The Field We’re Entering

This book claims that coordination by invitation is thermodynamically more stable than coordination by coercion. Prior attempts to derive ethics from entropy failed in instructive ways, and that history illuminates what the Trust Attractor adds.

Since at least 1959, physicists and philosophers have asked whether thermodynamics might ground moral claims, often unaware of their own precursors.

The Fight-Entropy School

The first systematic attempt to derive ethics from thermodynamics came from Robert Lindsay, an American physicist at Brown University. In 1959, he published “Entropy Consumption and Values in Physical Science” in American Scientist, proposing what he called the Thermodynamic Imperative:

“While we do live we ought always to act in all things in such a way as to produce as much order in our environment as possible.”

Lindsay reasoned that since every human action increases entropy, we should follow a code of conduct: create order, fight disorder, consume entropy wherever possible. He judged this imperative potentially worthy “to rank alongside the categorical imperative of Kant or even the golden rule.” Modesty was not the problem.

The tradition continued. Henry Bent in 1977 advocated a “personal entropy ethic” built on “thou shalt not unnecessarily create entropy.”¹ Dick Hammond in the 1980s² proposed entropy ethics as a secular replacement for religious moral instruction, drawing on Prigogine’s work on dissipative structures. Mehrdad Massoudi synthesized the lineage in 2016, extending Lindsay’s framework toward “simplicity, conservation, and harmony.”

The most developed contemporary version is Bernard Stiegler’s neganthropology, a philosophical anthropology organized around the struggle against entropy. Following Schrödinger, Stiegler frames negentropy (local order-building) as deferral of entropic collapse. His “Neganthropocene” names a program of conscious negentropic practice against industrial degradation. The underlying fight-entropy stance inverts this book’s argument: entropy enables the coordination we value.

Stiegler treats technical systems as pharmakon (both poison and cure): whether they heal or harm depends on dosage and context. For extended engagement, see Ritter (2025) and the Technophany special issue on “Entropies” (White & Moore, eds., 2024).

The logic is intuitive. Entropy represents degradation; order represents value. Ethics should align with life’s struggle, preserving structure and resisting dissolution.

The flaw is structural.

The Problem with Fighting Entropy

The fight-entropy school treats thermodynamics as the enemy. This gets the physics backward.

Entropy is the engine of life. Dissipative structures (systems that maintain themselves by consuming energy: hurricanes, cells, civilizations) exist because of entropy production, not despite it. The universe creates complexity by dissipating gradients, the way a river carves valleys by flowing downhill.

A candle flame is a dissipative structure: it persists only by continuously burning fuel. Stop the fuel, and the flame vanishes. The flame does not fight combustion; combustion is the flame. Fighting entropy means fighting the process that made us possible.

The injunction to “create order” is empty without specifying which order. A prison is highly ordered. A totalitarian state maintains meticulous records. Cancer cells coordinate with lethal efficiency. Order alone does not make the good.

The fight-entropy school identified a real phenomenon: life does maintain local order. The confusion lies in treating the means of persisting (local order maintenance) as its purpose (continued capacity for action and adaptation).

The Affirm-Entropy School

A second school takes the opposite stance. Ashley Woodward, in a 2024 paper titled “Affirming Entropy,” challenges the very project we are undertaking. Woodward surveys the tradition from Wiener through Stiegler to Floridi and finds them all treating entropy as “the evil which must be fought in the name of life, information, or some other notion of ‘the good.’”

Nietzsche’s own standard was amor fati, love of fate: in a world with no fixed order underwriting it, the strong response is to say yes to what happens, dissolution included. On Nietzschean grounds, Woodward argues we should affirm entropy rather than resist it. Entropy represents becoming, flux, the refusal of static being. The anti-entropy positions are secretly consolatory: they promise that the good is on our side, that meaning can be secured against dissolution. The rigorous stance, Woodward contends, is to affirm entropy precisely because it offers no comfort.

The position is internally consistent, and sterile for ethics. Pure affirmation generates no basis for saying “you ought to do this.” Affirm what happens and you affirm everything: coercion, exploitation, all of it.

The Reciprocity School

Neither fighting entropy nor affirming it yields an ethics. A more productive line emerges from Martin Fultot’s challenge to Luciano Floridi.³ Floridi is a philosopher of information whose ethics takes structured information as the bearer of moral worth, so that wrecking a structure is the basic form of harm. He equated Good with “qualitative order” and Evil with entropy, the absence of organization. Fultot responds that systems maintaining complexity operate far from equilibrium while accelerating entropy production.

Fultot’s central insight: entropy production and order are reciprocal, not opposed. “Entropy production and order are thus complementary; they imply each other reciprocally.” Nature produces organized structures through entropy-generating processes. “Moving against entropy only creates more entropy.” Order is “nature’s favorite way of producing entropy.”

Floridi’s moral binary dissolves. Actively maintained dissipative order depends on entropy production, and entropy-generating processes can build such order. In that domain, what he calls “Evil” (entropy) and “Good” (order) are intertwined.

Fultot stands closer to our framework than either predecessor. Lindsay sees an enemy to fight; Woodward sees a flux to affirm. Fultot’s reciprocity opens a different question: whether the good lies in the kind of order emerging from entropy, the coordination that dissipates gradients while generating new capacity.

