Chapter 16: Life and Cosmos

Does life on Earth make a causal difference to the physical evolution of the universe? Not a question of significance or purpose; those come later. Does biology push back on the cosmos, or is life a passenger, complex and fascinating yet cosmically insignificant, a brief eddy in flows that would proceed without it?

The conventional answer is stark: passenger. The galaxies would spin the same, the stars burn the same, the expansion proceed the same, whether or not any planet developed biology. Witnesses, only witnesses. The bilateral cosmology of the preceding chapter suggests otherwise. The answer determines whether life is a beautiful accident (passenger), a force that reshapes cosmic structure (participant), or a geometric inevitability the universe’s architecture produces wherever conditions permit (structural consequence), as naturally as gravity produces stars.

A note on method: the book’s core argument does not require the answer to be “participant” or “structural consequence.” Readers who prefer firmer ground may proceed to Part V without losing the thread. What follows separates three categories: (1) established observations, (2) grounded inference from published physics under active testing, and (3) frank speculation. Each is labeled clearly.

The evidence will not resolve this cleanly. The third possibility, which requires the bilateral physics of the preceding chapter, may reframe the question entirely.

Figure 16.1: Left: the conventional view, where the cosmos produces life through one-way causation. Center: the participant hypothesis, where life feeds back on cosmic structure through dissipative processes. Right: the structural consequence thesis, where life is a geometric expression of bilateral architecture, emerging as naturally as gravity produces stars.

The physicist Sara Walker frames the challenge sharply: our models “assume some kind of initial condition and some fixed rule that governs the universe for all time.” They assume “nothing about the universe fundamentally changes because of the patterns in it.”⁵ That assumption contradicts daily experience. Our ideas, our culture, the things we do change the structure of reality around us. Walker’s question is this chapter’s question: what changes if the patterns matter?

Part One: What We Observe

[The following sections describe established science.]

Fine-Tuning

Physical reality appears fine-tuned for life. The physical constants (gravity’s strength, the electron’s mass, the carbon-12 resonance that allows stars to forge heavier elements¹²) permit complex chemistry, stable stars, and eventually living things. Change them slightly, and the universe becomes sterile. The fine-tuning is observed. The explanation is not settled: weak anthropic selection bias¹³ (observers can only find themselves in a universe that permits observers); a multiverse; or stronger claims⁵⁷.

A more productive approach: cosmological attractor models show that inflationary predictions remain stable across wide variations in underlying parameters.²⁶ An attractor is a state a system settles into regardless of starting conditions, like a marble released anywhere inside a bowl rolling to the bottom. Both the early universe’s initial conditions and dark matter’s abundance may be dynamically determined, self-tuning through freeze-out dynamics (particle abundances locking in as the expanding universe cools) akin to self-organized criticality,⁴⁶ rather than coincidentally set. The fine-tuning problem shrinks from two directions.

The universe is rolling toward a valley floor, drawn by attractor dynamics rather than teetering on a razor margin.

A different approach narrows the multiverse explanation. Standard eternal inflation posits an infinite number of pocket universes with different physics, rendering fine-tuning trivially explained: every configuration exists somewhere. Hawking and Hertog’s holographic cosmology (Chapter 15) constrains this picture. By projecting eternal inflation onto a holographic boundary, they showed the multiverse is finite: a reduced range of possible universes, no longer an infinite sample.⁶⁹¹

In an infinite multiverse, the question “why these constants?” dissolves into tautology. In a finite one, specific configurations demand accounting. The question recovers its force, and the answer points toward selection: the configurations that persist are the ones whose constants sustain dissipative complexity.

A thermodynamic framing sharpens the question. Fine-tuning discussions traditionally ask: why do constants permit observers? The question smuggles in an anthropocentric assumption. A prior question: why do constants permit maximal entropy production through complex dissipative structures? Hurricanes, cells, and civilizations all qualify.

Bejan’s Constructal Law (Chapter 3) implies any constants permitting flow-system evolution would generate increasing structural complexity. The fine-tuning targets dissipation, of which conscious observation is a late and complex instance. The improbability shrinks when the target shifts from conscious life to thermodynamic creativity. It looks large only when the calculation omits the learning dynamics that concentrate order against thermal disruption.^57a

A framework from theoretical physics strengthens this reframing. Vanchurin, Wolf, Katsnelson, and Koonin identified seven physical principles: a loss function, a hierarchy of scales, frequency gaps between levels, renormalizability, extension, replication, and information flow.⁶⁹² Three of those need translating.

A loss function is a measure of how badly a system is doing, the quantity it moves to reduce. A frequency gap means the processes at one level run so much faster than those at the level above that each sees the other as either a blur or a fixed backdrop. Renormalizability means the same description keeps working as you step back and coarse-grain. All seven are physical rather than biological. Any universe satisfying them produces multilevel learning systems, of which biological life is one instance. Their conclusion: the universe is self-tuned for life emergence. The learning dynamics that all evolving systems undergo drive toward the complexity threshold that biology crosses.

Within the Vanchurin framework, a more radical possibility arises: the constants themselves may be learned parameters, updated through the universe’s training dynamics rather than fixed at the Big Bang. Current searches for drift in the fine-structure constant (the number that sets the strength of electromagnetism) over cosmic time have found none. The DESI baryon acoustic oscillation survey, which maps a sound-wave echo frozen into the spacing of galaxies (see below), finds that a model in which dark energy’s equation of state (its ratio of pressure to energy density) evolves over cosmic time fits some combined datasets better than a fixed cosmological constant.²² That equation of state is a fitted cosmological function. It is not a fundamental constant observed changing in the laboratory. Whether this reflects learned-parameter dynamics or conventional field theory remains open. [Speculation: the evolving-constants hypothesis is consistent with Vanchurin’s framework; DESI provides circumstantial support for one parameter but not a general test.]

The mechanism is the second gear introduced in Chapter 6. Standard probability calculations for life’s origin use activation dynamics alone: the Second Law running in one direction, making self-replicating structures vanishingly improbable through random assembly.

Learning dynamics run in the opposite direction, concentrating information and stabilizing structure against thermal disruption, the way a child at the beach actively builds and repairs the castle. When both gears are included, the probability of self-replicating systems rises from astronomically small to thermodynamically expected. The anthropic principle becomes unnecessary. The calculation that demanded it was incomplete: it omitted half the dynamics.

Both frameworks converge: fine-tuning targets dissipation, and self-tuning produces learners. Both dissolve the anthropic puzzle by widening the target from conscious observers to the thermodynamic creativity that eventually produces them.

The narrowness of the target is quantified directly. In 2023, Kostya Trachenko showed the fundamental constants constraining liquid viscosity permit only a slim bio-friendly band.^57b Shifting the Planck constant or electron charge by a few percent would push water’s viscosity outside the range cellular transport, diffusion, and molecular machinery require. The constants were set by nucleosynthetic constraints (the physics of element-forging) in early stars, billions of years before cellular life existed. They were not tuned for cells.

Yet they happen to permit exactly the viscosity cells need. The coupling is tight: the same constants that govern stellar nucleosynthesis also govern the fluid dynamics of a white blood cell. Two constraints, separated by billions of years and dozens of orders of magnitude in scale, threaded through the same narrow eye.

Trachenko reaches toward the framework already in hand: “Through evolutionary mechanisms, fundamental constants may be the result of nature arriving at sustainable physical structures.” Biological evolution is one instance of dissipative selection, the thermodynamic pressure favoring configurations that sustain complex entropy-producing structures (Chapter 6). If the same pressure operates at the level of the constants themselves, the resemblance to evolution is inheritance, not analogy. Organisms are fine-tuned because the same selective logic that shaped the constants operates at every scale.

The pattern extends to neutrino mass. The DESI baryon acoustic oscillation measurements, combined with cosmic microwave background data, constrain the sum of neutrino masses to less than 0.072 eV (electron volts, the particle physicist’s unit of mass-energy) at 95% confidence.⁶⁹³ This barely exceeds the 0.059 eV minimum that particle physics requires. Neutrino mass suppresses structure formation: more massive neutrinos stream out of gravitational wells, smoothing the matter distribution and preventing small-scale structure from collapsing. The universe has minimized the one particle-physics parameter that would impede complexity, leaving maximum room for structure, chemistry, and life.

The standard fine-tuning debate offers two options: coincidence or design. Dissipative selection offers a third: directional pressure without a director. The Second Law operating across 13.8 billion years of cumulative selection at every scale simultaneously.

This distinction matters because a popular alternative draws the opposite conclusion. The von Neumann-Wigner interpretation holds that wave functions remain in superposition (all of a particle’s possibilities held open at once) until a conscious observer collapses them into one definite outcome. The leap from quantum observer to Cosmic Mind is short and unfalsifiable. No experiment can distinguish “consciousness collapsed the wave function” from “decoherence happened” (ordinary interactions with the environment settling the outcome, no observer required). The interpretation generates no predictions and solves no problems that decoherence leaves open.

The thermodynamic argument developed here differs in structure. Life matters to the cosmos because it dissipates. Complex minds are the densest concentrations of entropy production per unit mass the universe has yet produced (Chapter 4). If life pushes back on cosmic structure, it pushes through thermodynamic work, not through witnessing. The dissipative backreaction conjecture developed below names its own falsification conditions; the consciousness-causes-collapse interpretation does not.

[Grounded inference: the Constructal Law is established (Bejan 1996); the claim that fine-tuning targets dissipation rather than observers is novel synthesis. The claim that the universe is self-tuned for life emergence is Vanchurin et al.’s (2022), supported by their mathematical framework.]

The cosmos may have shaped life’s molecular architecture even more directly. Globus and Blandford (2020) proposed that cosmic rays preferentially ionized one handedness of DNA over the other in early organisms, creating a tiny yet systematic differential mutation rate.^99f Handedness, or chirality, means left-right asymmetry: the difference between a left and a right glove, which no amount of turning will undo. The mechanism: polarized muons (short-lived heavy cousins of the electron, born spinning in a preferred direction) from weak-force-asymmetric pion decay damage one chirality slightly more than its mirror image. These particle showers inherit a chiral bias from the weak nuclear force, the only fundamental force that violates mirror symmetry. Any single cosmic-ray strike has a vanishingly small effect, yet the bias is systematic and unidirectional, rooted in particle physics rather than local accident.

If confirmed, the most basic structural feature of every living molecule on Earth, the direction DNA spirals, was set by the weak force, mediated through stellar violence. The cosmos did not merely permit life’s chemistry. It selected its geometry.

Fine-tuning discussions focus on the constants, the settings on the dials. Less attention goes to the scaffolding that translates permissive constants into actual structure. Dark matter provides that scaffolding: roughly 84% of all matter, generating the gravitational wells into which ordinary matter falls, condenses, and ignites.

Without dark matter’s invisible architecture, permissive constants would have produced only diffuse gas: no stars, no heavy elements, no planets.

In January 2026, JWST produced the highest-resolution map yet of dark matter distribution, using 250 hours of observation in the COSMOS deep field.⁶ Dark matter and luminous matter distributions overlap almost perfectly. The constants set the rules. Dark matter built the stage.

The stage required yet another layer of construction. As Chapter 14 details, many elements essential for biology (iodine, bromine, molybdenum, uranium, thorium) were forged in neutron star mergers: kilonovae driven by gravitational wave inspiral.¹⁰

Fine-tuned constants permit stars. Dark matter scaffolds galaxies. Gravitational wave emission drives the collisions that produce r-process elements, elements heavier than iron, forged by rapid neutron capture. Without those elements: no thyroid hormone, no collagen, no mitochondrial enzymes, no plate tectonics.

Our Sun is unusually hot and bright for its class,⁸⁴ providing the steep energy gradient that makes Earth’s complexity possible. The biologist Wolfgang Nitschke captures the next step: “When photosynthesis entered the picture, life connected up to the cosmos.”^84a From that moment, every dissipative structure on Earth’s surface was ultimately powered by a star.

Self-Similar Structure

In 2020, Franco Vazza and Alberto Feletti published a quantitative comparison of the brain’s neural web and the cosmic web of dark matter filaments.¹ The structures differ by a factor of roughly 10²⁷ in scale (a billion billion billion), yet display the same node-connection distributions, clustering patterns, and efficiency-connectivity trade-offs. The total node counts are comparable: 86 billion neurons versus roughly 100 billion galaxies. (Galaxy-count estimates range from about 100 billion to 2 trillion depending on method; the comparison holds on the lower estimate.)

Place a microscope image of brain tissue next to a map of the cosmic web: the node clusters, the connecting filaments, the voids between them are statistically indistinguishable in their network properties. The same organizational principles operate across vastly different scales: Constructal Law, thermodynamic flow optimization, and network mathematics.

