Derivations and bounds supporting the essay’s quantitative claims. Inline citations map to the reference section below. Sections A–D ground the established physics; E–F formalise the selection and typicality arguments; G is explicitly heuristic; H records what would falsify the picture.

A. The Inevitability of Singularities

The claim that “the geometry leaves no other option” is the content of the Penrose–Hawking singularity theorems.^[4][12] The mechanism is the focusing of light, governed by the Raychaudhuri equation. For a congruence of null geodesics with tangent $k^a$, expansion $\theta$, shear $\sigma_{ab}$, and vanishing rotation:

$\frac{d\theta}{d\lambda} = -\frac{1}{2}\theta^2 - \sigma_{ab}\sigma^{ab} - R_{ab}k^a k^b.$

The null energy condition, $R_{ab}k^a k^b \geq 0$ — equivalent through Einstein’s equations to $T_{ab}k^a k^b \geq 0$ — makes every term on the right non-positive. So

$\frac{d\theta}{d\lambda} \leq -\frac{1}{2}\theta^2 \quad\Longrightarrow\quad \frac{1}{\theta(\lambda)} \geq \frac{1}{\theta_0} + \frac{\lambda}{2}.$

If the congruence is anywhere converging, $\theta_0 < 0$ — the defining property of a trapped surface — then $\theta \to -\infty$ within affine length $\lambda \leq 2/|\theta_0|$. That divergence is a caustic: neighbouring light rays cross. Penrose (1965) globalised this: given a non-compact Cauchy surface, the null energy condition, and one closed trapped surface, the spacetime is null-geodesically incomplete.^[4] Incompleteness is the invariant definition of a singularity. Oppenheimer and Snyder had exhibited the explicit collapse in 1939;^[3] Penrose proved it was generic. Collapse past the trapped-surface threshold is forced, not contingent.

B. The Speed Limit of Computation

A logical operation is a transition between distinguishable states. The Margolus–Levitin theorem bounds its rate: a system with mean energy $E$ above its ground state passes between orthogonal states no faster than^[22]

$\nu \leq \frac{2E}{\pi\hbar}.$

Devoting all of a rest mass $m$ to the task, $E = mc^2$, gives the mass-normalised ceiling

$\boxed{\ \nu_{\max} = \frac{2mc^2}{\pi\hbar} \approx 5.4\times 10^{50}\ \text{ops}\,\text{s}^{-1}\,\text{kg}^{-1}.\ }$

Bremermann’s 1962 figure, $c^2/h \approx 1.36\times10^{50}$ bit-ops s⁻¹ kg⁻¹, is the same statement up to the order-unity convention relating bit-flips to orthogonal-state transitions.^[5] Humanity’s installed capacity is of order $10^{21}$ ops s⁻¹, so the gap to one kilogram at the ceiling is

$\frac{\nu_{\max}(1\,\text{kg})}{10^{21}} \approx \frac{5\times10^{50}}{10^{21}} \approx 10^{30}$

— the thirty orders of magnitude in the body. The bound is on rate per unit mass, independent of architecture, algorithm, and temperature.

C. The Memory Limit and the Black-Hole Computer

Speed is bounded by energy; memory is bounded by energy and size together. The Bekenstein bound caps the entropy of a region of radius $R$ enclosing energy $E$:^[6]

$S \leq \frac{2\pi k R E}{\hbar c}, \qquad I = \frac{S}{k\ln 2}\ \text{bits.}$

A black hole saturates it. Substituting the Schwarzschild radius $R_s = 2GM/c^2$ and $E = Mc^2$ recovers the Bekenstein–Hawking entropy,^[6][23]

$S_{\text{BH}} = \frac{k c^3 A}{4 G\hbar} = \frac{4\pi G k M^2}{\hbar c}, \qquad A = 4\pi R_s^2,$

so a one-kilogram black hole stores

$I = \frac{4\pi G M^2}{\hbar c\,\ln 2} \approx 3.8\times 10^{16}\ \text{bits.}$

Combined with the rate of §B, this is Lloyd’s ultimate computer:^[7] ~$10^{51}$ operations per second over ~$10^{16}$ bits.

