The Game Theory of the Quran
Seven secular research programmes converge on prescriptions a 7th century text already contained. A research programme, stated precisely enough to be attacked.
Game theory studies how self-interested agents should act when the outcome depends on what others do. It was formalised in the twentieth century, and it has been unkind to intuition. Strategies that feel virtuous often lose. Strategies that feel weak often win. The results are not opinions, they are consequences of the payoff structure.
This paper documents a pattern I cannot explain away. Across seven independent research programmes, all secular, all built by people with no interest in defending Islam, the conclusions keep landing on prescriptions the Quran already contains. Not vaguely. Specifically, at the level of mechanism.
I want to be careful about what that does and does not establish. This is not a proof, and the parts of the traditional apologetic that are easiest to attack are deliberately absent. What follows is a research programme: a claim stated precisely enough that it can be tested, a set of rival explanations stated at their strongest, and the specific findings that would sink it.
1. The claim, stated precisely
Three claims sit inside this framework, and conflating them is how the argument usually dies.
Claim A, descriptive convergence. Quranic social prescriptions correspond, at the level of mechanism rather than sentiment, to solutions that formal cooperation research independently arrived at. This is a claim about correspondence. It is the easiest to check.
Claim B, optimality. Those prescriptions are not merely one workable option among many, but close to the best available strategy for the environments humans actually inhabit: repeated interaction, noisy information, mixed populations of cooperators and defectors.
Claim C, inference to design. The correspondence is better explained by the text originating outside the 7th century than by any naturalistic account.
This paper argues A in detail, argues B for a subset of cases and flags where the word “optimal” is doing more work than it has earned, and treats C as the open question rather than the conclusion. Section 10 gives the strongest naturalistic rival I know of and does not pretend to have refuted it.
I should also state my position rather than hide it. I am Muslim. I hold the Quran to be the sole certain foundation of the religion, with the hadith corpus read critically and never permitted to override it. That commitment is a source of motivated reasoning and I have tried to build the paper so that a hostile reader can check each step without trusting me. Where the argument leans on material my own epistemology treats as probabilistic, I say so at that point rather than quietly borrowing the strength.
2. Why game theory is the right lens
Most moral instruction fails a simple test: it tells you to be good without telling you what happens when the other party is not. Game theory is useful precisely because it takes that seriously.
Three features of the formalism matter here.
Repetition changes everything. In a one-shot prisoner’s dilemma, defection dominates. In an indefinitely repeated one, cooperation becomes sustainable, because today’s defection is paid for tomorrow. Almost every institution that looks irrational in isolation makes sense once you notice it is converting a one-shot interaction into a repeated one.
Signals must be costly to be credible. Anyone can say they are committed. A signal only separates the committed from the opportunistic if it costs more to fake than it is worth. This is Spence’s result, and it explains an enormous amount of otherwise strange social machinery.
Mechanism design runs the logic backwards. Instead of asking how agents behave under given rules, it asks what rules produce the behaviour you want from agents who will pursue their own interest regardless. This is the framing under which the Quran’s institutional prescriptions become legible as engineering rather than as ritual.
With those three tools, the convergences below are not analogies. They are the same structures described in two vocabularies.
3. Convergence one, the cooperation strategy
In 1980 Robert Axelrod invited researchers to submit strategies for a repeated prisoner’s dilemma tournament. The winner, submitted by Anatol Rapoport, was the shortest programme entered: Tit for Tat. Cooperate on the first move. Then do whatever the other player did last. It won the rerun as well, with every entrant knowing the previous result.
Axelrod’s analysis of why it won identified four properties.
| Property | What it means | Quranic correspondence |
|---|---|---|
| Nice | Never defects first | “If they incline to peace, incline to it” (8:61) |
| Retaliatory | Punishes defection immediately | “Whoever transgresses against you, respond in kind” (2:194) |
| Forgiving | Returns to cooperation the moment the other does | “Whoever pardons and reconciles, his reward is with God” (42:40) |
| Clear | Behaviour is legible, so others can adapt to it | Rules stated publicly, in a text open to all |
The verse in 2:194 is worth reading closely, because the constraint is doing real work. Retaliation is permitted, and it is capped at proportionality, then immediately followed by a call to restraint. An uncapped punishment strategy loses badly in Axelrod’s environment. It triggers echo effects that neither player can exit.
Later work found Tit for Tat has a specific weakness: in a noisy environment, where a cooperative move is occasionally misread as defection, two Tit for Tat players lock into an alternating vendetta neither intended. The fix, identified by Nowak and Sigmund, is Generous Tit for Tat: forgive some fraction of defections unprompted, absorbing a small loss to break the cycle.