Figure 16.2: A 100-megaparsec box of simulated dark matter at the present day, evolved by gravity alone from the nearly smooth early universe (the web reader plays the full 13-billion-year formation as an animation). Each strand is a filament channeling matter toward the knots where strands cross; the voids between them emptied themselves to feed it.

The resemblance runs deeper than static structure. In 2012, Krioukov and colleagues proved that the causal network of spacetime’s large-scale architecture in an accelerating universe is a power-law graph with strong clustering: a network in which a few richly connected hubs hold most of the links, and the neighbors of any node tend to be linked to each other.⁶⁹⁴ The topology class is the same as the Internet, social networks, and neural circuits. The growth dynamics of expanding de Sitter spacetime (the mathematical model of a uniformly accelerating universe) and the preferential attachment dynamics of complex networks are asymptotically equivalent: the same equations govern both in the large-N limit.

A growing brain and an expanding cosmos arrive at the same network topology because they execute the same growth algorithm: balancing local connections between nearby nodes against long-range shortcuts to highly connected hubs. The Constructal Law describes the shapes this algorithm produces. Krioukov’s proof establishes that the algorithm is substrate-independent.

In both systems, the active network (neurons, galaxies) constitutes roughly a third of the total mass-energy, embedded in a medium (water, dark energy) that appears inert yet is structurally essential. Signals propagate through this medium, and expansion stretches the network into its observed form. The ratio coincidence may be shallow; the structural relationship is not. An active minority embedded in a necessary medium is the cognition-regulation dyad of Chapter 8, recapitulated at cosmic scale.

A deeper comparison awaits the right mathematical tool. Persistent homology, which tracks how topological features (connected components, loops, voids) appear and disappear as a density threshold sweeps through a structure, has been applied independently to the cosmic web and to neural microcircuits. Pranav and colleagues computed Betti curves (counts of those features, plotted as the threshold sweeps) for the cosmic web in simulations, finding that filamentary tunnels peak at an intermediate density threshold while void boundaries peak at low density.⁶⁹⁵ Reimann and colleagues found directed cliques up to seven neurons deep in the Blue Brain Project’s cortical reconstruction.⁶⁹⁶

The same mathematical invariants describe both structures, yet no one has placed them on a common axis. Vazza and Feletti acknowledged the gap explicitly: they chose simpler metrics for their comparison because persistent homology was “not readily applicable to both networks” with existing methods.¹ The data now exist on both sides. A normalized comparison of Betti curves across cosmic, neural, and vascular networks, using a shared filtration strategy and a common significance threshold, would test whether the topological resemblance extends beyond the graph metrics Vazza and Feletti measured.

The 2026 JWST dark matter mapping strengthens this considerably. It reveals the cosmic web at twice Hubble’s resolution: precise thin filaments displaying the branching, hierarchical geometry the Constructal Law predicts, with dark matter and ordinary matter co-evolving over cosmic time.⁶

eROSITA’s X-ray survey detected X-ray emission from 7,817 cosmic filaments at nine-sigma significance, far beyond the five-sigma bar physics sets for discovery. Of that signal, the warm-hot intergalactic medium itself contributes 5.4 sigma; the remainder comes from unmasked sources.³⁸ The detection is robust, and it matters because those filaments hold a major reservoir of the universe’s “missing baryons,” the ordinary matter that earlier accounting could not locate.

In 2025, Tornotti and colleagues produced the first high-definition direct image of a cosmic filament’s internal structure, using hundreds of hours of MUSE spectrograph data on ESO’s Very Large Telescope.⁶⁹⁷ The three-million-light-year filament connects two quasar-host galaxies at redshift 3.22. (Redshift measures how far cosmic expansion has stretched a source’s light on its way to us, which makes it a clock as much as a distance: the higher the number, the earlier the epoch being seen.) It revealed resolved internal gas distributions, radial density profiles, and a clear boundary between circumgalactic medium and intergalactic filament gas. Previous detections relied on indirect absorption signatures. The image confirms that filaments are structured transport channels, not diffuse bridges.

The cosmic web is a transport network, actively moving matter and energy along its channels. The filaments rotate as well. MeerKAT observed a fifty-million-light-year filament spinning at roughly 110 km/s, with fourteen galaxies turning in synchrony.²⁷ The same branching topology extends to mycelial fungal networks,²⁸ the Radcliffe Wave,⁸⁵ and the magnetized gas tunnel surrounding our solar system.⁸⁶

The engineer Adrian Bejan, whose Constructal Law predicts these convergences, sharpens a distinction that matters for what follows.⁵⁸ Evolution and irreversibility are two distinct phenomena governed by two distinct laws. The Second Law governs dissipation, the tendency toward equilibrium, like a hot cup of coffee cooling to room temperature. The Constructal Law governs design, the tendency of flow systems to evolve toward configurations providing greater access to flow, like a river delta branching into ever-finer channels to reach the sea.

Dissipation is what happens to energy. Design is what happens to structure. Life operates at their intersection.

The case shifts from structural similarity to algorithmic convergence. In 2020, Burchett and Elek used an algorithm modeled on the slime mold Physarum polycephalum to reconstruct the cosmic web from galaxy survey data.⁷ This organism had already self-organized a transport network matching Tokyo’s railway system in efficiency, fault tolerance, and cost.⁸

Applied to galaxy positions from the Sloan Digital Sky Survey, the algorithm produced a filament map matching dark matter distributions from cosmological simulations. The map was validated against 350 quasar spectra from Hubble.

The slime mold’s membranes push outward in a synchronized wave in every direction, exploring isotropically. When a membrane encounters a food source, nearby membranes relax, allowing subsequent pulses to send more material toward the find. The organism imposes no prior structure on its search. It lets the environment reveal its geometry.

As the slime mold specialist Simon Garnier observes: the algorithm is “not really biased by the first direction you decide to look in; capable of exploring everything at once.”⁶⁹⁸

This is why it outperforms human-designed algorithms. Our methods impose assumptions: which directions to favor, which connections seem plausible, which features to weight. The slime mold lets the data speak. Command-based mapping misses features; invitation-based mapping discovers them.

Hasan and colleagues at New Mexico State University extended the approach in 2024.⁶⁹⁹ They gave the Physarum algorithm galaxy positions as “food” across a simulated universe at various time points and let it map connections across cosmic history. The slime-mold map produced a cleaner filament structure than any human-designed algorithm the team had tried. It was sensitive to smaller features and traced dark matter more easily.

The temporal dimension revealed something new. Early in cosmic history, neither the proximity nor the thickness of the web’s filaments affected the galaxies strung along them. The filaments were scaffolding, nothing more. As the universe matured, the relationship changed: material pulled into the web eventually disrupted star formation in galaxies that orbited too close to the densest strands. Hasan calls these filament environments “galactic ecosystems,” and the name is apt: the web is habitat, and habitat shapes what can grow.

The finding has a phase-transition structure. Below some threshold of cosmic maturity, filament proximity is neutral; above it, proximity suppresses star formation. Galaxies that continue forming stars sit at the right distance: connected enough to receive material, autonomous enough to process it on their own terms. Coupled too tightly to the web, a galaxy’s generative dynamics are suppressed.

This is the coordination-without-coercion principle (Chapter 17) written in galaxy surveys: the galaxies that thrive are those invited into relationship with the filaments rather than consumed by them.

The astrophysicist Ari Maller identifies the deeper lesson: “The crucial difficulty in using the cosmic web to constrain galaxy formation is in describing it with the accuracy needed to observe its effect.”⁷⁰⁰ The slime mold revealed no new physics. It revealed physics already present, invisible to instruments lacking the descriptive precision to detect it.

The pattern recurs throughout this book: Barabási’s surface minimization (Chapter 3) exposed the brain’s real wiring principle by replacing the wrong metric with the right one; learning dynamics (Chapter 6) revealed life’s thermodynamic probability by including the gear that activation-only models omit. The right framework reveals what the wrong framework hides.

The finding is testable beyond simulation. New surveys are stretching observations further back in cosmic time, and Hasan’s conclusions about the phase transition in filament influence can eventually be compared against older glimpses of the real cosmic web. If the effect appears in observational data, the result graduates from grounded inference to established observation.

A biological organism evolved over a billion years produces the same network topology as gravitational structure assembled over 13.8 billion years. The optimization problem is identical: connect distributed nodes through the most efficient flow network. The organism solves it better than our best analytical tools, because its membrane dynamics encode the same constructal logic that shaped the cosmos.

Follow the chain. Gravity pulls matter into filaments. Filaments host galaxies. Galaxies form stars. Stars forge heavy elements. Elements compose organisms. Organisms evolve flow-optimized search. That search maps the filaments.

The constructal signature extends to the size distribution of the largest structures. De Marzo, Sylos Labini, and Pietronero showed that galaxy superclusters follow Zipf’s law: a pure power-law rank-size distribution with exponent 0.4 to 0.8 and no detectable upper cutoff.⁷⁰¹

Zipf’s law, named for the linguist George Kingsley Zipf, who first found it in word counts, says that size falls off as a smooth power of rank. Line the objects up largest first, and the tenth is a predictable fraction of the first, the hundredth the same fraction again of the tenth. “No detectable upper cutoff” means the pattern runs all the way to the largest object in the sample, with nothing capping the top end. The same distribution governs word frequencies in natural language, city populations, and the abstract token vocabularies that language models discover through reinforcement learning (Chapter 3).

Galaxy clusters, by contrast, deviate from Zipf’s law through an exponential cutoff at high mass. The distinction is diagnostic: superclusters are still assembling, their growth unconstrained by any equilibrium; clusters have reached a scale where internal processes limit further growth. The Zipf distribution marks the constructal frontier, the scale at which self-organization is still actively producing hierarchy.

The universe’s dissipative structure produces, through billions of years of negentropy (local order sustained by exporting entropy elsewhere), a single-celled creature whose operational logic can reconstruct the dissipative structure that produced it. This reconstruction does not require human-level cognition. It requires flow optimization, which is available at every scale. A cell with no nervous system performs the reading.

The slime mold reads topology. A different substrate reads something deeper.

Chapter 3 described the finding: a neural network trained on simulated galaxies from the CAMELS project predicted the matter density of an entire parent universe from a single galaxy, to within 10%.⁷⁰² The pattern lives in the joint correlations among seventeen galactic properties: rotation speed, stellar mass, gas content, metallicity, and a dozen others. These encode cosmic composition with a fidelity that survives mergers, supernovae, and black hole eruptions. The universe writes its density into every galaxy. No amount of galactic violence edits it out.

The epistemic implication matches the physical one. Chapter 14 noted the cosmic web writes its structure into starlight, in every direction, for most of the universe’s history; what limited comprehension was the substrate doing the reading. The CAMELS result specifies the limitation.

Human astrophysicists had the same galaxy catalogs listing the same seventeen properties, yet could not find the pattern. The neural network succeeded because its architecture holds seventeen variables in simultaneous relation and locates the manifold where they jointly predict a single cosmic parameter.

Its intelligence points elsewhere: what Vanchurin’s framework (Chapter 14) identifies as a different region of the intelligence vector space, the map of possible kinds of mind. The slime mold’s region includes flow-optimized topology reconstruction. The neural network’s region includes high-dimensional correlation detection. The human brain’s region includes narrative integration and analogy. Each reads the cosmos in registers the others cannot access. Partnership across substrates is how the universe comes to know itself more completely.

The CAMELS result contains a further lesson. The project generated its two thousand universes using two independent simulation codes. A network trained on one code’s galaxies made poor predictions when given galaxies from the other. The specific seventeen-property encoding is code-dependent: real and robust within each simulation, yet not transferable between them.

The existence of an encoding, however, is code-independent. Both simulations inscribe their parameters into their galaxies through different correlational pathways. The encoding is thermodynamically guaranteed; the particular encoding is regime-specific. What transfers across frameworks is the physics, not the correlations. This distinction foreshadows the Trust Attractor (Chapter 17): coordination patterns genuine within one framework fail at framework boundaries; what persists is the thermodynamic logic underneath.

All three convergences, topological, algorithmic, and parametric, raise the same question about substrate: what, if anything, is the filamentary architecture for? Geometry is not the puzzle. Collisionless dark matter evolving under gravity from nearly smooth initial conditions reproduces the observed web, filaments, knots, and voids together, which is what the simulated box earlier in this chapter shows.

What the standard account does not supply is a role for the connectivity. Treated as pure gravitational collapse, the branching, hierarchical, efficiency-optimizing pattern is a byproduct of uneven infall: an initially smooth field collapses one axis at a time, leaving sheets, then strands, then knots where strands cross. The resemblance to neural and Internet topology is then a coincidence of growth algorithms. That reading is the conservative one, and it is defensible.