The persistence caveat. A one-kilogram black hole is not a device. Its Hawking temperature and lifetime are^[24]

$T_H = \frac{\hbar c^3}{8\pi G M k} \approx 1.2\times10^{23}\ \text{K}, \qquad t_{\text{ev}} = \frac{5120\,\pi G^2 M^3}{\hbar c^4} \approx 8\times10^{-17}\ \text{s.}$

It evaporates almost instantly. The small black hole is the ultimate rate, not a standing machine. Persistence scales as $M^3$: a $10^9\,M_\odot$ hole has $t_{\text{ev}} \sim 10^{94}$ years, eighty-four orders of magnitude past the present age of the universe. The computational frontier and the deep-time substrate are one object at two masses — small holes saturate the rate, large holes supply the persistence and the energy of §D.

D. Energy Extraction from Rotating Black Holes

A Kerr black hole carries angular momentum $J$, written through the dimensionless spin $a_* = cJ/GM^2 \in [0,1]$. Its mass splits into an irreducible part, fixed by the horizon area, and an extractable rotational part. In geometrised units ($G=c=1$) the Christodoulou–Ruffini relation is^[10][25]

$M^2 = M_{\text{irr}}^2 + \frac{J^2}{4M_{\text{irr}}^2}, \qquad M_{\text{irr}} = M\sqrt{\tfrac{1}{2}\big(1+\sqrt{1-a_*^2}\,\big)},$

so the extractable fraction is

$\frac{E_{\text{rot}}}{Mc^2} = 1 - \frac{M_{\text{irr}}}{M} = 1 - \sqrt{\tfrac{1}{2}\big(1+\sqrt{1-a_*^2}\,\big)}.$

At maximal spin $a_* = 1$, $M_{\text{irr}}/M = 1/\sqrt2$, and

$\boxed{\ \frac{E_{\text{rot}}}{Mc^2}\bigg|_{a_*=1} = 1 - \frac{1}{\sqrt2} \approx 0.29.\ }$

Against fusion. Hydrogen-to-helium fusion liberates $\approx 0.007\,Mc^2$. The ratio $0.29/0.007 \approx 41$: black-hole spin is a store ~40× denser per unit mass than the process that lights a star.

Mechanisms. The Penrose process extracts $E_{\text{rot}}$ mechanically inside the ergosphere, $r < r_{\text{ergo}}(\theta) = (GM/c^2)\big(1+\sqrt{1-a_*^2\cos^2\theta}\,\big)$, where no observer can remain static and negative-energy orbits exist.^[9] The Blandford–Znajek mechanism extracts it electromagnetically: a horizon threaded by magnetic flux $\Phi$ radiates with power scaling as^[11]

$P_{\text{BZ}} \sim \frac{\kappa}{4\pi\mu_0 c}\,\Phi^2\,\Omega_H^2, \qquad \Omega_H = \frac{a_* c^3}{2GM\big(1+\sqrt{1-a_*^2}\,\big)},$

with $\kappa \approx 0.05$ and $\Omega_H$ the horizon angular velocity, giving $P_{\text{BZ}} \propto B^2 M^2 a_*^2$. For supermassive holes in galactic magnetic fields this reaches $10^{36}$–$10^{39}$ W, the luminosity class of observed quasar jets.

Reservoir size. The extractable energy of a maximal hole, $0.29\,Mc^2$, is $5\times10^{47}$ J for $10\,M_\odot$ and $5\times10^{55}$ J for $10^9\,M_\odot$. Against global primary energy use (~$6\times10^{20}$ J yr⁻¹) these are ~$10^{27}$ and ~$10^{35}$ years of supply — the second far exceeding the stelliferous era.^[8]

E. Cosmological Natural Selection as a Replicator Process

Smolin’s mechanism is selection on a heritable, mutating, differentially reproducing population, and takes the standard form.^[14] Let a universe’s low-energy constants be $p \in \mathbb{R}^d$, with fitness $W(p)$ = the expected number of black holes it forms, each seeding one offspring with constants $p' = p + \epsilon$ drawn from a narrow mutation kernel $K(\epsilon)$, $\langle\epsilon\rangle = 0$, $\langle\epsilon^2\rangle = \eta^2$ small. The cross-generation density evolves as

$f_{n+1}(p) = \frac{1}{\bar W_n}\int K(p-p')\,W(p')\,f_n(p')\,dp', \qquad \bar W_n = \int W f_n\,dp.$