The Quran prescribes this too, and marks it as the higher of two permitted options: “Repel evil with what is better, and the one between whom and you there was enmity will become as an intimate friend” (41:34). The verse does not merely permit generosity. It states the mechanism, that unprompted cooperation converts an adversary’s strategy, which is exactly what generous strategies do to the population dynamics.
One more piece completes the structure. A retaliatory strategy only deters if the capacity to retaliate is visible. “Prepare against them what you are able of force” (8:60) states that requirement, and the verse continues by naming deterrence rather than conquest as the purpose, then 8:61 immediately caps it by ordering acceptance of any peace offered. Visible capability, capped use, first move cooperative. That is the full specification.
Where I am overclaiming. Tit for Tat is not optimal in the sense of a theorem. It won particular tournaments against particular fields, and no strategy is optimal against all opponents. In evolutionary settings it is invaded by unconditional cooperators during quiet periods, and Win Stay Lose Shift outperforms it under some noise regimes. The honest version of Claim B here is narrower: the Quran specifies a nice, proportional, forgiving, transparent, credibly deterrent strategy, and that region of strategy space is where sixty years of cooperation research has repeatedly ended up. That is still a striking thing for a 7th century text to do. It is not a proof of mathematical optimality, and I will not use the word that way again.
4. Convergence two, institutions as mechanism design
Prescribing a good strategy to individuals is the easy half. The hard half is designing institutions that hold when individuals defect anyway. Four Quranic institutions read as mechanism design.
The prohibition of riba. Interest creates a payoff structure where the lender’s return is decoupled from the outcome of the venture. The borrower carries all the downside, the lender carries none. In repeated play across a population, this concentrates capital in a way that requires no malice from anyone. Piketty’s account of r greater than g describes the same engine from the other end: when the return on capital exceeds growth, inherited wealth compounds faster than earned wealth, and concentration follows mechanically.
The Quranic alternative is not the absence of finance. It is profit and loss sharing, in which the financier’s return is a share of the actual outcome. That converts an asymmetric game into a symmetric one. Both parties now face the same state of the world, so both have an interest in the venture succeeding and in reporting honestly. The lender acquires what Taleb calls skin in the game, which is the same condition stated in risk vocabulary.
Note what sits immediately next to the riba passage: the longest verse in the Quran (2:282) is a detailed contract-recording protocol. Write the debt down, specify the term, use a scribe, use witnesses. That is not piety, it is verification infrastructure, and cooperation research is unambiguous that enforcement without verification does not work.
Where I am overclaiming. Piketty’s r greater than g is contested by serious economists on both the empirical trend and the mechanism. And prohibiting interest does not eliminate returns to capital, it reallocates risk and changes the contract’s form. Some contemporary Islamic finance instruments reproduce interest economically while satisfying the letter of the prohibition, which suggests the mechanism is easier to name than to enforce. What survives the criticism is narrower and still substantive: the Quran prohibits precisely the risk-asymmetric contract that formal analysis identifies as the concentrating one, and mandates the risk-symmetric alternative.
Zakat as load balancing. A fixed proportional transfer from accumulated wealth to eight named categories of recipient (9:60) is a stabiliser on a system that otherwise runs away. The stated purpose is explicit and unusually mechanical for scripture: “so that it does not circulate solely among the wealthy among you” (59:7). Anti-hoarding injunctions (9:34, 104:2) close the obvious evasion. Winner-take-all dynamics are among the best documented destabilisers in the study of complex systems, and this is a designed damper on them.
The marriage framework as a signalling and enforcement structure. Unregulated mate selection is a market in which the cheapest strategy is to signal commitment without bearing its cost. The Quranic framework attacks this at three points. The mahr (4:4) is a transfer from the man to the woman, obligatory, hers alone, which is a costly signal in Spence’s exact sense. Family involvement embeds the pairing in a network that will still be present in ten years, converting a one-shot into a repeated game with third-party enforcement. Public announcement makes the commitment common knowledge, so defection carries reputational cost. Each element targets a distinct failure mode.
Consultative decision-making. The believers’ affairs are decided “by consultation among themselves” (42:38), and the Prophet himself is ordered to consult (3:159). Collective-choice arrangements turn out to be a necessary condition for durable common-pool governance, which brings us to the second Nobel laureate in this argument.
5. Five further convergences
The pattern is not confined to game theory narrowly construed. Five more research programmes, developed independently, land in the same place.