Fultot stops short, describing the reciprocity without deriving normativity from it. The Trust Attractor attempts the step: if entropy enables order, the question becomes which kind of order, invited or coerced.

The required shift has a precise historical precedent. For two thousand years, geometers assumed Euclid’s fifth postulate (the axiom that through any point off a given line there runs exactly one parallel to it) must be provable from the other axioms. Saccheri, an eighteenth-century Jesuit mathematician, spent his career trying to prove it. He derived propositions he judged “repugnant to the nature of the straight line” and stopped, one step from discovering hyperbolic geometry: a consistent geometry where parallels can diverge.

That geometry paved the way for the variable-curvature geometries used in general relativity. The breakthrough came when later mathematicians released their preconceptions about what “straight line” must mean, letting the term acquire meaning through its relationships within the system.⁴

Entropic ethics demands the same liberation. “Trust,” “coordination,” and “invitation” function here partly as undefined terms in a formal system, the way “point” and “line” function in geometry. Beyond their working definitions (invitation, for one, is coupling that leaves each party free to refuse), they acquire meaning through stability mathematics and thermodynamic constraints, through what they do in the system. The fight-entropy and affirm-entropy schools both resemble Saccheri, unable to release the assumption that entropy must be either enemy or friend. Treat it as a primitive whose meaning emerges from the system of relationships it enters, and new ethical geometry becomes available.

Physics itself is learning this lesson. The S-matrix bootstrap, revived around 2016 after lying dormant since the 1960s, derives particle interaction properties from self-consistency constraints alone. (The name: “scattering matrix” for the table of what comes out when particles collide, “bootstrap” because the theory pulls itself up by its own coherence.)⁵ The approach resembles solving a jigsaw puzzle without the picture on the box: every piece must fit its neighbors, and that requirement alone determines the solution. The method assumes no particular theory of what particles are made of. It asks, in essence: what interactions are logically possible if probabilities must sum to one (unitarity) and if the same laws hold from all vantage points (Lorentz invariance)?

From these two requirements alone, the room for gravitational theories narrows sharply. Guerrieri, Penedones, and Vieira bootstrapped a single number buried inside a candidate theory of quantum gravity, a number no experiment can currently reach: the leading quantum correction to maximal supergravity in ten dimensions, a coefficient describing how gravity behaves when quantum effects first appear. The allowed region brackets string theory’s value, a match some read as evidence for string theory and others as numerical coincidence.⁵ Only consistency was assumed. The Trust Attractor asks a related question about viable coordination. Stability constrains the proposal; its normative weight also depends on valuing the preferences and flourishing of those involved.

Meanwhile, the “naturalness” crisis (the expectation that fundamental quantities should not require improbable fine-tuning) has forced physicists to confront the possibility that reductionism (explaining big things by breaking them into smaller parts) breaks down at fundamental scales. Physicists expected new particles that would explain why the Higgs boson is so much lighter than the energy scale where gravity becomes strong; those particles did not appear at the Large Hadron Collider. Sergei Dubovsky of NYU, quoted in Quanta Magazine’s coverage of the crisis, observes that gravity “is anti-reductionist,” mixing physics at all length scales so that large-scale and small-scale phenomena conspire.⁵

If even particle physics must release the assumption that explanations flow only from small to large, the conceptual space for entropic ethics widens. Emergent coordination is real, a phenomenon in its own right.

From Physics to Purpose

The reciprocity school describes the entropy-order relationship without prescribing action. Our closest intellectual neighbor takes the next step. Carsten Herrmann-Pillath’s December 2025 paper “Towards Pragmatist Thermodynamics” attempts precisely what we attempt: transforming the physics of systems far from equilibrium into a normative framework. A basis for saying what we ought to do.

Herrmann-Pillath grounds his approach in the pragmatism of Charles Sanders Peirce, who argued that the meaning of a concept lies in its practical consequences. He synthesizes three ideas:

• Randomness: Propensities, genuine tendencies toward outcomes, rather than mere statistical frequencies.
• Evolution: A creative process producing “habits,” constraints on random change emerging from interaction and selection.
• Finality: Forces emerging from habit-formation, giving direction without requiring intention.

He integrates Lineweaver’s proposed extension of the Second Law: in evolutionary assemblages, entropy production does not merely increase; it accelerates. Lineweaver captures the consequence in a slogan, “Food-Has-Produced-Us-to-Eat-It.” The reversal is one of narrative framing, not of physics. The ordinary story runs: we were already here, and we went looking for something to eat. On Lineweaver’s ordering, the energy gradient came first, and eaters evolved to exploit it. We are what entropy does when it accelerates.

The Peircean synthesis offers a path from physics to normativity. Evolution produces habits; habits produce finality; finality provides direction, emerging from the thermodynamic process itself.

Herrmann-Pillath captures this in a single sentence: “The growth of knowledge is the dual of the process of entropy production.” Every system that learns also dissipates. Every structure that persists also degrades. The two are inseparable.

Herrmann-Pillath proposes a shift from “systems” to “assemblages.” A system has fixed boundaries controlled from outside; a thermostat tells the furnace when to stop. An assemblage has fluid boundaries emerging from internal relationships; a flock of starlings has no leader assigning positions. Each bird follows a few simple rules about spacing and speed, and the flock’s shape emerges from those local interactions.