For small $\eta$ this is a mutation–selection balance whose stationary density concentrates on local maxima of $W$. Expanding at a maximum $p_0$ ($\nabla W(p_0) = 0$, Hessian $H \prec 0$), the equilibrium spread is $\langle \delta p\,\delta p^\top\rangle \sim \eta^2 |H|^{-1}$. The empirical content is Smolin’s prediction:

$\nabla W(p_0) = 0,\quad H = \nabla^2 W(p_0) \prec 0 \;\Rightarrow\; \text{any small change in } p \text{ lowers black-hole yield.}$

This is what the neutron-star-mass and cosmological-constant critiques test, and contest.^[14][26]

The civilisation term. Split the yield as $W(p) = W_\star(p) + W_{\text{civ}}(p)$: black holes from stellar collapse, and those built by civilisations. $W_{\text{civ}}$ is supported only on the life-permitting set $L \subset \mathbb{R}^d$ — long-lived stars, stable chemistry, the conditions catalogued across the canon. If $\nabla W_{\text{civ}}$ is non-negligible on $L$, the selection maximum shifts toward life-permitting constants, and “selected for black holes” entails “selected for the civilisations that build them.” The requirement is a magnitude comparison,

$\|\nabla W_{\text{civ}}\| \gtrsim \|\nabla W_\star\| \quad \text{on } L,$

whose truth turns on whether engineered black holes are numerous enough to bend the landscape. That is unknown — the essay’s weakest beam, made quantitative. The same extension has been proposed independently.^[27]

F. Typicality, Self-Sampling, and the Measure Problem

The body argues the empty sky is the prediction. Formally, the lineage is a branching process with mean reproductive ratio $R = \langle W\rangle > 1$ per generation; after $n$ generations the count is $N_n \sim R^n$. A self-sampling observer reasons as a random draw from the observer-measure $\mu$ over the whole ensemble. Weighting by reproductive recency — each generation $R\times$ more numerous than the last —

$P(\text{generation } n) \propto R^n,$

so $\mu$ is dominated by the most recent generations. Those are, by construction, universes whose own civilisations have not yet filled them. The typical observer is therefore early, in a fertile and fine-tuned universe, with no visible peers — the observed situation.

The measure problem. The argument is conditional. $\sum_n R^n$ diverges; the ensemble is not finitely normalisable, so $\mu$ requires a regulator (a proper-time, scale-factor, or stopping cutoff). Different regulators select different “typical” observers, and some produce pathologies — the youngness paradox, Boltzmann-brain domination.^[20] No canonical measure is known. The qualitative conclusion (“early, alone”) is robust across the class of measures that weight by reproductive proximity; it is not derivable without committing to that class. This is the precise sense of “rigorous in spirit, not in arithmetic.”

G. Quantum-Gravity Computation: The Ladder and the Counting

§V is the frontier, drawn in lighter ink. The speculation is not a single leap but a ladder, and most of its rungs are already standing. This appendix separates them: what is established, what is seriously theorised, and what remains conjecture.

The ladder.

1. State space. A quantum computer of $N$ two-level systems explores a Hilbert space of dimension $2^N$ — every superposition one fixed physics permits.

2. Order space. Causal order need not be a fixed background. The process-matrix framework describes quantum correlations with no definite global causal order,^[18] and quantum theory admits transformations of operations not realisable by any fixed-order circuit.^[28] The laboratory instance is the quantum switch, where two operations act in a superposition of orders. Causal order is, demonstrably, a computational variable.

3. Causal-structure space. Lucien Hardy’s causaloid framework formalises computation with no definite background causal structure — his “quantum gravity computer.”^[17] Markopoulou’s quantum causal histories attach Hilbert spaces and local evolution to the events of a causal set, so a causal structure can itself carry quantum information.^[29]

4. Geometry. In general relativity causal structure is dynamical, and Malament’s theorem shows it fixes the spacetime metric up to a single conformal factor^[16] — to range over causal structure is therefore nearly to range over geometry. Lloyd runs the implication the other way: spacetime geometry derived from underlying quantum computation, as a superposition of geometries.^[30] Computation and geometry are not separate primitives.

5. The singularity boundary. Where classical causal order ends and quantum gravity must take over. The Horowitz–Maldacena proposal treats the singularity as a final-state boundary condition on the interior^[31] — a precedent for reading it as an information-theoretic boundary, though the proposal is contested.