Ostrom, commons governance. Elinor Ostrom won the 2009 Nobel for showing that communities routinely manage shared resources sustainably without either privatisation or state control, and for identifying eight design principles that successful cases share. The correspondence is close enough to be worth tabulating.
| Ostrom principle | Quranic or early-Islamic correspondence |
|---|---|
| Clearly defined boundaries | The ummah as a bounded membership with explicit entry |
| Rules congruent with local conditions | Ijtihad and the recognition of urf as a legal source |
| Collective-choice arrangements | Shura (42:38, 3:159) |
| Monitoring | Hisbah, grounded in the injunction to enjoin right and forbid wrong (3:104) |
| Graduated sanctions | Expiations scaled to means and to the offence (5:89, 58:3 to 4) |
| Conflict-resolution mechanisms | Sulh, with a mandated arbitration procedure (49:9 to 10, 4:35) |
| Recognised right to organise | Communal autonomy in religious and civil affairs |
| Nested enterprises | Household, local congregation, regional community, ummah |
Several of these rows are weaker than the others and I am not going to hide them. Congruence with local conditions and the right to organise are properties of the juristic tradition that developed around the text, not statements in the text. Monitoring is in the same position: hisbah as an institution, with an appointed muhtasib, is a later development, and what the Quran itself supplies in 3:104 is a general injunction to enjoin right and forbid wrong, not a monitoring mechanism. And nested enterprises describes the layered structure of most human societies, so that row cannot discriminate between this text and almost any other. If you restrict the comparison to what the Quran actually states, the defensible score is four clear correspondences out of eight, five if a general injunction counts as monitoring, with the rest supplied by later institutional development. That is a weaker headline than the table suggests, and I would rather report it than defend the table.
Taleb, antifragility. Systems that gain from disorder share a structure: optionality, skin in the game, and a barbell of extreme conservatism in the core combined with aggressive flexibility at the edges. The Quran states adversity as a strengthening mechanism rather than an anomaly to be explained away (2:214, 2:155 to 157), builds skin in the game into its financial architecture, and pairs a small fixed core of non-negotiables with wide latitude in implementation. The claim here is structural correspondence, and Taleb’s framework is itself informal, so this convergence is suggestive rather than decisive.
Turchin, cliodynamics. Turchin’s mathematical models of secular cycles identify elite overproduction, meaning more claimants to elite position than positions available, and declining asabiyyah, meaning collective solidarity, as the recurring drivers of instability. He took the second concept directly from Ibn Khaldun. The Quran’s historical narrative sections describe the same cycle repeatedly: a people prosper, grow arrogant, abandon the framework, and are replaced. Its counter-mechanisms map onto Turchin’s drivers with unusual directness. Anti-hoarding constrains elite wealth accumulation. Status assigned by taqwa rather than lineage or wealth (49:13) attacks the credential race at its root. Mandatory collective ritual maintains exactly the solidarity variable his models track.
Kahneman and Tversky, behavioural economics. The Quran addresses biases not formally described until the 1970s. Hyperbolic discounting, our systematic overvaluing of the near term, is named directly: “But you prefer the worldly life, while the Hereafter is better and more lasting” (87:16 to 17). That is a statement of the bias and of its correction in two clauses. Social proof is warned against explicitly: “If you obeyed most of those on earth, they would lead you astray” (6:116), which contradicts the default human heuristic. The endowment effect is undercut by reframing ownership as trusteeship (57:7). And the vivid asymmetric depiction of reward and punishment reads as deliberate exploitation of loss aversion in the service of long-horizon behaviour.
Network theory. Five daily congregational prayers produce dense local clustering. Weekly assembly (62:9) connects the local clusters into a larger component. The annual pilgrimage (22:27) mixes the global network once a year, adding long-range edges between otherwise distant clusters. High clustering plus sparse long-range shortcuts is the exact recipe for a small-world topology, which is optimal for propagating information rapidly while preserving local cohesion. This is one of the cleanest structural correspondences in the set, and it is generated purely by the schedule.
The tension I am not going to paper over. My working notes also mapped the isnad, the hadith transmission chain, onto authenticated distributed systems. I have removed it, and the reason matters. My own position is that hadith reports are probabilistic and cannot serve as independent foundations. Using the impressiveness of the isnad apparatus to bolster a case about the Quran’s origin would be borrowing authority from a source I otherwise subordinate. It would also be a weak argument: chain authentication verifies transmission, not truth of content, which is precisely the objection.
6. The deeper frame, reality as a designed game
If the prescriptions match, the interesting question is structural rather than merely comparative. The Quran does not present itself as good advice. It presents existence as a bounded test with a designer, rules, players, choices, and terminal payoffs. That is the definition of a game.