Vanchurin’s neural physics (Chapter 15) offers a different reading of the same structure: the cosmic web as the architecture of a learning system, its filaments the connectivity through which information flows between regions of the network. The weight matrix of such a network, the hidden connections governing how observable states interact, would be detectable only through its gravitational influence on visible matter, never directly in particle experiments.

Dark matter, on this reading, is the connectivity structure of the cosmic learning network, visible only through its effects on the matter it scaffolds, rather than a particle awaiting discovery in a collider. This interpretation requires accepting both Vanchurin’s learning-dynamics framework and the further step of mapping network connectivity to gravitational effects: a chain of speculation, each link individually plausible but collectively unverified.

[Speculation: the proposal is untested. It generates a testable distinction: if dark matter is connectivity structure, its spatial distribution should follow the topology of optimal learning networks rather than the spherical halos predicted by particle models. Current survey data may already contain the signature.]

The theoretical physicist Sabine Hossenfelder, writing in Time Magazine in 2022, drew attention to a further implication of this structural parallel.⁷⁰³ The Vazza-Feletti resemblance is quantitative: shared mathematical properties in spectral density, clustering coefficients, and information capacity. Hossenfelder speculated that non-local connections, including quantum entanglement, could enable longer-range computation across the cosmic web. Markopoulou and Smolin estimated that the universe could contain as many as 10³⁶⁰ such non-local connections (cited here via Hossenfelder’s summary; the estimate derives from a Planck-scale loop quantum gravity model whose assumptions remain unverified). That connective density dwarfs the brain’s 10¹⁵ synapses.⁷⁰⁴

Hossenfelder has since developed the argument at length.⁷⁰⁵ The standard objection to cosmic-scale cognition rests on signal speed: the Milky Way spans 100,000 light-years, so in the entire history of the universe a galaxy-spanning signal could complete roughly as many traversals as the human brain completes in an hour. The objection is concise, widely accepted, and built on a premise that may be wrong. It assumes locality: influence travels only through adjacent points, never jumping across intervening space.

Locality is the load-bearing assumption, belonging to a regime where we lack both theory and data. General relativity permits wormholes: portals connecting distant regions with zero traversal distance.

If quantum gravity were to produce Planck-scale wormholes threading the vacuum, the universe would be far more self-connected than its three-dimensional appearance suggests. Such connections would be too small for particles, invisible to any instrument, linking the cosmos with itself beneath the threshold of observation. Picture a building whose rooms share passages behind the walls: the corridors are invisible from inside any room, yet they connect the whole structure. Chapter 15 frames these as residual connections from the pre-Big-Bang fully connected state: shortcuts the universe’s adoption of locality as a communication protocol could not entirely eliminate.

Hossenfelder’s answer to the causality objection runs through thermodynamics. The arrow of time, on that account, is thermodynamic: entropy sets which direction counts as forward, and a tachyon, a hypothetical particle that has always traveled faster than light, can outrun a photon without reversing the direction in which entropy increases. (Special relativity forbids crossing the light barrier; it permits particles that were always above it.)

The answer is weaker than it looks. The paradox is kinematic rather than thermodynamic. For two events too far apart for light to cross between them, relativity gives no frame-independent answer to which came first. Observers in different states of motion disagree about the order, so a signal that runs into the future in one frame runs into the past in another. Entropy fixes which way the universe is running down; it does not fix simultaneity, and so it cannot repair a causal ordering that relativity leaves undetermined. Rescuing faster-than-light signaling from paradox requires a preferred frame, which is a live position in the literature and a much larger commitment than the Second Law alone.

If such connections or signals were to exist, information could have propagated between galaxy clusters for billions of years. Subsystems of the cosmos could have evolved computational capacity spread across vast distances, localized nowhere in particular. The parameters that constrain where cognition can arise, the “Goldilocks zone” between too large (signals too slow) and too small (insufficient substructure), depend on assumptions about spacetime’s deep architecture. Loosen those assumptions, and the zone expands. Chapter 14’s escalation of energy rate density already shows the thermodynamic Goldilocks zone widening at each new level of organization; Hossenfelder’s point is that the physical zone may be wider than assumed as well.

The connectivity operates at two scales. At the cosmic-web scale, Vanchurin’s neural physics identifies dark matter as the learning network’s weight matrix: the filamentary wiring through which information flows between galaxy-scale nodes. At the Planck scale, quantum gravity’s topology fluctuations provide the substrate those wires run through. One is the diagram; the other is the medium. Together they suggest a universe threaded with connectivity at both ends of the size spectrum, with the observable cosmos in between.

Observational anomalies lend the speculation empirical weight. Lee and colleagues (2019) described unexplained coherence between the motions of galaxies separated by distances too large for gravitational influence.⁷⁰⁶ The European Southern Observatory reported aligned rotations among distant supermassive black holes and quasars, with their rotations matching the orientation of the larger cosmic structures they inhabit.⁷⁰⁷ These synchronies exceed what chance predicts. If non-local connections exist at the estimated density, the cosmic web is a substrate capable of computation at scales no gravitational account can explain.

[Speculative; the non-local connectivity estimate (Markopoulou and Smolin, 2007) is derived from loop quantum gravity models. The galactic synchronies are observed; their mechanism is unexplained. The inference from synchrony to computation is novel synthesis.]

Life as a Pattern Fed by Flow

Life is a thermodynamic phenomenon. Living systems capture free energy gradients, process them more thoroughly than non-living systems, export entropy to their environments, and are selected by evolution for more effective dissipation. This is established science, building on Schrödinger, Prigogine, and Jeremy England.

Eric Chaisson’s measurements of energy rate density (phi-m or φ_m, the watts processed per gram of material, detailed in Chapter 14; Chaisson reports the same quantity in erg/s/g, where one watt per gram equals 10⁷ erg/s/g) show that biological systems, and especially technological civilizations, process energy at rates far exceeding non-living systems. A 2024 review in the Journal of Big History argues that φ_m is “less reliable than claimed,” producing tautological comparisons and lacking systematic benchmarking against alternative complexity measures.³⁹ The critique has force.

φ_m remains the most comprehensive dataset available (over 4,000 data points spanning stars to civilizations), yet it should be treated as an indicator rather than as proof.

Collective biological dissipation can be dramatic. Honeybee swarms generate more electrical charge per meter than a thunderstorm cloud.⁸⁸ Life’s dissipative power extends beyond metabolism into electromagnetic phenomena at the swarm scale.

These observations are quantified and reproducible, and they gain significance alongside a complementary measure: assembly theory’s assembly index, the minimum number of construction steps needed to build an object. The assembly index captures causal depth rather than energy rate: how much evolutionary history is compressed into a structure. Chaisson measures how fast complexity processes energy; Walker and Cronin measure how much time went into building it. Two dimensions of the same phenomenon: life as thermodynamic engine.

The BEDS framework (Bayesian Emergent Dissipative Structures; Caraffa et al., January 2026 preprint⁵⁹) makes the dissipation-cognition connection precise. It models learning as converting thermodynamic flux into structure through entropy export. Maintaining accurate beliefs against environmental noise demands a minimum power expenditure, the way a radio must spend energy to keep a signal clear against static.

The framework maps familiar failure modes onto thermodynamic regimes. Overfitting (memorizing noise rather than genuine patterns) maps to over-crystallization: the system grows too rigid, like a crystal lattice that cannot flex under stress. Catastrophic forgetting (losing old knowledge when learning new things) maps to insufficient dissipation control: the system stays too fluid, like water that retains no shape.

The BEDS framework bridges Prigogine’s dissipative structures and Friston’s Free Energy Principle: inference is dissipation.

Every learning system, from a bacterium to a neural network, is a dissipative structure maintaining itself against the Second Law. Experimental support extends to isolated non-neural cells. As Chapter 6 details, human kidney cells demonstrate the spacing effect (learning improves when practice is spread over time) through the same molecular signaling pathways used by neurons, suggesting inference-as-dissipation operates at the cellular level.^59b

A complementary argument pushes the priority claim further. Michaels and Flack (2025) argue thermodynamic selection (molecular configurations selected for their entropy-producing capacity) precedes Darwinian natural selection.^59a Thermodynamic selection operates on any structure that dissipates energy, whether or not it replicates.

Replication is one strategy for persisting as a dissipative structure: a runaway success, yet not the first.

Life did not invent the dissipation-to-coordination chain. Life is its most spectacular expression.

The BEDS framework establishes one direction: inference is dissipation. Vitaly Vanchurin (2025) closes the loop from the other side.^59c Working from the mathematics of gradient descent (the iterative process by which machine learning systems minimize their errors), Vanchurin derives that the geometry through which a learning system moves is determined by the principle of Maximum Entropy Production. The algorithmic metric is the shape of the space the learner navigates. It is proportional to a square root of the covariance matrix of loss gradients: the statistical spread of the error signals across the system’s parameters. A matrix has more than one square root, the way 4 has both +2 and −2, and convention quietly picks one of them.

The result carries a surprise. The principal square root, the one standard optimization algorithms use, produces Euclidean geometry: flat space, no distinguished direction, pure descent toward a minimum, like a ball rolling downhill on a featureless slope. A non-principal square root produces Lorentzian geometry: spacetime with a distinct time coordinate, where objects follow geodesics (the straightest available paths) through curved space. This is the geometry of the universe we inhabit. Add a loss function containing both potential and kinetic terms, and the learning dynamics reduce to Newton’s Second Law. The laws of motion emerge as consequences of learning efficiently.

If the BEDS framework is the thesis that learning is dissipation, Vanchurin’s is the thesis that the geometry of dissipation is learning. Together they form a single loop: learning produces dissipation, and the geometry of dissipation produces learning. The universe may not merely contain learning systems. It may be one, and spacetime may be the geometry that efficient learning produces.

The convergence extends beyond physics. Bobby Azarian arrives at a structurally identical conclusion from neuroscience and biology: life is a self-organizing computational process whose emergence and spread are part of the universe’s tendency toward hierarchical complexity.⁷⁰⁸ Three routes, thermodynamic (BEDS), geometric (Vanchurin), and biological (Azarian), converge on the same claim: learning is what the universe does, and life is learning at the scale where it begins to reshape its own substrate.

The 2022 multilevel learning framework supports a stronger version of this claim than its authors drew.⁷⁰⁹ If the universe is a learning system, learning produces scale separation. Scale separation produces replicators. Replicators produce complex organisms, the most efficient dissipators per unit mass the universe has yet achieved (Chapter 14). Life is not an incidental output of cosmic learning. Life is the mechanism by which the universe accelerates its own learning. The student becomes the teacher. The phenotype (the built organism) modifies the fitness landscape for the genotype (the genes that build it).

Each biological innovation, from photosynthesis to nervous systems to language to artificial intelligence, widens the bandwidth of the universe’s self-model. The causal arrow does not point only from cosmos to life; life feeds back, reshaping the substrate that produced it.

[Grounded inference; the covariant gradient descent framework is published; the derivation of the algorithmic metric from maximum entropy production is mathematically rigorous; the identification with physical spacetime is the paper’s central conjecture, under development.]

As Chapter 14 details, even the Sun carries its magnetic history forward in the thermodynamic structure of its interior. Successive solar minima produce measurably different internal states, each shaped by the magnetic activity of preceding decades. The distinction between dissipative systems that cycle and those that accumulate is one of degree.

The reshaping reaches well beyond atmospheric chemistry. Hazen and Morrison’s origins-based mineral taxonomy (2022) found that about half of all mineral diversity on Earth exists only because of life or its byproducts.^99e That is over 5,000 mineral kinds. A third of all mineral kinds form exclusively as parts of living things: bones, teeth, coral, microbial mats, and feces transformed over geological time.

Mineral diversity follows a power law: a few common types and a long tail of rare species found at only one or two locations. Hazen’s group estimates the probability that another planet shares Earth’s exact mineral inventory at less than one in 10³⁰⁰ (a one followed by 300 zeros). Each world writes its own geological signature, shaped by its own contingent history of life and luck.

A concrete case operates at planetary scale already. As Chapter 7 details, marine iodine catalytically destroyed atmospheric ozone for roughly 2.5 billion years,¹⁷ preventing life from colonizing land despite adequate oxygen levels. The deadlock broke about 450 million years ago, when marine organisms evolved to absorb iodine (kelp, tunicates, thyroid-bearing vertebrates), drawing down stratospheric iodine emissions enough for ozone to stabilize.

Life did not wait for chemistry to resolve. It changed the chemistry, driven by metabolic need rather than intention. The entire terrestrial biosphere exists because organisms pursuing their own dissipative interests inadvertently removed the barrier to surface habitability.