The counting. Causal-set theory makes the top of the ladder countable. A four-volume $V$ is a locally finite partial order of $N \sim V/\ell_P^4$ elements,^[15] and the number of partial orders on $N$ labelled elements grows as^[32]

$\log_2 \#\{\text{posets on } N\} \sim \frac{N^2}{4},$

so the space of causal structures has size $\sim 2^{N^2/4}$ against the $2^N$ states of $N$ qubits. The exponent scales as $N^2$, not $N$: the architecture-space is super-exponentially larger than the state-space within any one architecture. That ratio is the only precise content behind “ranging over architectures, not states.”

The caveats. Not every causal structure is admissible — only those compatible with quantum mechanics — so the relevant space is constrained, not arbitrary. And the black hole is where these threads meet: the complexity-equals-action conjecture ties computational complexity to a region of black-hole spacetime and casts black holes as the fastest computers nature permits.^[33]

The boundary of the claim. Rungs 1–2 are established, in theory and in the laboratory. Rungs 3–4 are serious theoretical frameworks. Rung 5, and the essay’s further step — that a computation at the singularity boundary could range over the law-space itself — is conjecture. §V climbs the ladder the literature has built, then names the last rung as speculation. That honesty is the section’s licence to reach.

H. What Is Falsifiable, and What Is Not

The essay depends on its optimism, so its limits should be stated honestly. The essay makes three claims of decreasing security, and they should not be confused.

The pattern is the strongest claim, and the most testable. Across independent domains, perception, computation, energy, materials, propulsion, optics, the universe turns out richer than assumed and the access gap turns out closable. This is the empirical content of the prior eight essays. It would be weakened by domains where abundance proves illusory or access structurally impossible. The domains were not selected to fit. Each rests on mainstream physics and observation established for unrelated reasons.

Cosmological natural selection is the proposed explanation, and it makes testable predictions. Smolin argued that if our universe is near-optimal for black hole production, small changes to the constants should reduce that production. The most discussed test concerns the maximum mass of neutron stars, since a higher limit would open an additional formation channel and imply our universe is not at an optimum. The prediction is contested. Observations of massive neutron stars have been used against it, and critics argue that black hole production could be raised, not lowered, by changing other parameters such as the cosmological constant, which would undercut the claim that our universe sits at an optimum.^[14][26] The status is open, not settled.

The civilisation extension is the frontier, and the weakest beam. That civilisations can engineer black holes whose parameters seed offspring universes lies far beyond any experiment. The move is not unique to this essay: the proposal that cosmological selection might favour technology, not only stellar collapse, has been raised independently.^[27] It carries one observational consequence: by typicality, a randomly placed observer should find themselves early and alone in a fine-tuned universe rather than late in a crowded one, which is what we find. But this rests on the cosmological measure problem (§F), the unsolved question of how to count observers across a possibly-infinite ensemble. Until that is solved, “more likely” is rigorous in spirit, not in arithmetic. Section V is further out still: a direction the physics suggests, not a result it has reached.

What would weaken the whole picture: domains where the access gap proves uncloseable, a demonstration that black holes do not produce offspring universes, that the physical constants cannot vary across them, or that the information and energy limits derived in §B–§D have been miscalculated. The argument is built to be wrong if the physics is wrong. That is the point.

[1] Ashby, N. (2003). “Relativity in the Global Positioning System.” Living Reviews in Relativity 6, 1. (Open version: arXiv:gr-qc/0303117.) (Why GPS requires both special- and general-relativistic corrections to keep accurate time and position.)

[2] Schwarzschild, K. (1916). “On the Gravitational Field of a Mass Point According to Einstein’s Theory.” Sitzungsberichte der Königlich Preussischen Akademie der Wissenschaften, 189–196. (The first exact nontrivial solution of Einstein’s field equations, later understood as the Schwarzschild black-hole solution.)

[3] Oppenheimer, J. R. & Snyder, H. (1939). “On Continued Gravitational Contraction.” Physical Review 56, 455–459. (The classic model of unhalted gravitational collapse, the link between stellar death and black-hole formation.)