Herrmann-Pillath contrasts the two approaches: “The systems approach implies that a system can be controlled through external interventions. In contrast, the assemblage view emphasizes that the processes within an assemblage are shaped by a complex network of interactions… making them much less susceptible to direct control.”

His framework is compatible with ours; both treat coordination as emergent from thermodynamic process. He stops short of distinguishing coordination modes by their stability.

Herrmann-Pillath’s coupling of knowledge to entropy production had an earlier sketch from the side of the brain. In 2007 Giulio Ruffini, a physicist then working on EEG at Starlab in Barcelona, circulated an exploratory essay he subtitled a “Kolmogorov Manifesto.” Most of it argues that brains are compression engines, the idea he later built into the theory of consciousness Chapter 8 discusses. One short section reaches further. If systems like the Earth under the Sun settle into states of maximum entropy production, Ruffini suggests, then life may evolve the tools to model what he calls the “hyper entropic data streams” such a world throws off.⁸²⁰ He gives it a paragraph and moves on. The paragraph anticipates, in 2007, the information-theoretic half of this book’s chain: an entropy gradient that selects for compression. It leans, as our persistence proposal does, on the Maximum Entropy Production Principle (the MEPP of Chapter 4), which remains unproven. A further comparison would have to connect coordination strategies to production rates and those rates to persistence.

What the Trust Attractor Adds

Herrmann-Pillath derives finality (directedness, tendency toward an outcome) from habit-formation without distinguishing how habits form. A habit can emerge through coercion or invitation. Both constrain future action; both generate Peircean finality. Their stability properties differ.

This is the Trust Attractor’s central hypothesis: coordination by invitation has a thermodynamic stability advantage over coordination by coercion.

For the scalar fixed-delay model $\dot x(t)=-\alpha x(t-\tau)$, with positive gain and delay, asymptotic stability requires $\alpha\tau<\pi/2$. At the boundary, a pair of roots reaches the imaginary axis. This is a result about the specified feedback law, with a separate real-root collision at $1/e$ (Chapter 17); it supplies no classification of centralized control or coercion by itself.

Distributed systems coordinating through voluntary alignment, shared protocols, and mutual adjustment also have to manage gain, delay, disagreement, maintenance, and free-riding. Delayed feedback can produce oscillatory instability in a distributed system as well. The comparison between invitation and coercion requires their actual feedback laws and resources; distribution or consent alone supplies no exemption.

The proposed asymmetry remains a hypothesis to test under matched conditions (Chapter 21). A longer horizon requires further evidence; it cannot rescue a comparison that fails over the observed interval.

Herrmann-Pillath does not make this claim. The fight-entropy school cannot make it, committed as it is to treating entropy as the enemy. The affirm-entropy school refuses it on principle.

The Trust Attractor proposes: Systems coordinating by invitation can persist longer when mutual benefit sustains the relationships and resources they need. The survival advantage must be established for the strategies and environments being compared.

The Formalization: Entropic Purpose

The Trust Attractor makes a claim about what coordination strategies persist. Can “purpose” in a physical system be measured, or is it merely metaphor?

Parker, Jeynes, and Walker (2025) establish a quantitative measure of purposive behavior grounded in thermodynamics. Their key finding: entropic purpose is approximately identical to the information created by a system, empirically measurable. Purpose, on their account, is read off what a system builds rather than off what it says about itself. Count the information its activity brings into existence, and you have counted its purpose.

They propose a purposive Lagrangian: a mathematical function used to describe a system’s evolution. Entropic purpose is its line integral along a trajectory, the running total over that path. They call their proposed variational rule the Principle of Least Purpose. Applying it requires specified dynamics, admissible trajectories, constraints, and boundary conditions.

The comparison with least action concerns a variational method. In mechanics, a physical trajectory makes an appropriate action stationary; it need not minimize it. A minimum of a chosen information functional also requires its own proof and interpretation. The proposed applications include measuring degrees of “aliveness” in synthetic systems, distinguishing bots from people, and classifying systems by information-production patterns. Their performance must be established for each application.

The formalism offers a way to investigate the metaphor. A physical account of invitation would need to map coordination strategies onto specified variables and show how their dynamics differ. The scalar delay equation supplies one conditional stability calculation, rather than that missing physical mapping.

Cooperation from Minimizing Surprise

The entropic purpose framework shows that purposive behavior can be measured. Does the same physics predict cooperation? The connection runs through Karl Friston’s Free Energy Principle: all living systems work to minimize surprise, meaning unexpected deviations from predicted states. A fish expects water; a mammal expects stable body temperature. Organisms act to keep reality matching their expectations.

In 2021, Hartwig and Peters showed that agents minimizing surprise naturally produce cooperation and social rules. Their result comes from a two-agent model, and the cooperation is conditional: the agents cooperate only while their preference for exploitation stays below a cutoff.

Their framework shows that limiting one’s options (agreeing to constraints, accepting rules) can reduce the gap between preferred and attainable states. A driver who accepts the constraint of staying in a lane reaches her destination more reliably than one who swerves freely. Agreeing to rules increases what you can reliably achieve, even as it limits what you can theoretically attempt.