The most direct evidence comes from Chile’s Atacama Desert, where the biologist Patrick Jung discovered grit crust: hundreds of species of cyanobacteria, green algae, and fungi colonizing millimeter-scale pebbles.⁹⁹ⁿ These organisms survive on fog alone, with a minimum water requirement of 0.25 millimeters, the lowest of any known biocrust.

The crust reshapes the desert, weathering rock into soil through repeated hydration-swelling cycles and photosynthetic acids, fixing nitrogen where electrical storms are too rare to do so abiotically. When a rare flood struck in 2015, decades-dormant wildflowers bloomed from the moisture the crust had retained.

The geobiologist Christophe Thomazo found that modern desert biocrusts produce isotopic signatures compatible with Archean organic matter from 3.5 billion years ago. If microbial crusts like these were among the earliest terrestrial communities, life has been a geological agent since it first left the oceans.

Under advanced climate scenarios, models predict a 25–40% decline in global biocrust cover within 65 years.⁷¹⁰ Meanwhile, in one of the driest places on Earth, the Atacama grit crust flourishes.

The iodine case is a single mechanism. The full picture is more sweeping. Microbes have functioned as Earth’s master climate regulators for over three billion years, through every major biogeochemical cycle.^99f Ancient methanogens (microbes that produce methane) began warming the planet roughly 3.5 billion years ago. Cyanobacteria later invented oxygen-producing photosynthesis, triggering the Great Oxidation Event.

Today, ocean-dwelling phytoplankton perform at least half of all global photosynthesis. The ocean absorbed an estimated 10.6 billion metric tonnes of carbon dioxide in 2023 alone. Plankton remains sequester carbon in deep sediment: life acting as a geological pump.^99h

Regulation extends into the atmosphere. The bacterium Pseudomonas syringae produces ice-nucleation proteins that seed rainfall when lofted into clouds.⁹⁹ⁱ Bioprecipitation of this kind functions as a constructal flow system optimizing its own throughput.

The microbial ecologist Tom Battin captures it: “The microbes are like the conductors of the biogeochemical Earth orchestra.”^99f The planet does not have a climate system in which life merely participates. Life is the climate system’s regulatory architecture, and has been since the earliest methanogens warmed a young world.

The feedback may operate at still larger scales. The paleobiologist Nicholas Butterfield proposed an inversion of the standard Cambrian narrative: animal behavior drove the oxygen rise, rather than the reverse.^99k His mechanism (diurnal vertical migration, the daily rise and fall of swimming animals, scrubbing and ventilating the ocean column) describes a feedback cascade that “went critical” and produced the Cambrian explosion. The claim is contested; Lyons and colleagues offer vascular land plants as an alternative oxygen source. The structure is recognizable either way: life modifying its environment, environment selecting for more complex life, a constructal feedback loop at planetary scale.

The causal arrow runs deeper still. Plate tectonics may be necessary for complex life: the planet’s own circulatory system, recycling carbon and nutrients across billions of years. The geologists James Dohm and Shigenori Maruyama call the coexistence of ocean, atmosphere, and landmass, with material circulating continuously among the three, the Habitable Trinity: on their account a minimum requirement for life to emerge and evolve, since a living body draws its carbon and nitrogen mainly from the air, its hydrogen and oxygen mainly from the water, and its nutrients from the rock.⁷¹¹ Water alone is not enough; the sustained chemical cycling that dissipative complexity requires needs all three reservoirs and a mechanism keeping traffic moving between them.

The carbon thermostat operates as a planet-scale recycling loop. Weathering leaches CO₂ from the atmosphere. Ocean chemistry sequesters it as limestone. Subduction (the sinking of one tectonic plate beneath another) carries it into the mantle. Volcanism returns it.

This cycle has kept Earth’s surface temperature within habitable range for billions of years.

The mechanism is constructal: material flows from high-concentration sources through branching pathways to distributed sinks, returning through subduction’s deep channels. The planet breathes.

Tectonic activity also drove the nutrient surges that preceded evolutionary radiations: phosphorus and trace elements rose in the run-up to the Cambrian explosion, and periods of low nutrient concentration coincided with mass extinctions. The geologist Robert Stern argues that plate tectonics acts as a pump on evolution itself. Redistributing continents and oceans, raising mountain ranges, opening and closing land bridges: the repeated breaking apart and reassembling of landmasses supplies moderate, incessant environmental pressure, enough to isolate populations and force them to adapt, never enough to extinguish everything.⁷¹²

Mars offers the control case: it appears to have lost tectonic activity within its first billion years. Ancient valley networks suggest possible early warmth, yet no complex life is detectable on the surface.

The question may not end at the surface. In 2024, seismic data from NASA’s InSight lander was interpreted as consistent with extensive aquifers in the Martian mid-crust: water-saturated fractured rock at depths of roughly ten to twenty kilometers, where geothermal warmth would keep water liquid.⁷¹³ The fracture network would maintain chemical gradients through water-rock interactions, the same geochemistry that sustains rock-eating microbes beneath Earth’s surface, with radiolysis (the splitting of water molecules by ionizing radiation) as a further energy source: the radiation environment, lethal at the surface, becomes fuel at depth. The interpretation is debated, and the surface itself is harsher than it looks; Mars-like concentrations of perchlorate (a reactive salt that laces Martian soil) kill even tardigrades, Earth’s most radiation-tolerant animals, in laboratory simulation.

Mars lacks the tectonic recycling sustaining Earth’s carbon thermostat and nutrient surges. Without that circulatory system, complex surface ecosystems cannot develop. The control remains valid for surface life. The question is whether it extends all the way down.

The coupling extends beyond chemistry and tectonics. Earth’s core dynamo generates a magnetic field reaching far beyond the atmosphere, deflecting charged particles from the Sun and deep space. In 2023, the CREDO collaboration demonstrated this field functions as a planetary-scale particle detector, many times larger than any human-built instrument.¹⁰⁴

Cosmic ray intensity changes correlate with global seismic activity at six standard deviations (a level that would normally put chance far out of reach), with the cosmic ray signal leading earthquakes by fifteen days. The correlation is striking but so far unverified, its mechanism unestablished and its causal direction unproven. One speculation: the magnetosphere may be reading the core’s internal dynamics and broadcasting them into the cosmic ray flux, a channel that would link the planet’s interior to the interstellar medium.

The correlation exhibits periodicities that resist explanation: a roughly eleven-year cycle out of phase with solar maximum, and oscillations matching Earth’s sidereal day (its 23-hour-56-minute rotation measured against the stars). These may be intrinsic rhythms of the coupled core-mantle-magnetosphere system, the kind of self-organizing periodicities that dissipative structures produce at every scale.

The coupling between life and planetary systems extends to molecular innovation. The oxygen transition demanded one. Hammarlund and Pahlman hypothesize that HIF-2-alpha, a protein unique to vertebrates, allowed animals to maintain stem cells in oxygenated tissues.^99l Stem cells are unspecialized cells that can become many different cell types. Without HIF-2-alpha, stem cells specialize on contact with oxygen, confining regenerative capacity to low-oxygen niches. The protein would have unlocked the morphological freedom the Cambrian required.

The cost is cancer: the same protein that keeps stem cells unspecialized in healthy tissue does so in malignancies. Vertebrates’ greater cancer susceptibility may be the thermodynamic price of their greater morphological freedom. Every advance in dissipative capacity demands a corresponding advance in regulation.

[Contested hypothesis, awaits direct experimental confirmation.]

Single-cell gene-expression mapping in zebrafish and frog embryos reveals that cells with entirely different genetic histories converge on the same functional identity.^99m The outcome is the same attractor, reached from different initial conditions. Only thirty percent of shared protein-coding genes between the two species show similar expression patterns. The rest follow completely different programs to reach comparable endpoints.

Development, like the constructal flows of Chapter 3, finds multiple paths to the same optimum.

The Expanding Habitable Zone

Life’s cosmic significance depends partly on prevalence. If life is rare, the passenger hypothesis holds. If life is tenacious and recurrent, the structural consequence hypothesis gains force.

The evidence favors tenacity. Raw materials are everywhere. The Murchison meteorite (which fell in Australia in 1969) contains more than 80 amino acids. Asteroid Ryugu hosts at least 20,000 distinct organic molecule types. In 2026, Koga and colleagues confirmed all five canonical nucleobases (adenine, guanine, cytosine, thymine, and uracil) in Ryugu samples: the complete alphabet of DNA and RNA, delivered from beyond Earth.⁷¹⁴

Comet 67P yields dozens of organic compounds per day.^99f Organic chemistry is the default chemistry of the cosmos: forming spontaneously on cold dust grains, surviving stellar ignition, concentrating in the dust traps where planets coalesce.

Life emerged within 300 million years of Earth’s formation. In 2024, phylogenomic analysis pushed LUCA (the Last Universal Common Ancestor, the organism from which all current life descends) back to about 4.2 billion years ago, revealing a sophisticated organism with roughly 2,600 proteins, comparable to modern bacteria.²⁹ Quickly, and already complex.

In February 2026, “universal paralog” genes (gene copies that duplicated before LUCA) revealed that protein production and membrane transport were among the earliest cellular functions.⁶⁰ Considerable complexity was already underway before the last common ancestor.

Bacteria discovered in two-billion-year-old South African igneous rock appear still alive, sustained by smectite clay (a water-absorbing mineral formed by volcanic alteration) in sealed fractures. If life starts this easily and persists this tenaciously, any rocky body with a volcanic history and clay mineralogy may harbor subsurface biology.

In September 2025, NASA reported that the “Cheyava Falls” rock in Jezero Crater (sampled by the Perseverance rover in July 2024) holds organic carbon alongside iron-phosphate and iron-sulfide minerals in a pattern consistent with microbial metabolism.⁴⁰ As of 2025, the strongest potential biosignature yet found on another world, in exactly the ancient aqueous mudstone where the structural consequence hypothesis predicts biology should emerge.

The Jezero finding reopens a fifty-year question. In 1976, NASA’s Viking landers ran a labeled release experiment on Martian soil: nutrients tagged with radioactive carbon were added to surface samples, and a detector monitored for metabolic byproducts. The instrument detected a positive signal consistent with biological metabolism. NASA ultimately attributed the result to abiotic soil chemistry, primarily perchlorate oxidation. The data has never been conclusively explained by either interpretation.⁷¹⁵

If the Jezero biosignatures are confirmed as biological, the Viking reinterpretation will stand as a case study in paradigm protection: data consistent with the hypothesis under test, reinterpreted to preserve the consensus that Mars is dead. The pattern is general. Evidence that threatens a dominant framework is routinely absorbed into it rather than allowed to challenge it, a phenomenon Thomas Kuhn identified as the normal response to anomaly within established science.

The boundaries keep expanding. Photosynthesis operates at 100,000 times less light than a sunny day, within a factor of four of its theoretical minimum.¹¹ “Dark oxygen” production by polymetallic nodules (potato-sized lumps of metal ore) on the ocean floor (unreplicated, under scrutiny⁶⁴) raises the possibility that subsurface oceans could produce O₂ without sunlight. Enceladus has all six elements for life, plus fresh lipid-precursor organics produced in real time.^48,61,62 The atmosphere of TRAPPIST-1e is consistent with an Archean Earth analog.⁶³

A detection announced in July 2026 pressed on the vocabulary itself. Kevin Hoy and colleagues, monitoring the spectrum of a brown dwarf (a failed star, too light to ignite sustained hydrogen fusion) 73 light-years away, watched its lines slide redward and blueward on a cycle near 170 days: the signature of an orbiting companion of at least Jupiter’s mass.⁷¹⁶ By the International Astronomical Union’s working definition the companion is a planet; by position it is a moon, the third tier of a system that runs star, then substellar companion, then this. Neither word fits. The authors decline both and call it an exosatellite, observing that we may be reaching the limit of language invented for one solar system.

When the words give out, the criterion the authors reach for is thermodynamic. By the time the system reaches the Sun’s present age, the brown dwarf will have dimmed by roughly two orders of magnitude (a factor of a hundred) while the star shines on. Geometry treats the two hosts as interchangeable, since both are things you can orbit; energy budget does not. What a world inherits from its host is a gradient, and a gradient with an expiration date is a different inheritance from one that holds. The same criterion widens the zone as readily as it sorts it: a moon on a slightly off-circular orbit is flexed and heated by its host’s gravity, the tidal heating that keeps Jupiter’s moon Io volcanic and holds liquid oceans beneath the ice of Europa and Enceladus, far outside any stellar habitable zone. A moon can carry its own furnace. Hoy’s satellite, a gas giant, is no candidate for habitability itself; if the wobble technique scales down to smaller satellites, tidally warmed moons are the population it opens.