[4] Penrose, R. (1965). “Gravitational Collapse and Space-Time Singularities.” Physical Review Letters 14, 57–59. Awarded the Nobel Prize in Physics 2020 for the discovery that black hole formation is a robust prediction of general relativity. See Nobel Prize in Physics 2020.

[5] Bremermann, H. J. (1962). “Optimization through Evolution and Recombination.” In Self-Organizing Systems, eds. Yovits, Jacobi & Goldstein, 93–106. Spartan Books. (The maximum computation rate of a physical system bounded by its mass-energy; commonly cited as ~1.36 × 10⁵⁰ bits per second per kilogram.)

[6] Bekenstein, J. D. (1973). “Black Holes and Entropy.” Physical Review D 7, 2333–2346. (Upper bound on the entropy, and hence information, of a bounded region; saturated by black holes.)

[7] Lloyd, S. (2000). “Ultimate Physical Limits to Computation.” Nature 406, 1047–1054. (Open version: arXiv:quant-ph/9908043.) (The “ultimate laptop”: maximum operations per second for a given mass, maximised by a black hole.)

[8] Adams, F. C. & Laughlin, G. (1997). “A Dying Universe: The Long-Term Fate and Evolution of Astrophysical Objects.” Reviews of Modern Physics 69, 337–372. See also Dyson, F. J. (1979). “Time Without End: Physics and Biology in an Open Universe.” Reviews of Modern Physics 51, 447–460. (The deep-time fate of the universe; the long epoch after stellar fusion ends.)

[9] Penrose, R. (1969). “Gravitational Collapse: The Role of General Relativity.” Rivista del Nuovo Cimento 1, 252–276. (The Penrose process: energy extraction from a rotating black hole’s ergosphere.)

[10] Christodoulou, D. (1970). “Reversible and Irreversible Transformations in Black-Hole Physics.” Physical Review Letters 25, 1596–1597. (Irreducible mass; up to ~29% of a maximally spinning black hole’s mass-energy is extractable.)

[11] Blandford, R. D. & Znajek, R. L. (1977). “Electromagnetic Extraction of Energy from Kerr Black Holes.” Monthly Notices of the Royal Astronomical Society 179, 433–456. (The electromagnetic mechanism powering relativistic jets from rotating black holes.)

[12] Hawking, S. W. & Penrose, R. (1970). “The Singularities of Gravitational Collapse and Cosmology.” Proceedings of the Royal Society A 314, 529–548. (Extension of the singularity theorems to cosmological spacetimes, including the Big Bang.)

[13] Ashtekar, A. & Singh, P. (2011). “Loop Quantum Cosmology: A Status Report.” Classical and Quantum Gravity 28, 213001. (Open version: arXiv:1108.0893.) (The quantum “bounce” replacing the classical initial singularity.)

[14] Smolin, L. (1992). “Did the Universe Evolve?” Classical and Quantum Gravity 9, 173–191. See also Smolin, L. (1997). “The Life of the Cosmos.” Oxford University Press. On the falsifiability discussion and the neutron-star-mass critique, see Smolin, L. (2006). “The status of cosmological natural selection.” arXiv:hep-th/0612185. (Cosmological natural selection; its predictions and their contested status.)

[15] Bombelli, L., Lee, J., Meyer, D. & Sorkin, R. D. (1987). “Space-Time as a Causal Set.” Physical Review Letters 59, 521–524. (Spacetime as a discrete partial order of events; geometry emergent from causal structure.)

[16] Malament, D. B. (1977). “The Class of Continuous Timelike Curves Determines the Topology of Spacetime.” Journal of Mathematical Physics 18, 1399–1404. (A spacetime’s causal structure fixes its topology and its metric up to a conformal factor — the basis for “order plus number equals geometry.”)

[17] Hardy, L. (2007). “Quantum Gravity Computers: On the Theory of Computation with Indefinite Causal Structure.” arXiv:quant-ph/0701019. (Introduces the “quantum gravity computer” and the causaloid framework: computation where causal structure is not a fixed background.)

[18] Oreshkov, O., Costa, F. & Brukner, Č. (2012). “Quantum Correlations with No Causal Order.” Nature Communications 3, 1092. (Open version: arXiv:1105.4464.) (The process-matrix framework: local quantum mechanics with no predefined global causal order, realised in toy form by the quantum switch.)