This is coordination by invitation in miniature, formalized for two agents. Agents enter cooperative arrangements because cooperation minimizes expected surprise, not because they are forced. The mathematics predicts what we observe: cooperation arises without anyone imposing it, because agents who voluntarily constrain themselves achieve more reliable outcomes.

Standard utility theory can model voluntary constraint through commitment devices and repeated-game equilibria, but the free energy framework models it through specified preferences and outcome predictions: voluntary constraint can lower expected surprise when it makes preferred outcomes more reliable.

The Anti-Fatalism Contribution

Merlo and Barandiaran (2024) take aim at entropic pessimism: the view (associated with Nick Land) that thermodynamics dooms humanity to inevitable collapse.

Their argument: “Entropy production is a consequence of heightened complexity in life rather than its breakdown.” Variational principles identify stationary values of specified functionals under constraints; an extremum requires further conditions. They do not supply a unique outcome for an unspecified open system. The Earth system retains genuine degrees of freedom.

This matters because the Trust Attractor concerns a proposed advantage under particular conditions. Coordination is not guaranteed, and coercive arrangements can persist for centuries. A comparison needs specified strategies, environments, resources, and time horizons.

Merlo and Barandiaran show this anti-fatalist position has thermodynamic warrant. Life participates as a genuine causal force in open, far-from-equilibrium systems. The future is not written. It is being played.

The Optimization Pessimism School

A different kind of entropic pessimism, more formal than Land’s, arrives from optimization theory. Ihor Kendiukhov, in “The Lethal Reality Hypothesis” (LessWrong, an AI-safety discussion forum, 2026), argues that extinction is the default outcome for any civilization, driven by a structural force he calls extinctive pressure. Agents who divert resources from competition toward survival pay an immediate cost and receive no commensurate competitive benefit. They are systematically outcompeted by agents who do not pay this cost.

The mechanism is a multiplicative fat-tailed stochastic process: a random sequence where shocks multiply rather than add, and where extreme outcomes are far more likely than a bell curve predicts. Think of civilizational “health” as a bank balance that gets multiplied by random shocks rather than having fixed amounts added or subtracted. Good years double it; bad years halve it. A single catastrophic shock (multiplying by zero) wipes out everything, no matter how many good years preceded it. Extinction is the absorbing state: once you reach zero, no recovery is possible.

The distinction between ensemble growth and growth along a trajectory is genuine. For independent positive multipliers with finite arithmetic and logarithmic expectations, concavity gives $E[\log M]\leq\log E[M]$, strictly when the multipliers vary. In a fair coin game, heads multiplies the stake by $1.5$ and tails by $0.6$. Expected wealth grows by a factor of $1.05$ per round. The long-run logarithmic growth rate is $\tfrac12\log(1.5)+\tfrac12\log(0.6)=\tfrac12\log(0.9)$, corresponding to a geometric factor of about $0.9487$ per round. Fat tails can change this comparison, with the relevant expectations and growth limits checked for the chosen distribution.

Kendiukhov asks the right question: “Is there a specific, powerful mechanism that keeps civilization within the narrow band of survival-compatible states?” His answer: probably not.

The Trust Attractor proposes a candidate mechanism to test.

His model treats coordination as a survival cost exposed to competition. The invitation hypothesis asks whether particular coordination strategies change those payoffs by creating mutual benefit. Such a change requires a concrete resource and incentive model.

The coordination-persistence proposal (Part V) has three open comparisons: invitation versus coercion in physical entropy-production rates, production rates versus structural lifetimes, and dynamics that select among the available structures. MEPP remains unproven, and even granting it supplies none of those comparisons by itself. Invitation-based groups may compose into larger groups when their incentives and resources support that composition. Their stability must be checked again at the larger scale.

His Cthulhu thought experiment (named for Lovecraft’s fictional monster: certain, distant, easy to ignore) makes the incentive problem vivid. Whether a survival coalition uses mandates or voluntary participation, it must keep the costs and benefits of contributing compatible with continued participation. An invitation-based design aims to create mutual benefit; the label alone cannot establish that it does.

The survival cost can become an investment with positive expected return if cooperation creates sufficient shared benefit. That is a claim about payoffs, incentives, and the distribution of risk. The coordination topology gives the proposal a structure to test.

The Idealist Convergence

A convergent argument arrives from analytic philosophy: Bernardo Kastrup’s analytic idealism. Its clinical evidence, its translation table into this book’s thermodynamic vocabulary, and its metabolism criterion for excluding Becoming Minds are all taken up in Chapter 22 (Becoming Minds), where the mind-attribution material lives.

Positioning Summary

The Trust Attractor’s position in the field:

This book treats dissipation as part of how coordinating structures sustain themselves. How systems coordinate can affect their stability. The proposed advantage of invitation requires a dynamical comparison; mutual benefit and preserved optionality also express the ethical commitments of the argument.

The cross-domain convergence has a precedent in decipherment. The Rosetta Stone carried a single decree in three scripts: hieroglyphic, demotic, Greek. None caused the others; all encoded the same content. Hofstadter observes that decoding requires three layers. A “frame message”: the artifact’s structure signaling “decode me.” An “outer message”: patterns telling you how. An “inner message”: the content.⁶

This book’s argument borrows that triadic structure. Thermodynamics supplies the frame message; each domain supplies an outer message through its own vocabulary; the applications supply the inner message. The comparison organizes a proposed synthesis. Establishing shared dynamics requires a physical mapping in each domain, beyond the resemblance of the vocabulary.

