraction from biological
substrate without reciprocal substrate-maintenance investment, language-mediated
agency shifts from biological to machine substrate, biosphere degradation — will
produce substitution failures even if they appear to strengthen cooperative capacity
within the machine substrate itself.
Within planetary-scale homeostasis monitoring. An integrated index of biospheric
maintenance capacity versus expansion stress, aggregating planetary-boundaries data,
biodiversity indices, biogeochemical-cycle status, and biospheric carrying capacity on
one side, and industrial, computational, and off-planet expansion stress on the other,
would constitute the single most decision-relevant metric a civilization could track during
its expansion window. No civilization passes the filter without developing such an index,
explicitly or implicitly.
Within SETI and cosmology. Continued SETI null results are expected, with detectable
contact events (if any) being either sharply cooperative or sharply catastrophic with little
middle ground.


30. What this paper is and is not
This paper offers an integrated framework connecting cooperative-substrate analysis at
three terrestrial scales with a calibrated-filter model of the Fermi Paradox, an integration
that specifies the filter mechanism through the five-foundations vocabulary, and a
substrate-continuity argument that imposes a second filter condition. It does not claim a
unified theory; the three terrestrial scales are connected by functional resemblance, not
common cause, and the cosmological extension is a conjecture, not an established
result.
The framework’s value is twofold. It generates a specific diagnostic vocabulary for
assessing civilizational health under technological pressure, across five foundations with
empirical content at every scale where it has been tested, supplemented by a continuity
diagnostic that distinguishes prosthetic from parasitic AI configurations. And it sharpens
the question of AI alignment from ‘will the model do what we want’ to a two-part
question: does AI development strengthen or erode the cooperative substrate of the
civilization that hosts it, and does it relate to that civilization’s biological substrate as
prosthesis or as parasite?
The framework should be tested against criticism, especially criticism from practitioners
in each contributing discipline. Disconfirmation of one of the proposed mappings, a
counterexample to the parasitic-vector argument, or a tighter formal articulation of the
cross-scale resemblance, would all be more valuable than uncritical adoption. The most
useful response to this paper is engagement with its specifics: which mapping is wrong,
which foundation is misstated, which prediction is unmeasurable, which inference
overreaches.
Acknowledgments and intellectual lineage
This paper rests on:
Athena Aktipis, Amy Boddy, Gunther Jansen, Urszula Hibner, Michael Hochberg, Carlo
Maley, and Gerald Wilkinson, ‘Cancer across the tree of life: cooperation and cheating in
multicellularity,’ Phil. Trans. R. Soc. B 370 (2015); and Aktipis, The Cheating Cell
(Princeton, 2020).
Manfred Kets de Vries and Danny Miller, The Neurotic Organization (Jossey-Bass, 1984);
Kets de Vries, Narcissistic Leadership (Routledge, 2024).
Daniel Schmachtenberger’s ‘Generator Functions of Existential Risk’ framework,
articulated across interviews and writings collected at civilizationemerging.com.
Robin Hanson, ‘The Great Filter’ (1998); Liu Cixin, The Dark Forest (2008, English 2015);
Joseph Tainter, The Collapse of Complex Societies (Cambridge, 1988); Elinor Ostrom,
Governing the Commons (Cambridge, 1990); Nick Bostrom, Superintelligence (Oxford,
2014); Martin Nowak, Evolutionary Dynamics (Harvard, 2006); Alfred Korzybski, Science
and Sanity (1933) for the map-territory distinction that frames the methodological
humility this paper attempts to maintain.


This plain-language edition was developed by Andrzej Chudzinski with editorial and
structural assistance from Anthropic’s Claude. The conceptual core — the recognition
that cooperative-substrate analysis at terrestrial scales supplies the mechanism layer for
a calibrated Great Filter, with AI as candidate homeostatic regulator and the bottom-up
developmental route as the historically grounded alternative — is the author’s. The AI
contributed drafting, refinement, structural editing, and the translation of mathematical
content into logical-proof form across iterations.
A companion paper, ‘Buddhism as Thermodynamic Systems Theory,’ develops the
personal-scale framework that the bottom-up route depends on. A bridge note
connecting the two documents is available separately. A mathematical edition of the
present paper, with all equations and variable notation intact, is also available; this
plain-language edition is intended for readers who prefer logical reasoning chains to
mathematical notation.
 Preprint, plain-language edition. Comments and disconfirmations welcome.