[19] Maldacena, J. & Susskind, L. (2013). “Cool Horizons for Entangled Black Holes.” Fortschritte der Physik 61, 781–811. (Open version: arXiv:1306.0533.) (The ER=EPR proposal: entanglement and wormhole geometry as two descriptions of one structure.)

[20] Barrow, J. D. & Tipler, F. J. (1986). The Anthropic Cosmological Principle. Oxford University Press. See also Carter, B. (1974). “Large Number Coincidences and the Anthropic Principle in Cosmology.” In Confrontation of Cosmological Theories with Observational Data, IAU Symposium 63, 291–298. Reidel. (The standard reference works on anthropic reasoning and observer-selection effects.)

[21] Weinberg, S. (1987). “Anthropic Bound on the Cosmological Constant.” Physical Review Letters 59, 2607–2610. (The landmark anthropic prediction constraining the cosmological constant from the requirement that galaxies form, the multiverse approach’s clearest success.)

[22] Margolus, N. & Levitin, L. B. (1998). “The Maximum Speed of Dynamical Evolution.” Physica D 120, 188–195. (Open version: arXiv:quant-ph/9710043.) (A quantum speed limit: the maximum rate at which a system can move between distinguishable states, set by its energy.)

[23] Bardeen, J. M., Carter, B. & Hawking, S. W. (1973). “The Four Laws of Black Hole Mechanics.” Communications in Mathematical Physics 31, 161–170. (The formal correspondence between black-hole mechanics and thermodynamics underlying the Bekenstein–Hawking entropy.)

[24] Hawking, S. W. (1975). “Particle Creation by Black Holes.” Communications in Mathematical Physics 43, 199–220. (Hawking radiation: black holes emit thermal radiation at temperature $T_H \propto 1/M$ and evaporate over a time $\propto M^3$.)

[25] Christodoulou, D. & Ruffini, R. (1971). “Reversible Transformations of a Charged Black Hole.” Physical Review D 4, 3552–3555. (The black-hole mass formula relating total mass, irreducible mass, angular momentum, and charge.)

[26] Vilenkin, A. (2006). “On cosmic natural selection.” arXiv:hep-th/0610051. (A critique of cosmological natural selection, arguing black hole production could be increased by altering parameters such as the cosmological constant, weakening the claim that our universe is near-optimal.)

[27] Shainline, J. M. (2020). “Does cosmological evolution select for technology?” New Journal of Physics 22, 073064. (arXiv:1912.06518.) (An independent development of cosmological natural selection in which technological civilisations, not only stellar collapse, drive black hole production and so universe reproduction.)

[28] Chiribella, G., D’Ariano, G. M., Perinotti, P. & Valiron, B. (2013). “Quantum Computations without Definite Causal Structure.” Physical Review A 88, 022318. (Open version: arXiv:0912.0195.) (Quantum theory permits transformations of operations not realisable by any circuit with a predefined causal order.)

[29] Markopoulou, F. (2000). “Quantum Causal Histories.” Classical and Quantum Gravity 17, 2059–2072. (Open version: arXiv:hep-th/9904009.) (Causal sets with Hilbert spaces and local evolution attached to events — causal structure carrying quantum information.)

[30] Lloyd, S. (2005). “A Theory of Quantum Gravity Based on Quantum Computation.” arXiv:quant-ph/0501135. (Spacetime geometry derived from underlying quantum information processing, yielding a superposition of spacetimes.)

[31] Horowitz, G. T. & Maldacena, J. (2004). “The Black Hole Final State.” Journal of High Energy Physics 2004(02), 008. (Open version: arXiv:hep-th/0310281.) (A final-state boundary condition imposed at the singularity — precedent, though contested, for the singularity as an information-theoretic boundary.)

[32] Kleitman, D. J. & Rothschild, B. L. (1975). “Asymptotic Enumeration of Partial Orders on a Finite Set.” Transactions of the American Mathematical Society 205, 205–220. (The number of partial orders on $N$ labelled elements grows as $2^{N^2/4 + o(N^2)}$.)

[33] Brown, A. R., Roberts, D. A., Susskind, L., Swingle, B. & Zhao, Y. (2016). “Complexity Equals Action.” Physical Review Letters 116, 191301. (Open version: arXiv:1509.07876.) (Ties computational complexity to a region of black-hole spacetime; casts black holes as the fastest computers in nature.)