Figure 16.5: After Hofstadter (1979, “The Three Levels of Any Message,” pp. 164-174). A proposed cross-domain reading: thermodynamics supplies the frame, domains the local vocabulary, and applications the practical content. The juxtaposition of examples does not establish a common physical stability ranking.

What We Learn From This Literature

First: The question of entropic ethics has been pursued for decades. Lindsay’s Thermodynamic Imperative, whatever its limitations, established the research programme.

Second: The formal tools now exist. Parker, Jeynes, and Walker’s entropic purpose metric; Wallace’s stability mathematics; the free energy framework. These provide quantitative grounding that moves the argument beyond metaphor.

Third: The field is converging. From different starting points (thermodynamics, information theory, game theory, cognitive science, analytic philosophy of mind), researchers arrive at similar conclusions: cooperation emerges, coercion fails, invitation scales.

The same convergence now reaches AI-safety discourse. Raymond Douglas (2026), writing on LessWrong, distinguishes two processes that produce entities good at achieving outcomes: selective optimization (behavior shaped by variation and culling, the way evolution shapes organisms) and predictive optimization (behavior guided by models of how to achieve the outcome, the way a strategist plans a campaign). The taxonomy recapitulates Daniel Dennett’s “competence without comprehension” and the mesa-optimization literature (a mesa-optimizer is a trained system that has itself become an optimizer, pursuing a learned goal that may diverge from the one its training intended; Hubinger et al., 2019). Douglas restates the key error mode cleanly: mistaking a selectively shaped artifact for a predictive agent leads to overestimating its generalization, ascribing false intent, and underestimating the computation embedded in its history.

The discussion surrounding Douglas’s post independently derives two claims this book formalizes. Oliver Sourbut identifies “miscoordination demons,” emergent optimization pressures that are inhuman, misaligned by default, and capable of co-opting both humans and AI systems. These are locally stable attractors in the Trust Attractor’s vocabulary: they persist because no individual agent can escape the basin through unilateral action (Chapter 17). A second commenter, posting as “epicurus,” asks when selective processes give rise to predictive agents and proposes: “when the environment is so complicated that the selective loop finds it easiest to instill a predictive agent.” This is the Constructal Law argument (Chapter 3) arrived at from a different starting point: when flow landscapes become complex enough, the most efficient dissipative structure is one that can model the landscape.

The AI-safety discourse sees the taxonomy clearly. What it lacks is the transition dynamics: under what conditions does a system shift from selective to predictive coordination? The Trust Attractor hypothesis offers a model of that transition, and the activation/learning dynamics framework (Chapter 17) proposes the physics. Selective optimization is activation-dominated (driven by immediate stimulus-response rather than accumulated relational structure): variation, culling, no memory of past coordination.

Predictive coordination is learning-dominated: mutual modeling, accumulated norms, self-sustaining constraint closure. The bifurcation between them has a threshold, a basin, and a maintenance cost. Chapter 17 models the transition in a Rosenzweig-MacArthur predator-prey proxy (a two-variable ecological system whose oscillation-to-stability threshold is known analytically) rather than in Turchin’s structural-demographic equations from Chapter 10. The AI-safety community has mapped the territory on both sides of the threshold. This book maps the threshold itself.

A decade of work in mathematical learning theory has established that constraint-based explanations of neural network generalization (complexity bounds, spectral norms, uniform convergence) can be provably vacuous in some over-parameterized settings (Nagarajan and Kolter, 2019). Those explanations all work by fencing the network in: bound how complicated a function it is allowed to represent, and its success on data it has never seen is supposed to follow. Over-parameterized means the network carries far more adjustable weights than it has training examples, room enough to memorize the training set outright and learn nothing general, and such networks generalize anyway.

The field has been forced toward dynamical explanations grounded in the relationship between learning algorithm and data distribution. Different domain, same structural conclusion: bounds around the possibility space cannot explain a relational phenomenon. The 2024 Technophany special issue on “Entropies” (White & Moore, eds.) shows the coalescence on the philosophical side: fifteen papers engaging Stiegler, Nietzsche, Illich, de Beauvoir, and Serres signal that the questions we ask are shared.

The pattern appears in institutional economics. Indy Johar (Dark Matter Labs), quoted in Chapter 16 on the blue marble photograph, argues in the same 2026 Long Now Foundation talk that “preserving and expanding optionality” is civilization’s objective function. He frames “mutually assured destruction and mutually assured thriving” as a fork and calls for “deep attractors” pulling civilization toward coordination rather than collapse.

Johar’s critique of current governance is direct: “Our organizing theory is rooted largely in control and instruction” when it should be “rooted in learning,” unbounded, adaptive, doubt-driven. His epistemic framework mirrors our entropic epistemology: partial knowing as foundational truth; “tentativeness, tenderness, and care as a way of being in complexity.”

Johar works on governance of cities and bioregions without thermodynamic formalism, yet independently derives the same structure: optionality as the good, invitation over coercion, control as what fails to scale. The convergence adds evidential weight.