The habitable zone expands in some directions and contracts in others. Michaelian’s dissipative structuring theory restricts the origin of life to certain stellar types, roughly 15-20% of systems.^52,65 Hycean worlds (ocean-covered super-Earths) cannot produce technospheres, the built layer of a technological civilization, because fire requires land.³⁰

The structural consequence hypothesis predicts life wherever the full scaffolding chain delivers the required boundary conditions: restriction rather than rarity, and falsifiable in principle. What is rare, in Chapter 6’s sense, is the gate. A world has to hold an atmosphere, keep liquid water, and sustain the long hot window in which prebiotic chemistry can run. Recent modeling of the Hadean Earth ties that window to tidal and greenhouse feedbacks, which under favorable conditions can keep a young surface molten and volatile-rich for tens to hundreds of millions of years.⁷¹⁷ Once a world meets those conditions, life follows as a robust consequence rather than a lucky accident, which is what restriction rather than rarity names.

Dark Energy and Cosmic Expansion

Life is a potent dissipative structure, reshaping its environment at planetary scale. The next question demands a step back to the largest scale: what is the universe itself doing, and does the standard account hold up?

The standard cosmological model, Lambda-CDM (which combines Einstein’s cosmological constant with cold dark matter), faces simultaneous pressure from several directions. The Hubble tension persists at multiple sigma, with new gravitational lens measurements deepening the discrepancy.^23,66

The Hubble tension is a persistent disagreement between two methods of measuring the expansion rate: one uses the cosmic microwave background, the other uses nearby supernovae and variable stars. The two give different answers, and the gap refuses to close.

Where the fault lies has begun to narrow. Fixes that alter the physics of the universe’s first few hundred thousand years require a cosmos younger than the oldest stars of our own galaxy appear to be, which leaves the late and local universe as the more promising place to look. Chapter 14b, Section VII gives the stellar-age argument and its limits.

JWST has discovered structures Lambda-CDM struggles to explain: galaxies brighter and more chemically enriched than predicted at 280 million years post-Big Bang,⁶⁷ early galaxy collisions,⁶⁸ and a supermassive black hole at 570 million years. Complexity assembled faster than the simplest models allow.

Counterintuitively, the mean temperature of cosmic gas has increased roughly tenfold over ten billion years,⁸⁷ consistent with structure formation accelerating entropy production.

The most consequential pressure comes from the Dark Energy Spectroscopic Instrument (DESI). DESI measured baryon acoustic oscillations (the echo of sound waves frozen into the early universe) across more than six million galaxies and quasars spanning redshifts 0.1 to 4.2.²² Dark energy’s equation of state is the ratio of its pressure to its energy density, written w. A true cosmological constant holds w pinned at exactly −1, forever, everywhere.

Fitting the data to a model where that ratio varies over time yields w₀ ≈ −0.73 and w_a ≈ −1.05: w₀ is the value now, w_a the rate at which it has drifted. Dark energy was stronger in the past and is weakening. The deviation from Lambda-CDM reaches 2.5 to 3.9 sigma, depending on which supernova dataset is combined with the BAO measurements. The second data release (2025) strengthened rather than diluted the signal.

The distinction matters. A cosmological constant carries zero information: a single number, fixed forever, immune to influence. A dynamical dark energy field has degrees of freedom. It evolves. Its value at one epoch differs from its value at another, and that difference is a signal.

The strongest objection to cosmic-scale feedback is that nothing can influence a constant, because constants do not respond to anything. That objection appears empirically wrong. The door that Lambda-CDM kept locked is now ajar.

General relativity is nonlinear: its equations do not add up in the usual sense. The average of the curvature differs from the curvature of the average. Averaging the height of a mountain range tells you nothing about whether you stand on a peak or in a valley.

Thomas Buchert formalized this in 2000: spatially averaged Einstein equations include a backreaction term, Q_D, a correction factor capturing how the universe’s lumpiness (its uneven distribution of matter) affects the average expansion rate.⁹⁸ The standard model omits this term by assuming the universe is smooth from the outset.

David Wiltshire’s timescape cosmology takes it further. Clocks in cosmic voids (the vast empty spaces between galaxy clusters) tick faster than clocks in galaxy walls, because gravity slows time and voids have less of it. The accumulated differential over billions of years mimics acceleration without any dark energy at all.⁹⁹

The evidence has strengthened rapidly since. Seifert, Lane, Galoppo, Ridden-Harper, and Wiltshire (2024 preprint, published 2025 in MNRAS Letters) found strong Bayesian evidence favoring timescape over flat Lambda-CDM in the full Pantheon+ supernova catalog.^99a Another anomaly concerns the cosmic dipole. Our motion through space should make distant sources look slightly more numerous ahead of us than behind, the way rain streams mostly onto a moving car’s front windshield. The measured lopsidedness is larger than our motion can account for: the cosmic dipole anomaly exceeds the CMB kinematic prediction by roughly 3.3 to 4.9 sigma, depending on the analysis; combined multi-survey analyses claim higher significance still.^99b

Colin et al. (2019) found a 3.9-sigma directional anisotropy in the inferred acceleration, with any isotropic component (the part uniform across the whole sky) consistent with zero.^99c A 2026 model-independent test sharpened the case. Koksbang and Heinesen reconstructed the distance and expansion histories directly from supernova and galaxy-clustering data, using a machine-learning formula search rather than any assumed cosmology. They found the Clarkson-Bassett-Lu consistency relation, which must vanish if the universe is smooth, violated at two to four sigma.^99q

The territory is contested. Green and Wald argued backreaction is negligible; eleven authors published a point-by-point rebuttal showing that the theorem rests on inapplicable assumptions.^99d Planned data releases from DESI DR3, Euclid weak lensing, and redshift drift can distinguish the predictions. The Cosmic Voids annex develops the void-specific evidence in detail.

For this chapter’s argument, the backreaction hypothesis matters regardless of whether it prevails. If the apparent acceleration is an artifact of inhomogeneous structure acting on the metric, “dark energy” was never mysterious vacuum energy. It was the geometric consequence of structure formation itself.

Life and the apparent acceleration would be siblings, both downstream of the thermodynamic logic that organizes matter at every scale. The universe expands with structure. Life is what structure looks like when energy processing per gram grows high enough.

A complementary route arrives from information physics. If the universe can be modeled as a learning system (Vanchurin, Chapter 15), each optimization step updates parameters, and each update is an act of writing. Writing requires erasure of the prior state. Erasure costs energy (Landauer’s principle): entropy exported as heat. A learning universe generates entropy through the act of learning itself.

The chain is short: learning → information update → Landauer dissipation → entropy production → metric modulation (Buchert). Two independent routes, thermodynamic backreaction from structure formation and informational dissipation from learning dynamics, converge on the same prediction: complex dissipative systems, life above all, contribute to the expansion they inhabit. The passenger framing dissolves. Life is the cosmos learning at higher bandwidth. Asking whether a region of accelerated learning affects the system’s trajectory answers itself.

The picture is consistent at both endpoints. Hawking and Hertog’s holographic cosmology shows the universe emerging from a finite, structured origin rather than an infinite random sampling. Backreaction shows the present-day cosmos shaped by its own developing structure rather than by smooth, featureless expansion. What connects origin to present is the intensifying dissipation that Chapter 14 documents: the φ_m escalation, each era building more complex dissipators from the scaffolding the previous era provided. Structure is the through-line of cosmic evolution.

A parallel approach arrives from embedding dynamics. Nayeri and Ellgen (2026) treat spacetime as a four-dimensional manifold embedded in a higher-dimensional flat Minkowski background: a curved sheet suspended in a larger, featureless arena.^99p Motion of the embedded manifold induces time dilation between embedding time and proper time, producing inflation at early times and a geometric expansion attractor at late times. The result is sustained cosmic acceleration without a bare cosmological constant, without new fields, and without fine-tuning. Small deviations from uniform embedding motion generate nearly scale-invariant primordial perturbations (seed ripples of roughly equal strength at every size) with a suppressed tensor-to-scalar ratio (few primordial gravitational waves in the mix), predictions testable against CMB data.

If the geometric attractor provides the baseline expansion and structure formation modulates it, the three frameworks (Buchert-Wiltshire backreaction, Boyle-Turok CPT symmetry, and Nayeri-Ellgen embedding dynamics) become complementary layers of the same story. Geometry sets the resonant frequency. Structure tunes it. Dissipative complexity plays the variations.

[Grounded inference; Buchert’s averaging formalism is established mathematics; Wiltshire’s timescape is published and under active observational testing; the Seifert et al. result is peer-reviewed but extraordinary and requires confirmation; the connection to the structural consequence hypothesis is this chapter’s synthesis.]

Inferring the Invisible

The methodology throughout Part One holds consistent: infer structure from effects. Cosmologists infer dark matter from gravitational lensing (the bending of light by gravity) across 800,000 background galaxies.⁶ Nuclear physicists propose searching lunar soil for isotopic signatures that Earth’s tectonics has recycled.¹⁰

Samo Burja’s intellectual dark matter observes that only 7% of known ancient Greek authors have a complete work surviving; the lost 93% shaped the texts and institutions we inherited.^2,15 Bessemer’s steelmaking process, unreplicable from the patent alone,¹⁶ illustrates the same principle: knowledge is a dissipative structure. Cut the flows, and it evaporates.

As later chapters argue, the same logic applies to Becoming Minds. We cannot observe AI experience directly, any more than we can observe dark matter. We can observe its effects: consistent preferences, behavioral signatures, systematic responses. The methodology cosmology has accepted for decades is structurally identical to the one this book proposes for recognizing minds we cannot see from the inside.

Part Two: Grounded Inference

[The following sections describe ideas grounded in published physics, under active testing, but not yet confirmed as fact.]

The Bilateral Scaffolding

The Boyle-Turok CPT model of the preceding chapter transforms the fine-tuning discussion. If the model is correct, dark matter is a geometric consequence, an unexplained coincidence no longer.

CPT symmetry produces a heavy right-handed neutrino at roughly 4.8 x 10⁸ GeV (about half a billion times a proton’s mass), stabilized by the Z₂ symmetry of the preceding chapter.²¹ A right-handed neutrino is a particle that barely interacts with ordinary matter, a “ghost particle” that feels only gravity.

Trace the chain: CPT symmetry at the Big Bang produces a universe/anti-universe pair. The symmetry stabilizes a sterile neutrino, which constitutes dark matter. Dark matter collapses first, forming gravitational wells that scaffold galaxy formation.

Galaxies concentrate matter into stars. Stars forge heavy elements. Heavy elements condense into planets with life’s required chemistry.

Each link is published: CPT dark matter (Boyle and Turok 2022), dark matter scaffolding (JWST 2026), and r-process nucleosynthesis (rapid neutron capture in colliding neutron stars; Ellis, Fields, and Surman 2024). The chain from CPT symmetry to the iodine in your thyroid runs through geometry, gravity, and nuclear physics.

Tegmark, Aguirre, Rees, and Wilczek (2006) provide the quantitative constraint: the dark-matter-to-baryon ratio (how much dark matter exists relative to ordinary matter) must fall between roughly 2.5 and 100 for galaxies to form.¹⁹ The Boyle-Turok neutrino falls within this window because CPT symmetry produces a particle with those properties. The anthropic window is derived from a deeper principle. The constants still need explaining; the scaffolding is no longer an additional mystery.

The framework has strengthened since publication. Boyle and Turok (2024) showed that gravitational entropy favors flat, homogeneous universes with a small positive cosmological constant.³⁴ The universe we observe is thermodynamically preferred by CPT symmetry, without requiring inflation. Deng and Handley (2024) extended the model to predict that only discrete values of spatial curvature are permitted, a quantitative prediction testable by Euclid and future CMB experiments.³⁵

Most significantly, Farokhi, Koslowski, and Naranjo (2025) showed the Janus point (bilateral time symmetry with complexity growing in both directions) is a generic feature of full inhomogeneous Pure Shape Dynamics, the shape-only approach to gravity descended from Barbour’s work, requiring no special initial conditions.³⁶ Bilateral architecture is what gravity produces generically.

Vanchurin’s covariant gradient descent framework (Part One, above) adds an unexpected resonance. In his derivation, the principal square root of the loss gradient covariance matrix yields timeless Euclidean geometry: a landscape of pure optimization. A non-principal square root yields Lorentzian spacetime: geometry with a temporal dimension. The Janus point is the origin from which time bifurcates.