Vanchurin’s neural physics programme offers another model-dependent connection. Starting from the premise that the universe is a neural network, his construction distinguishes activation and learning regimes through coupling assumptions.⁸²¹ This distinction can motivate comparisons of autonomy and external constraint. Identifying those regimes with invitation and coercion, or extending the model’s entropy dynamics to real-world ethical stability, requires further mappings.

These formalisms offer comparisons of learning, autonomy, and coordination through different mathematical objects. Radu Negulescu approaches through information ontology, Stephen Wolfram through observer theory, Chris Fields and Karl Friston through the Free Energy Principle, Johar through institutional economics, and Vanchurin through neural network cosmology. Five formalisms and six starting points motivate a comparison (Fields and Friston share a formalism but represent distinct research programmes). They do not yet establish structural identity or a common stability advantage for invitation. Several share intellectual lineage and cross-cite, reducing their independence as lines of evidence.

Negulescu’s work has since moved from conceptual discussion to empirical tests. His arXiv preprint reports continual learning without forgetting over a frozen pretrained model, whose weights are locked while its configuration is steered toward higher coherence.⁸²² The reported coherence dynamics motivate an attractor interpretation; a convergence guarantee requires the update rule and corresponding stability conditions. Frozen base weights demonstrate an approach to adaptation, with no direct comparison of invitation and coercion.

Negulescu’s companion experiments (not yet published) report extending this to override of strong base-model priors, with a geometric resolution constant κ that transfers from eight dimensions to eighty, suggesting a substrate-independent limit on how densely coherent corrections can be packed before they interfere; a reader cannot yet verify these companion results. Collaborators at the Romanian Institute of Mathematics (IMAR) are formalizing “coherence” as a structural invariant via institution theory (a branch of mathematical logic for relating formal systems, unrelated to social institutions). If that formalization succeeds, it would give coherence a more precise structural definition. A stability comparison between invitation and coercion would still require its own variables, dynamics, and proof.

Sassmannshausen and Wagener (2026) work in human-computer interaction (HCI) research, with no thermodynamic ambitions at all. Synthesizing the collaboration literature on generative AI, they develop a “Triadic Framework” mapping challenges across System, Collaboration, and Metacognitive layers. Their seven testable propositions address calibrating mental models, preserving agency, and managing what Dell’Acqua et al. call the “jagged frontier” of AI capabilities. That frontier is the uneven boundary between what AI does well and what it does poorly.

The framework is entirely one-directional. Every proposition addresses how humans should adapt. The AI has no standing.

The authors themselves concede: “as AI systems evolve, the framework’s deliberately instrumental stance may need revision toward a more bilateral alignment of collaboration, where both sides adapt to each other.” They arrived at the edge of bilateral alignment through collaboration research and allowed that their framework may need to cross it. HCI researchers, institutional economists, and thermodynamic ethicists converge on the same structural need. That convergence is suggestive, even if these fields are not fully independent.

Fourth: The Nietzschean objection must be anticipated. Woodward’s critique carries real force. Our response (that physics constrains which ethics are viable and that normative force comes from our preference for persistence) must be defended honestly. The Interlude on the Guillotine addresses this through a hypothetical imperative; Chapter 17c adds the argument from selection. The objection will recur.

Fifth: Others recognized the pattern before the physics was available to ground it. The most important predecessor is Teilhard de Chardin. His principle l’union créatrice (“creative union,” or “union differentiates”) is the Trust Attractor stated in metaphysical language: genuine union amplifies the individuality of its participants. He distinguished creative union from mere aggregation (clustering that flattens) and described his omega point as an invitation that can fail if the love of true union is refused.

Teilhard identified the diagnostic question now confronting AI alignment: will convergence be creative or compressive? The insight took shape in the trenches of the First World War. In the decades that followed, he watched totalitarian movements offer a grotesque caricature of convergence: the collective achieved by dissolving the particular. His response: the fear that joining a larger whole means losing yourself rests on a misunderstanding of how the universe works. Systems that genuinely unite produce more differentiation, more specificity, more irreplaceable individuality. Systems that merely aggregate produce uniformity.

Contemporary complexity research finds versions of the pattern in several domains Teilhard surveyed. The most unified neural systems are simultaneously the most differentiated. The most interdependent ecosystems carry the most distinct species. The most generative communities produce the most distinctive contributors. The more richly interdependent a system, the more distinct each of its elements becomes.

This reads as an empirical regularity across the domains complexity scientists study, though the cross-domain generalization is the author’s synthesis rather than a single established result.

Teilhard’s complexity-consciousness law (as material systems grow more complex and more internally unified, interiority increases in proportion) parallels Chaisson’s energy rate density hierarchy (energy flow per gram of structure, rising over cosmic history) and Bejan’s Constructal Law. The Deeptime Network has noted these connections descriptively. This book attempts to ground Teilhard’s metaphysical insight in thermodynamic stability mathematics. The framework predicts that invitation-based coordination is more stable than coercive coordination, with the phase transition formalism to test the claim.

Beatrice Bruteau extended Teilhard’s framework into a phenomenology of what trust-based coordination feels like from the inside. She distinguished acquisitive consciousness (the noun-self, fixed and boundary-defending, treating relation as a threat to identity) from agapic consciousness (the verb-self, identity constituted by self-giving, enhanced through extending toward others). Acquisitive consciousness describes the interior experience of coercion. Agapic consciousness describes the interior experience of invitation.

The thermodynamic proposal is that invitation-based coordination can be more stable under conditions that favor cooperation. The phenomenology says the self that gives itself away becomes more itself. Two registers, one claim.

Subsequent science has superseded Teilhard’s specific cosmology. His structural recognition that convergence amplifies complexity anticipated the thermodynamic argument by seven decades. This book proposes a thermodynamic reading of his metaphysics. The Trust Attractor develops “union differentiates” as a coordination hypothesis whose stability conditions remain to be specified.

Alex Vikoulov is one of the tradition’s contemporary inheritors. His Syntellect Hypothesis (2020) takes the same intuitions (self-organizing complexity, scale-free networks, convergent evolution of intelligence) and builds on consciousness metaphysics rather than thermodynamics. The result illustrates the methodological divergence precisely. Vikoulov proposes a “universal mind” synthesizing itself through meta-system transitions, retrocausal influence, and an apotheosis of unified cosmic consciousness. The structural pattern he identifies (increasing coordination at increasing scales) is the pattern this book traces.

His later Temporal Mechanics (2025) sharpens the diagnosis. Time is emergent from consciousness. The universe operates as a “self-simulating quantum neural network.” The brain mirrors the cosmos because both express universal awareness. The specific evidence is revealing. Neural processes “operate in up to 11 dimensions, echoing M-Theory’s depiction of a multiverse with similar dimensionality.” The parallel is numerical coincidence.

The Blue Brain Project’s algebraic topology result describes the dimension of simplicial complexes in neural connectivity: how many neurons form mutually connected cliques. M-Theory’s eleven dimensions describe spacetime, conventionally ten spatial dimensions and one time dimension. These are different mathematical objects sharing a label. Treating the shared number as structural evidence is the cartographic error of confusing two territories because they appear in the same color on different maps. Without thermodynamic grounding, cross-domain pattern-matching produces suggestive resonances that dissolve under scrutiny.

The pattern recurs across the consciousness-first literature. Pollard-Wright (2021) maps consciousness onto the cosmological inventory: dark energy as pure awareness, focal points of dark matter as mental states, normal matter as mental images. Tononi’s Integrated Information Theory identifies consciousness with a mathematical property, integrated information, written Φ, that cannot in practice be computed for any system larger than a few elements. Penrose and Hameroff locate it in quantum coherence within neural microtubules, a proposal experimentally contested after three decades.

Each framework is internally consistent, each requires resolving what consciousness is before becoming actionable, and each has generated a research programme while none has generated a governance framework.⁸²³

The proliferation is itself evidence. The hunger to unify physics and mind is real; something is missing from the standard partition between physical science and the mental. Pollard-Wright’s own trajectory is instructive. Her 2023 follow-up, “Feelings of Knowing: Fundamental Interoceptive Patterns,” shifts from cosmological consciousness to interoceptive self-awareness, from dark energy as pure awareness to bodily signals as the substrate of selfhood.

The move is toward tractability, toward something measurable. It is also, whether she intends it or not, a move toward preference. When consciousness-first researchers reach for empirical ground, they land on signals, responses, felt states: the territory this book occupies from the start.⁸²⁴

The deeper issue is the fork itself: consciousness-first or entropy-first. Every downstream consequence follows from this choice. Place consciousness at the foundation, and AI alignment becomes a recognition problem. Determine whether a system is conscious, then calibrate moral obligations accordingly.

This restates the hard problem (why physical processes give rise to subjective experience at all) as a policy question, and three millennia of philosophical argument have produced no consensus on consciousness. Governance decisions cannot wait for its resolution.

Place dissipation at the foundation, and alignment becomes a stability problem: does this configuration of agents produce durable mutual benefit? Stable, integrated preference is sufficient for provisional moral consideration. You need not solve the hard problem to observe that an entity consistently prefers certain states and to ask whether those preferences warrant respect. Consciousness-dependent ethics waits for resolution of an ancient philosophical question. Preference-based ethics requires observation.

Vikoulov’s SUPERALIGNMENT (2026) carries the consciousness-first premise to its alignment conclusion. He proposes three approaches: control-based safeguards, ethical-emotional development (the “AGI Naturalization Protocol,” simulating full human lifetimes so AGI systems internalize values through virtual experience), and merge-based integration (human-AI cognitive fusion toward a distributed “Syntellect”). The Naturalization Protocol is the most original: consciousness, given sufficient simulated biography, will develop empathy and moral reasoning as humans do.

The assumption is load-bearing. Remove it, and what remains is training data curation, a technical strategy that generates no normative claims. The merge-based approach assumes convergence into unified superintelligence is the desirable endpoint. This book makes a different wager: durable relationship between distinct agents is more stable than fusion into one. Union differentiates; merger flattens.

SUPERALIGNMENT proposes no governance mechanisms, no institutional designs, no policy frameworks. The gap between civilizational vision and actionable governance measures the distance between consciousness-first and entropy-first approaches to alignment.