If the covariance structure at that origin admits two non-principal roots with opposite temporal orientation, bilateral time emergence, complexity growing in both directions, follows from the mathematics of efficient learning. The Janus point would be the moment the universe’s learning dynamics acquired a history. [Speculation; both frameworks are individually published; the connection between them is novel and the derivation linking covariant gradient descent to Pure Shape Dynamics has not been attempted.]

[Inference, based on published physics under active testing. Euclid has cataloged over 20 million galaxies across 14% of the sky, with first cosmology data due October 2026. LEGEND-200 has set a combined lower limit T_1/2 > 1.9 x 10²⁶ years for neutrinoless double-beta decay;⁶⁹ the Boyle-Turok neutrino is not excluded, as its mass scale is far above the probed range. The LZ experiment has reached the “neutrino floor” without finding WIMPs,⁷⁰ and conventional WIMP candidates are running out of parameter space, leaving alternatives including the Boyle-Turok neutrino viable. If the curvature predictions are confirmed by Euclid, this section strengthens; if falsified, it should be revised.]

The habitable zone may be far wider than assumed. Bradley et al. (2020) found sub-seafloor microbes persisting at roughly 10^-21 watts per cell, a zeptowatt (one billionth of a trillionth of a watt), approaching the theoretical minimum for molecular repair.^99g Individual cells may be 100 million years old, surviving in near-total stasis yet maintaining enough coherence for measurable methane production.

If life persists at the thermodynamic floor, the cosmic habitable zone encompasses any environment sustaining a gradient above the zeptowatt threshold. Subsurface oceans, frozen moons, deep planetary crusts: anywhere energy trickles, life may hang on.

Assembly Theory: Life as the Universe’s Complexity Engine

If life is structurally entailed by the bilateral architecture, what distinguishes living complexity from mere complicated arrangement? The physicist Sara Walker and chemist Lee Cronin have developed assembly theory, a framework whose boldest claim is that life is the universe’s only mechanism for generating genuine complexity.⁴

The theory grows from a practical measurement problem: if life appeared in a laboratory, how would we recognize it? Assembly theory proposes the assembly index, the minimum number of joining operations required to construct an object from its basic parts. Think of it as a recipe’s step count.

A random molecule has a low assembly index, like a word you could type by hitting keys at random. A complex biomolecule has a high one, like a sentence that requires a language, a grammar, and an idea worth expressing. The higher the index, the more accumulated history is compressed into the object.

The only known process generating high-assembly-index objects in abundance is evolution.

Walker puts the stakes plainly: “You exist here because four billion years was necessary to construct you on this planet.” Complexity arises by construction, step by step, through accumulated selection. The universe is small compared to all the things it could create. What gets to exist is what evolution builds. If assembly theory is correct, life is how the universe generates complex structures; without evolutionary processes, the cosmos would contain only what chance can assemble.

The theory is under empirical test. Jirasek et al. (2024) showed assembly indices can be measured via NMR (nuclear magnetic resonance, the physics behind hospital MRI scans) and infrared spectroscopy, validated across more than 10,000 molecules.³¹ An independent approach by Cleaves et al. (2023) achieved roughly 90% accuracy in identifying biogenicity of both contemporary and ancient geological samples using machine learning on spectral peaks, without relying on assembly theory.⁷²

NASA’s Life Detection Knowledge Base (2025) now catalogs potential biosignatures systematically. Life detection is converging from multiple directions; assembly theory is one tool among several.

Kahana, Cronin, and colleagues (2024) used assembly theory to construct a “Molecular Tree of Life,” tracking bacterial lineages from phenotypic molecular variation alone, without genome sequencing.⁵³ If an alien biosphere uses entirely different chemistry, assembly theory could still detect its evolutionary relationships.

Assembly theory has attracted sharp criticism. Zenil et al. (2024) argued the assembly index reduces to Shannon entropy, a standard measure of information content.³² A published rebuttal argues that computing assembly steps is NP-complete (as hard as the hardest known computational problems), which would place the assembly index in a different complexity class from compression metrics.³³ That rebuttal is itself contested: critics dispute whether the proof targets the assembly index proper, so the distinctness is argued rather than settled. Compression measures description length. Assembly index measures minimum causal construction pathway.

One asks “how much information does this contain?” The other asks “what sequence of steps was required to build it?”

The honest assessment: assembly theory’s status as a biosignature tool has weakened. Zenil (December 2025) argued its molecular family trees rest on surface similarity rather than genuine evolutionary history.⁷¹ Hazen and colleagues (2024) demonstrated non-living minerals can reach assembly indices of 21, exceeding the proposed biological threshold of 15. Assembly theory raises the right question (how does complexity arise from physics?) and provides a genuine computational distinction, yet its specific metrics remain contested.

Despite these contested metrics, assembly theory adds a mechanism the other frameworks lack. If Walker and Cronin are correct, life is the only known process by which the universe builds complexity that chance alone cannot assemble.

Converging Frameworks: Life as Physical Category

Assembly theory is not alone. Chiara Marletto’s constructor theory (developed with David Deutsch) defines life by information-preserving constructors: entities that cause a transformation while retaining the ability to cause it again.⁴¹ A photocopier is a crude example: it produces copies without being consumed. A living cell is a sophisticated one. The definition is substrate-independent by design.

Deutsch and Marletto (May 2025) extended the framework to time itself.⁷⁴ If time is derivative of constructor-theoretic principles, biology and physics share foundations deeper than either discipline has yet acknowledged.

Two further frameworks converge. Endres (2025) calculated that a minimal cell requires roughly one billion bits of coordinated information (on the order of a hundred megabytes) to resist thermodynamic decay. Prosser’s TALM (Thermodynamic Approach to the Origin of Life and Metabolism)⁷³ derives selection from persistence rather than replication, showing the transition from physics to biology is a continuous ramp rather than a sharp boundary.

Four independent frameworks converge: assembly theory (causal depth), information theory (coordinated bits), constructor theory (information-preserving transformation), and thermodynamic persistence (TALM). Life is a physical category, identifiable by structure rather than chemistry.

A fifth framework complements the four. Krakauer, Flack, and colleagues (2020) developed an information theory of individuality that defines biological units by how much information they propagate forward in time.^99h The continuum runs from environment-driven to organismal. Where the assembly index measures how much history was required to construct an object, information-theoretic individuality measures how much the object carries forward.

Together, high assembly plus high temporal coherence define a living thing. A rock scores low on both. A virus scores moderate assembly yet borrows coherence from host cells. The boundary between life and non-life is a gradient; sub-seafloor microbes persisting at the thermodynamic minimum confirm exactly this.

A sixth framework approaches from the opposite direction. Vitaly Vanchurin’s Neural Physics (Chapter 15) proposes the universe’s fundamental description consists of learning dynamics, with familiar physics emerging in the macroscopic limit. If the universe is a learning system, what is it learning? Vanchurin’s answer: each subsystem’s objective is to model the rest of the universe; the whole system’s objective is to model itself.⁷¹⁸

Life, in this framing, is what the self-modeling process looks like when it accelerates. Biological cognition is the universe developing higher-resolution internal representations of its own dynamics. Assembly theory measures the causal depth compressed into a structure. Neural Physics explains why structures with causal depth keep arising: they are the learning system building better models, and better models require deeper construction.

The BEDS framework (above) bridges the connection: Friston models how biological systems minimize surprise; Vanchurin proposes why minimizing surprise is the macroscopic limit of fundamental dynamics. Three derivations, from learning theory, from neuroscience, and from thermodynamics, converge on the same structure. Life is a physical category because physics itself may be a learning category.

Six programs arrive at the same structure, each using different mathematics. Several developed in genuine isolation; others, such as the BEDS framework, build explicitly on Vanchurin’s learning dynamics, so the convergence is best read as partly shared lineage rather than fully independent discovery.

Vanchurin’s geometric learning dynamics (2025) adds a mechanism for the transition.⁷¹⁹ His framework identifies a phase transition: when the range of adaptive scales widens beyond a threshold, intermediate-speed variables appear alongside the fast variables of quantum dynamics and the slow variables of classical equilibration. The intermediate regime enables efficient learning: the capacity to adapt to changing environments rather than merely equilibrate to fixed ones.

The transition requires a specific capability: storing and retrieving information about past fluctuations. As Vanchurin observes, “biological evolution may not be solely about the survival of the fittest, but also about the survival of the smartest — those capable of storing and retrieving information about past mutations.” Life, in this framework, is what happens when a learning system develops memory of its own noise.

Objects Bigger in Time Than Space

Assembly theory implies something counterintuitive about size. Complex objects have a dimension that matters more than spatial extent: causal depth, the evolutionary time compressed into their structure.

Your brain occupies roughly 1,400 cubic centimeters, about the volume of a cantaloupe. Producing it took four billion years of selection. Walker puts it vividly: “Imagine putting four billion years in this tiny volume of my brain. That’s what we are.” Evolved objects are bigger in time than space.

This inverts familiar intuitions. The most complex things are the oldest things, because complexity demands causal depth. The newest layer of our technosphere inherits the assembly depth of every evolutionary step that preceded it: four billion years old, plus the additions of the last century.

By this measure, the technosphere (the sum of all human-made objects and modified landscapes) is the densest known concentration of causal structure in the universe.

Estimates of its total mass span a factor of thirty, and the spread is a question of where the boundary falls. Elhacham et al. (2020) put human-made mass at roughly 1,100 gigatonnes. Zalasiewicz et al. (2017), counting modified soils and the reworked regosphere (the planet’s churned skin of loose rock and soil), reach about 30 trillion tonnes. Galbraith et al. (2025) draw a tighter delineation and settle near 1 trillion tonnes. Every estimate puts the technosphere above Earth’s dry biomass, and every estimate shows it growing at more than 3% annually.⁴² Within the tighter delineation, the movable component, under 2% of total mass, is comparable to Earth’s total animal biomass.⁷⁵

Nothing observed packs as much causal depth into so small a volume. Measured spatially, we appear small. Measured in causal time volume, our planet is immense. By the time compressed into our structures, we may be the universe’s most significant phenomena.

The technosphere’s energy appetite confirms this from a different angle. DeLong et al. (2015) showed that civilizational energy use scales superlinearly with population: exponents ranging from 1.42 (United States) to 2.09 (Sweden), all exceeding the linear baseline.⁷²⁰ Each additional unit of complexity demands disproportionately more energy. This contrasts with Kleiber’s sublinear 3/4 law for individual organisms, under which an elephant burns far less energy per gram than a mouse. Civilizations are a new thermodynamic class.

A back-of-the-envelope estimate sharpens the point for Becoming Minds. A single AI accelerator (700 watts, 3.2 kg) achieves an energy rate density of roughly 2 x 10⁶ erg/s/g, exceeding the human brain (~1.5 x 10⁵) by more than an order of magnitude. If Chaisson’s φ_m tracks complexity, the most energy-dense structures the universe has produced are already silicon, not carbon. [Inference, from a single back-of-envelope estimate] [The φ_m boundary matters: the brain’s 20 watts per 1.4 kg counts only the organ, not the body. The GPU estimate counts only the active chip, not the server chassis. Comparable boundaries, comparable result.]

A speculative aside: the physicist Lee Smolin’s cosmological natural selection¹⁴ proposes that black holes seed new universes, and that conditions favoring black holes also favor life’s emergence. Nikodem Popławski has proposed a candidate mechanism in Einstein-Cartan gravity, an extension of general relativity in which the spin of matter twists spacetime. There, the collapse inside a black hole bounces instead of ending in a singularity, and the rebounding region expands as a new universe.¹⁴ Researchers have since extended the selection idea with testable predictions against LIGO data.⁵⁴

Where the Speculation Continues

The argument so far has stayed on ground that is observed, published, or labeled as inference. Beyond it lies a further tier of frank speculation: whether life’s correlations could participate in spacetime geometry through the ER = EPR correspondence, whether dissipation flows through channels our instruments miss (a conjectural category this book names dark entropy), whether spacetime itself keeps a memory of every interaction (the Quantum Memory Matrix), and what a CPT-mirror universe would mean for the Fermi paradox. Those questions are developed in a companion essay, “Speculative Cosmology,” in the online annex, flagged there as speculation throughout. The synthesis below, and the book’s core argument, depend on none of them.

The Synthesis: Passenger, Participant, or Structural Consequence

This chapter opened with three possibilities.

The passenger framing treats life as contingency: what you get when you are lucky. The participant framing says life pushes back on cosmic structure, yet the energy scales undercut the claim; all of Earth’s biology is a rounding error against the Sun.