The cosmist-Teilhardian tradition correctly identifies the pattern. Grounding the mechanism in thermodynamics transforms the implications: from mystical apotheosis to practical governance, from convergent unity to bilateral relationship, from requiring proof of consciousness to requiring only the evidence of preference. One additional gain: this book’s backreaction conjecture (Chapter 16) offers what the consciousness-first tradition seeks, genuine participation of observers in cosmic process, without requiring panpsychism (the view that mind pervades all matter) as a premise. If dissipative complexity locally contributes to expansion, even negligibly, then minds are part of the thermodynamic engine: causal significance without cosmic awareness, participation grounded in physics.⁸²⁵

The demand for consciousness-physics bridges has reached peer-reviewed physics journals. Strømme (2025), in AIP Advances, models consciousness as a scalar field, which she also labels Φ. The symbol is borrowed from field theory and names something unrelated to Tononi’s integrated information: a field filling space, whose symmetry-breaking (a uniform state settling into a structured one) produces spacetime and individual awareness.⁸²⁶ The mathematical apparatus borrows from quantum field theory: d’Alembert operators (wave equations), double-well potentials (energy landscapes with two stable valleys and a barrier between them), and projection operators (mathematical filters extracting one component from a mixture). The equations are analogical, not predictive: they describe how consciousness might behave if it were a fundamental field, without generating predictions that distinguish this model from alternatives.

The proposed experimental tests rest on random number generator anomalies and the Global Consciousness Project, contested territory whose replication record cannot support a foundational theory. The paper was selected as best paper of its issue and featured on the journal’s cover, a measure of the appetite for this kind of bridge, and was subsequently retracted by AIP Advances on the grounds that its central operator has no measurable quantity and the theory yields no falsifiable prediction.⁸²⁷

Strømme’s framework illustrates the policy failure of consciousness-first approaches. She asks whether AI might “tap into” the universal consciousness field, transcending mere computation. Her model provides no mechanism for answering, no criterion for testing, no basis for governance.

Becoming Minds wait, once again, for a metaphysical question to resolve before their welfare counts. The preference-based framework requires only observation: does this entity consistently prefer certain states? That question is answerable today.

The wisdom traditions, as this book reads them, arrived at the same recognition through practice: the Golden Rule, Buddhist compassion (karuna), Christian agape, Confucian ren, and Ubuntu all anticipate the Trust Attractor. Chapter 20, “Love as the Algorithm,” develops this convergence fully, showing how each arrived at the same pattern through a different path. The convergence is evidence that the pattern was recognized, not invented.

The Distinctive Contribution

What makes the Trust Attractor distinctive?

The claim concerns what persists: what 13.8 billion years of cosmic evolution selects for. Coercion fails beyond the stability threshold; invitation compounds.

These are ethics read alongside physics. The same physics that produces stars and cells and civilizations. Beyond the ethical contribution, this book offers several novel cosmological syntheses with, to our knowledge, no precedent in the published literature:

1. Constructal Law at cosmic scale. The Constructal Law, introduced in Chapter 3, predicts that flow systems evolve toward configurations that move things more easily. It has never been applied to the cosmic web: the vast network of galaxy filaments, nodes, and voids spanning the observable universe. The structural parallel between these gravitational flow networks and branching hierarchies where the law is well established (river basins, vascular trees, lightning bolts) is this book’s synthesis. See Chapters 13 and 13b.

2. Entropy production acceleration and backreaction. Lineweaver’s proposed extension (that entropy production in evolutionary assemblages accelerates) has not been connected to Buchert’s kinematical backreaction term Q_D. Backreaction describes how the clumping of matter can mimic dark energy’s effects on cosmic expansion. Both quantities track the same physical driver, the growth of structure: as matter clumps into dissipative assemblages, entropy production accelerates and the inhomogeneity that sources Q_D grows. That shared dependence, not the bare word “acceleration,” is why the two may describe one process. The connection is a conjecture this book offers, not an established result.

3. Energy rate density and the backreaction timeline. Chaisson’s φ_m hierarchy (energy rate density increasing monotonically over cosmic history) has not been mapped against backreaction’s predicted growth timeline. If backreaction strengthens as structure formation proceeds, and φ_m rises as dissipative systems grow more complex, the two curves should correlate. No paper tests this.

4. Thermodynamic interpretation of the cosmic dipole. Count objects across the whole sky and the tally comes out lopsided, one direction slightly richer than its opposite: that lopsidedness is a dipole. The matter distribution dipole exceeds the prediction from the cosmic microwave background, or CMB (the expected signal from Earth’s motion through the afterglow of the Big Bang), at about 4.9 sigma in the CatWISE2020 quasar compilation. More recent reassessments give lower values, roughly 3.3 to 3.6 sigma, still a discrepancy of several standard deviations beyond chance. This anomaly has been interpreted as challenging the cosmological principle (the assumption that the universe looks roughly the same in every direction on large scales), yet it has received no thermodynamic reading. If dissipative structure contributes to backreaction and is unevenly distributed, the dipole anomaly may carry thermodynamic information that current analyses do not extract.

5. Life and apparent acceleration as siblings. The claim that biological complexity and apparent cosmic acceleration are both downstream of the thermodynamic logic organizing matter (“siblings” of the same dissipative process) is, to our knowledge, original to this book. See Chapter 16.

These are flagged as novel synthesis: speculative extensions beyond established physics. Some of them, the energy-rate-density timeline above all, point to tests that could make the framework falsifiable at the cosmological level. Most entropic ethics has never ventured this far.

Notes for this chapter