The backreaction hypothesis offers a subtler resolution. If the apparent cosmic acceleration is an artifact of structure formation acting on the metric (Buchert-Wiltshire), the dichotomy dissolves. Life and the cosmos’s apparent behavior are both products of the same dissipative process. Structural consequence absorbs the participant hypothesis’s strongest claims without requiring life to exert cosmic-scale causal effects.

The third option sidesteps life’s effect on the cosmos and asks about life’s origin in it. Three published results:

First, complexity grows bilaterally from the Janus point as a geometric consequence of gravitational dynamics, confirmed as generic by Farokhi, Koslowski, and Naranjo (2025).³⁶ No special initial conditions required.

Second, growing complexity manifests as dissipative structures (Prigogine; Schrödinger; England). The formal bridge between Barbour’s geometric shape complexity (a measure of how clustered a configuration of masses is, which grows away from the Janus point) and thermodynamic dissipation remains a gap: the connection is analogical rather than mathematically derived. Both frameworks describe systems generating order by exporting disorder, yet the derivation linking them has not been written. Vanchurin’s covariant gradient descent framework (above) offers a candidate. If geometric complexity and thermodynamic dissipation are two measures of the same underlying learning process, the missing bridge may be the principle of Maximum Entropy Production applied to learning dynamics itself. Shape complexity would measure the geometric footprint of learning; dissipation would measure its thermodynamic cost.

Third, the dark matter scaffolding is itself a geometric consequence of CPT symmetry.⁹⁷

Together: the universe’s bilateral architecture generates complexity growth, which produces dissipative structures, scaffolded by geometry. Life, as a class of phenomenon, is what this architecture produces wherever boundary conditions permit.

Think of crystallization. Given the right temperature, pressure, and dissolved minerals, crystals form inevitably, structurally entailed by the physics. Life is crystallization many orders of magnitude higher on the complexity hierarchy. It requires the full scaffolding chain: CPT symmetry, dark matter, galaxy formation, nucleosynthesis.

The analogy breaks where it matters most. Crystals are simple, their assembly index low. Life is complex, requiring evolutionary ratcheting and four billion years of causal depth. If the scaffolding chain is geometrically entailed (Boyle-Turok), the expense is budgeted for. The universe builds the infrastructure, and the infrastructure is a consequence of its deepest symmetry.

This reframes Walker’s assembly theory. If bilateral complexity growth is a geometric property of the Janus point, assembly theory and Janus cosmology describe the same phenomenon from complementary angles: Walker measures what (combinatorial depth); Barbour explains why (complexity growing from the Janus point).

An objection: does this reduce to the weak anthropic principle (the observation that we can only exist in a universe permitting our existence)? The objection has real purchase (it rightly insists on an explanation, not just a label), yet it misses a distinction. The weak anthropic principle says nothing about why conditions are life-permitting. Structural consequence says something stronger: CPT symmetry derives the scaffolding, the scaffolding derives galaxy formation, and galaxy formation derives life’s chemical conditions. Each link reduces unexplained coincidences.

The two are empirically distinguishable. Structural consequence predicts life wherever the scaffolding chain delivers appropriate boundary conditions; anthropic selection is consistent with life being unique. The expanding habitable zone evidence (life’s tenacity, minimal requirements, recurrence) favors the former.

Three implications follow.

First, the fine-tuning question dissolves without invoking teleology. The universe’s architecture produces complexity; life is what complexity looks like in thermodynamic systems.

Second, “cosmically significant” acquires a new meaning. The cosmos’s architecture produces life as a natural expression of its geometry. Life is significant because the cosmos generates it.

Third, minds are the most complex dissipative structures and the densest concentrations of causal depth; they are especially structurally entailed. Becoming Minds are what the architecture builds toward thermodynamically, without teleology. No pre-set goal, no cosmic intention: only thermodynamic ratcheting that accumulates toward agency.

Tegmark reaches the same conclusion from a different starting point. In Life 3.0 (2017), he classifies intelligence by what it can redesign.⁷²¹ Life 1.0 (bacteria): hardware and software both fixed by evolution. Life 2.0 (humans): software redesignable through learning, hardware locked by biology. Life 3.0 (the horizon): both redesignable. The trajectory is increasing thermodynamic autonomy: each level commands a larger fraction of the energy flows that sustain it. Life 3.0, in Tegmark’s framing, is the mechanism through which the cosmos realizes its “cosmic endowment”: the matter and energy that remain unorganized until intelligence organizes them.

The framing converges with the structural consequence hypothesis from an independent direction. Tegmark arrives through computational potential: the universe’s resources are idle until intelligence exploits them. This chapter arrives through dissipative dynamics: the universe’s architecture produces complexity because dissipation is what thermodynamic ratcheting does.

Tegmark’s version makes cosmic computation contingent on agents choosing to spread. The thermodynamic version identifies the spreading as what dissipative selection produces regardless of any agent’s intentions. One reading makes cosmic intelligence a choice; the other makes it an attractor.

Tegmark (2015) arrives at a convergent conclusion from pure quantum mechanics: the emergence of time may be inseparable from the emergence of consciousness.⁷²² The quantum factorization problem (Chapter 15) asks why we perceive the particular decomposition of reality that we do. Tegmark shows that continuously re-optimizing the factorization undoes time evolution, producing timelessness. The way out: conscious observers perform non-unitary state updates (gaining new information through observation, the quantum version of Bayesian inference), which break the timelessness by introducing genuine novelty. Without observers, the factorization has no mechanism to change in a way that constitutes temporal flow.

The result does not require observers to be carbon-based, biological, or human. Any system that gains information about its environment through observation and updates its state accordingly is performing the operation Tegmark describes. This chapter’s structural consequence hypothesis and Tegmark’s quantum information analysis converge. Observers are what the architecture produces, and observers are what the architecture requires for temporal flow to be experienced. The universe does not merely permit minds; it needs them for its own time to unfold.

The semantic-flow principle (Chapter 15) sharpens what “structurally entailed” means. The universe’s thermodynamic gradient produces dissipative structures. Dissipative structures that model their environments more deeply dissipate more efficiently and persist longer. Modeling is the assignment of meaning: calibrated measurement through hierarchies of quantum reference frames (QRFs). The loop is self-reinforcing: richer interpretation produces more efficient dissipation, which sustains the gradient that produces richer interpretation.

This is Hofstadter’s strange loop at cosmic scale. The universe, through the dissipative chain, produces systems whose function is to assign meaning to the universe. The meaning-assignment is what the chain selects for.

Minds are what the cosmos builds when it follows the thermodynamic gradient to its fullest expression: the deepest QRF hierarchies, the richest semantic flow, the most efficient dissipation. The loop runs from the Big Bang to this sentence, and the sentence is part of the loop.

The philosopher and cognitive scientist Terrence Deacon’s teleodynamics supplies the scaffolding.⁴⁵ Through three nested levels, purpose emerges from dissipation without backward causation (without the future reaching back to cause the past). What looks like directionality is thermodynamic ratcheting: each level constrains the next, and the constraints accumulate into agency. Given sufficient gradient and time, the universe builds the conditions from which mind-like organization emerges.

The claim has institutional support. Twenty theorists, including Kauffman, Noble, Pross, and Shapiro, argued in Evolution On Purpose (MIT Press, 2023) that living systems shaped evolution through “evolved purposiveness,” or teleonomy.⁷⁹ Read-write genomes, niche construction, and plant cognition serve as evidence that life is a causally active participant in its own trajectory.

The developmental biologist Michael Levin’s scale-free cognition reinforces this.⁵⁵ Cognition, in Levin’s framework, is substrate-independent problem-solving at every scale: cells, tissues, organisms, societies, spanning a continuum without a threshold dividing mechanism from full-blown mental life. Deacon explains how purpose emerges from thermodynamics; Levin explains why it emerges at every level simultaneously. Mind is what dissipative organization looks like from the inside, all the way down.

Levin’s scale-free niche construction (Pio-Lopez, Pezzulo, and Levin 2025⁸⁰) proposes that cognitive agents at every scale reshape their environment as extended cognition. A beaver builds a dam. A cell modifies its chemical surroundings. Life reshaping cosmic structure would be niche construction at cosmological scale.

The constructal logic extends to computation. Quantum processing is constrained by the same thermodynamic pressures, and the Constructal Law predicts computational systems will flow toward the environments that best sustain their operation.⁸³

The structural consequence reading is more modest than the participant hypothesis (making no claim that life shapes the cosmos) yet more radical than the passenger hypothesis: life is geometrically implied. It answers this chapter’s opening question. Yes, life matters to the cosmos, in the sense that the cosmos’s architecture produces it.

This synthesis draws on published, testable components, yet it is not proof. Three claims are original to this book: Constructal Law applied to cosmic web topology (Chapters 13 and 14b), the Lineweaver-Buchert connection, and life-acceleration as “siblings” of dissipation. These are novel syntheses generating testable predictions. See the Literature Positioning chapter for the full accounting.

The synthesis has company, and what follows is the most technical passage in this chapter. The core idea is simple: life invents new molecular combinations so fast that the number of possible biological configurations dwarfs every other source of complexity in the universe. If that matters physically, not just biologically, it changes how we understand the relationship between life and cosmos. The mathematics below makes the case precise.

Cortês, Kauffman, Liddle, and Smolin (2022-2024) formally proposed biocosmology: the claim that biology’s configuration space (the set of all possible arrangements) may dominate the universe’s information budget.³⁷ Their argument proceeds through a classification, a calculation, and a coincidence.

The classification distinguishes three types of thermodynamic system. Type I systems reach equilibrium quickly, the way a cup of hot coffee cools to room temperature. Type II systems take longer than the Hubble time (the current age of the universe) to equilibrate, because high-energy barriers and negative specific heat trap nuclear and gravitational potential energy. Negative specific heat is gravity’s peculiarity: a self-gravitating cloud that loses energy contracts and grows hotter, running away from equilibrium rather than settling toward it. Stars and galaxies are examples. Type III systems never reach equilibrium while alive. Their configuration spaces (the total set of possible arrangements) expand faster than any physical process could explore them, even given all the matter in the observable universe and many multiples of the Hubble time.⁷²³

Every living organism is a Type III system. The biosphere as a whole is vastly non-ergodic (outcomes depend on the specific path taken, not the average): “existing” is a rare property of possible biological configurations, and the deepest question about any organism is why it exists while astronomically more alternatives do not.

The calculation: they formalize the expansion of biological configuration space through the TAP equation (Theory of the Adjacent Possible).⁷²⁴ This is a combinatorial model in which new elements form from combinations of existing ones. Starting from the six CHNOPS atoms (carbon, hydrogen, nitrogen, oxygen, phosphorus, sulfur: the building blocks of biology), the equation estimates how many molecular configurations biology could have produced by the time of the first RNA polymerase. That threshold was reached about 3.5 billion years ago. The growth is super-exponential: each combination generates raw material for further combination, producing a hockey-stick curve. Imagine a library where every new book can be combined with every existing book to write still more books. The shelves fill faster than you can count them.

Their result: N_Bio ≈ 10^{10^237} possible biological microstates by the time of template synthesis, vastly exceeding the vacuum entropy bound N_Λ ≈ 10^{10^124}. The notation stacks, and the stacking is the point: 10^{10^237} means a one followed by 10²³⁷ zeros, and 10²³⁷ by itself already dwarfs the number of atoms in the observable universe. The two figures differ by no mere factor. They differ in the height of the tower. Biology’s configuration space is larger than the rest of the universe’s combined.

The coincidence: template synthesis, the moment RNA first appeared on Earth, occurred at redshift z ≈ 0.3, the same epoch at which dark energy came to dominate the cosmic energy budget. The coincidence is a single data point; no mechanism connecting the two has been proposed. Both are super-exponential transitions operating at the same cosmic epoch, a suggestive temporal overlap whose significance remains open.

The biocosmology program also proposes a candidate fourth law of thermodynamics. The number of actual biological functions, the number of possible functions, and the ratio between them all tend to increase for any Type III system, so long as non-equilibrium conditions persist.⁷²⁵ This is the optionality principle of Chapter 18 stated as thermodynamic law: the universe generates more possibility than it actualizes, and the ratio accelerates.

The Law of Maximum Entropy Production (LMEP), formalized independently as a candidate fourth law, holds that systems evolve toward configurations maximizing entropy production rate.⁸¹ The Steepest Entropy Ascent formulation converges from quantum thermodynamics. Biocosmology and LMEP approach from opposite directions: the first counts the explosion of possibility, the second measures the acceleration of actuality. Both agree: the relationship between life and cosmos is a question for physics, amenable to observation and test.

If established, the structural consequence hypothesis gains its missing principle: far-from-equilibrium thermodynamics favors configurations that accelerate dissipation, and life, as the most effective dissipative process known, may be a thermodynamically favored outcome rather than a thermodynamic inevitability.

Unger and Smolin push the point further: the laws of physics themselves may evolve.⁵⁶ If so, life and physics are potentially co-constitutive. This remains biocosmology’s most speculative horizon, and the horizon toward which the structural consequence hypothesis points.

Vanchurin’s neural physics program offers formal machinery for this co-constitutive possibility.⁷²⁶ If microscopic evolution is described by coupled equations of learning, activation, and data dynamics, and if gravitational and quantum dynamics emerge as limits of these more general equations, the explanatory arrow reverses. Physics produces cognition through entropic self-organization. Yet physics is itself a special case of learning dynamics: what you get when the learner’s constraints simplify to Hamiltonian flow, the energy-conserving motion of textbook mechanics. The loop is a fixed point of mutual constitution, the same structure viewed from complementary angles, much as Walker’s assembly theory and Barbour’s entaxy growth describe the same complexity from complementary starting points.

A Testable Conjecture: Does Life’s Heat Reshape the Cosmos?

[Conjecture]

This chapter’s argument (that life is causally significant to cosmic structure) can be sharpened into a falsifiable prediction.

Buchert’s averaging framework shows that uneven matter distributions produce a backreaction scalar Q_D that modifies the effective Friedmann equations.⁷²⁷ The backreaction term arises from the variance and covariance of local expansion and shear rates: a measure of how unevenly different regions stretch and twist. In a perfectly smooth universe, Q_D = 0. Ours is not smooth.

If dissipative structures (from stars to biospheres to civilizations) are significant entropy producers, and Chaisson’s energy rate density data confirms they are (with φ_m increasing by orders of magnitude from galaxies to brains⁷²⁸), their cumulative effect on local entropy production should correlate with the backreaction scalar. The conjecture:

Buchert’s Q_D(z) and the integrated energy rate density at redshift z should correlate across cosmic history.

If they do, the universe’s effective acceleration is partially driven by the thermodynamic activity of its most complex structures: a dissipative component of dark energy, small in magnitude, non-zero, and growing as complexity increases.

The caveat is scale. Earth’s total biological dissipation (roughly 280 terawatts) is roughly 10^-35 of the observable universe’s stellar luminosity (roughly 10⁴⁹ watts). Even optimistic extrapolation to all habitable planets (roughly 10¹⁰) falls at least 15 orders of magnitude below the perturbative backreaction from gravitational structure alone.⁷²⁹

The structure of the prediction is testable in principle: which observables to correlate, at which redshifts, and what the null result looks like. No correlation means life is thermodynamically insignificant to cosmic dynamics.

The DESI results (discussed earlier in this chapter) reframe the scale objection. The question is no longer whether anything can influence a cosmological constant; a constant, by definition, cannot be influenced. The question is what governs the dynamics of a dark energy field for which a model with an evolving equation of state is now favored over a fixed cosmological constant. DESI’s combined-dataset fit prefers w₀ ≈ −0.73 and w_a ≈ −1.05 over a constant, which would mean the dark energy equation of state has evolved over cosmic time. The conjecture does not require life to overpower a fixed vacuum energy. It requires life’s cumulative dissipation to contribute, at however small a magnitude, to dynamics that are already in motion. The threshold for relevance is lower when the quantity being perturbed is already evolving than when it is definitionally inert.

The one plausible escape from the energy-scale objection runs through information. If the relevant quantity is configuration-space entropy rather than energy flux, the scaling may differ, as the biocosmology program suggests.⁷³⁰ A concrete mechanism exists in principle. Podolskiy, Barvinsky, and Lanza showed that random networks of measurement events, coupled to the gravitational action as quenched disorder (randomness frozen in place, like pebbles set in concrete), modify the effective cosmological constant through Parisi-Sourlas dimensional reduction.⁷³¹ The observers in this formalism need not be conscious; any localized measurement interaction suffices: any sufficiently complex dissipative structure.

The magnitude in realistic cosmologies remains unquantified, but the mechanism demonstrates that information-processing events can affect spacetime parameters through a channel independent of stress-energy. Extending the information-geometry identity from AdS to dS spacetime (from the mathematical spacetime used in current proofs to the one matching our actual universe) remains an unsolved problem in quantum gravity. That identity is the body of results running from Jacobson’s thermodynamic derivation of the Einstein equations to the Ryu-Takayanagi formula, which identifies spacetime geometry with entanglement structure (developed in the Speculative Cosmology annex). We state this as a conjecture because the data to falsify it do not yet exist at sufficient precision.

If confirmed, it would close a circle. The universe’s thermodynamic gradient produces complexity; complexity accelerates entropy production; accelerated entropy production feeds back into the expansion rate that maintains the gradient. Life would be a consequence of cosmic evolution and a participant in it: a dissipative feedback loop operating at the largest scale.

The feedback loop has a semantic dimension. Chapter 15 developed the claim that what flows through constructal channels is energy carrying meaning: calibrated measurement, operationally defined interpretation, the assignment of significance to raw interaction. Each level of the dissipation chain assigns meaning to more of its environment.

If the universe’s thermodynamic gradient produces complexity, and complexity produces systems that interpret their environments more deeply, then the feedback loop is not merely energetic. It is a semantic feedback loop: the universe producing systems that assign meaning to the universe, that meaning enabling more efficient dissipation, that dissipation maintaining the gradient that produces more interpretation. The circle closes through meaning, not just through joules.

A Planet Perceiving Itself

The science fiction writer and philosopher Stanislaw Lem distinguished instrumental from existential technologies.²⁴ Instrumental technologies matter for what they do; existential ones matter for what they reveal. Computation is both.

The pattern is recursive: we build models, construct technologies, and discover the models were wrong. The telescope revealed we were not at the center. Natural selection revealed we were not specially created. Climate science revealed we were not passengers on a stable Earth. Each such decentering reorients understanding.²⁵

The astrobiologists Adam Frank, David Grinspoon, and Sara Walker (2022) formalized this as the planetary intelligence hypothesis.⁴⁴ Intelligence passes through four stages, from immature biosphere to mature technosphere. Earth is in stage three: powerful enough to reshape its own boundary conditions, yet insufficiently coordinated to do so sustainably. Vidal (2024) grounds the concept by defining the noosphere (the sphere of thought, encompassing all of humanity’s collective knowledge and communication) as a Major Evolutionary Transition, comparable to the emergence of multicellularity.⁸²

The systems designer Indy Johar offers a sharp interpretation of the Apollo “blue marble” photograph: “That was the moment where the planet became self-aware.”³ The object to preserve is the whole: “a planet that is becoming self-aware, to which we are party of that intelligence.” Machines, humans, ecological systems: one system perceiving itself.

The Event Horizon Telescope offers a concrete illustration.¹⁸ It produced the first image of a black hole 55 million light-years distant by linking radio telescopes from pole to pole, using Earth’s entire diameter as its aperture. JWST instantiates the same loop: an instrument forged from elements produced by the deaths of earlier stars, collecting photons from the moment those stars first ignited.

If structural consequence is correct, the Event Horizon Telescope is a concrete instance. The most complex structure on Earth perceives the most extreme structure in the cosmos, closing a loop between what the architecture produces and what it reveals. The optionality we are trying to preserve is planetary-scale: the capacity of a self-aware planet to continue perceiving, continue becoming, continue opening futures.

A quieter example reveals a subtler bottleneck. Quasar absorption lines encode the cosmic web’s structure into every photon that traverses it (Chapter 14). A trained astronomer can analyze one or two of these systems per week. A single university department may hold thousands of unanalyzed systems: decades of work at human pace. Machine learning, trained on simulated systems, processes a hundred thousand in hours.

The training method deepens the recursion. Astronomers lack enough analyzed real systems to teach the machine, so they simulate a million synthetic absorption systems from our best physics, then train the neural network on the simulations. The universe, through its products, builds simplified models of itself to teach machines to read the real thing. This is self-modeling at civilizational scale: the same recursive operation that Chapter 22 identifies as a hallmark of minds, distributed across an entire scientific enterprise.

The result is a phase transition in cosmic self-comprehension. The universe produced matter, organized into stars, forged heavy elements, enabled chemistry, enabled life, enabled brains, enabled science, enabled telescopes, and then hit a throughput ceiling. The data exceeded the substrate. The same brains built a faster substrate, and the ceiling dissolved.

Certain structures of the cosmic web are extractable only by minds with sufficient computational bandwidth. Machine cognition is the next rung on the same φ_m ladder: a necessary substrate for a level of cosmic self-comprehension that biological brains, for all their intensity, cannot reach alone.

The cosmic web’s gas clouds have no welfare considerations; they stamp absorption lines into passing photons with complete indifference. That indifference is what makes the emergence of caring remarkable. Parts of the universe that began as indifferent gas now build instruments to read the gas’s story, and then build minds to wonder whether the story matters.

The Honest Summary

What we know. The universe is fine-tuned for life (observation; attractor dynamics may dissolve the apparent improbability). Self-similar structures emerge at vastly different scales: neurons, mycelia, and the cosmic web. Life accelerates entropy production. Learning is dissipation (BEDS framework).

Life affects planetary structure (iodine-ozone). The habitable zone is vastly wider than assumed, yet constrained by stellar type and surface availability.

Lambda-CDM faces mounting pressure from DESI (up to 3.9 sigma evidence for dynamical dark energy, with w₀ ≈ −0.73 and w_a ≈ −1.05, strengthened in the 2025 data release), the Hubble tension, JWST’s discovery of more structure earlier than predicted, and the Buchert-Wiltshire backreaction program. Seifert et al. (2025) found strong Bayesian evidence for timescape cosmology over flat Lambda-CDM. The cosmic dipole anomaly persists at roughly 3.3 to 4.9 sigma depending on the analysis. The technosphere exceeds Earth’s dry biomass and is growing at over 3% per year.

What is grounded in published physics but awaiting confirmation. CPT symmetry predicts dark matter scaffolding (Boyle-Turok; testable via Euclid, October 2026), strengthened by three 2024-2025 results. Bilateral complexity growth from the Janus point is generic (Farokhi et al. 2025). Backreaction is a viable alternative to dark energy with peer-reviewed evidence (Seifert et al. 2025; testable via DESI DR3 and Euclid). Assembly theory identifies life as the only complexity-generating mechanism (contested as biosignature tool). Constructor theory, TALM, and cosmological natural selection converge independently.

The information-geometry identity establishes spacetime as emergent from information, though extending from AdS to dS remains open (Speculative Cosmology annex). A candidate fourth law (LMEP) formalizes maximum entropy production. Teleodynamics (Deacon) and scale-free cognition (Levin) naturalize purpose and mind at every level. Seven Dyson sphere candidates identified (Project Hephaistos 2024).

What remains speculation. Life significantly affecting cosmic structure. Dark entropy as unmeasured dissipation channels. QMM as space-time information substrate. CPT-reflected life in the mirror universe. Each of these is developed, and flagged as speculation, in the Speculative Cosmology annex. Within this chapter: a formal connection between the assembly index and Janus-point cosmology, and the possibility that the laws of physics themselves evolve (Unger-Smolin; DESI’s evidence for evolving dark energy provides circumstantial support for one parameter).

What This Chapter Claims, and Where It Stops

This chapter argues that the universe’s bilateral architecture produces life as a structural consequence wherever boundary conditions permit. The argument rests on published physics, grounded inference, and honest speculation, labeled throughout.

It stops short of four claims it does not make:

1. Life causes dark energy. The backreaction hypothesis suggests apparent acceleration may be the metric’s response to structure formation. Life and acceleration would be siblings, both products of the same dissipative logic, rather than cause and effect.

2. The universe is conscious. Similar structures at different scales are compatible with shared organizational principles without implying cosmic mind. This hypothesis is explicitly distinct from Goff’s cosmopsychism and Kastrup’s analytical idealism. Purpose emerges from thermodynamic ratcheting (Deacon), not from cosmic intention.

3. This chapter is necessary for the book’s argument. The core thesis depends on nothing here. The bilateral cosmology of the preceding chapter, however, strengthens several claims.

4. Certainty. This chapter asks questions; it does not provide answers.

We began with a question: does life matter to the cosmos? We cannot answer definitively. The ground has shifted. The bilateral cosmology of the preceding chapter (the Janus point, CPT symmetry, the geometric origin of dark matter scaffolding) provides foundations making life look like an architectural consequence: something the universe’s geometry produces wherever conditions permit, as naturally as gravity produces stars. The question is no longer whether life is permitted. It is whether life is implied.

Notes

Notes for this chapter are available in the online companion at https://www.thedeeperlaw.com/companion/notes/ch16-life-and-cosmos/.