The framing is explicit. Life is described as play and diversion relative to what follows (57:20), the hereafter as “the real life” (29:64), the purpose of creation as a test of conduct (67:2, 18:7), and the burden as never exceeding capacity (2:286), which in formal terms is a guarantee that the feasible strategy set is non-empty for every player.
The part I find most interesting is the information structure. A standard objection to religion asks why a designer who wants belief would leave the evidence ambiguous. In game-theoretic terms an answer is available, and it has more structure than the objection assumes. If the payoffs were common knowledge and verifiable, cooperation would be the dominant strategy for every agent regardless of type. Everyone would cooperate, and the game would separate no one, because the observed behaviour of a virtuous agent and a purely calculating agent would be identical. A test that distinguishes types requires that the payoff not be commonly known. The uncertainty is not a defect in the design, it is the condition that makes the design capable of measuring anything.
The text states the same requirement in its own vocabulary. “There is no compulsion in religion, the right course has become distinct from error” (2:256). Sufficient signal to make a choice meaningful, insufficient proof to make it forced. “We guided him to the way, whether he be grateful or ungrateful” (76:3). This is a separating equilibrium described in the seventh century.
Where I am overclaiming. Two problems, and the first is close to fatal for the formal version of this argument. Screening models exist to solve an uninformed principal’s problem: the whole machinery of separation is needed because the principal cannot observe type directly. The Quran’s God is not that principal. He is described as knowing what the breasts conceal (40:19) and as nearer to a man than his jugular vein (50:16), so no behavioural separation is required for Him to know types. The text’s own framing is that the test makes conduct manifest in actuality (29:2 to 3), for the record and for the agent, not that it produces information the designer lacks. That is a coherent theological position, but it is not Spence’s, and I should not borrow Spence’s authority for it. Second, the claim that common knowledge of the payoffs would collapse separation is too quick. With payoffs of the magnitude the text describes, cooperation is close to dominant under any non-trivial credence, which is Pascal’s old point, and ambiguity alone does not restore separation without assumptions about discounting and risk attitudes that I have not stated. The defensible claim is a correspondence of structure: the information environment the text describes and explicitly defends is the one a type-separating test would require. The formal machinery does not transfer cleanly, and this section is an analogy with unusual internal consistency, not a derivation.
7. The 99 Names as a system specification
This section is more speculative than what precedes it, and I flag it as such.
Read as a specification document, the divine attributes cluster by function. Design attributes (al-Khaliq, al-Alim, al-Hakim) assert intentional construction. Fairness attributes (al-Adl, al-Haqq, ar-Raqib) establish that the payoff structure is credible and that actions are observed, which is the precondition for any enforcement to bite. Mercy attributes (ar-Rahman, ar-Rahim, al-Ghaffar) establish that the punishment branch is not absorbing, so a defector always has a path back, which is exactly the forgiveness property from Section 3 stated as a property of the system rather than of the player. Power attributes (al-Qadir, al-Jabbar) make enforcement credible. Relational attributes (al-Mujib, al-Wali) keep the channel two-way.
The attributes also arrive in complementary pairs. The one who constricts and the one who expands, the one who honours and the one who abases, the outward and the inward. A system governed by balanced opposing forces is the standard architecture of dynamic stability. A specification consisting only of mercy attributes describes a system with no enforcement. One consisting only of power attributes describes one with no recovery path. The pairing is the design.
The furthest step, and the one I hold most loosely, is this: several attributes are not expressible without creation. Forgiveness requires someone who errs. Mercy requires someone in need. Acceptance of repentance requires someone who strays. If those attributes are essential rather than incidental, then creation, moral agency, error, and judgement follow as a deductive cascade from the designer’s nature rather than as arbitrary choices. I find this compelling and I recognise that it is metaphysics, not evidence. It belongs in the research programme, not in the case.
8. Eschatology as civilizational diagnostic
Here I have to be most careful, on two fronts.
The first is epistemic. The “minor signs” tradition is hadith material, not Quranic. By my own stated standard it cannot serve as evidence. What it can legitimately do is generate hypotheses, which are then tested against independent data. That is the only weight I put on it in this section, and readers who reject the corpus entirely can skip to the demographic argument, which does not depend on it.
The second is political, because this section is where a structural argument most easily degrades into a grievance. I am describing failure modes of a system, not the moral worth of the people living in it. I am also a beneficiary of that system, writing from inside it, in its languages, using its institutions.
With those stated: the traditional signs read as a diagnostic checklist. Widespread ignorance despite unprecedented access to information. Intoxication as routine self-medication. Collapse of the marriage structure. Normalised dishonesty in commerce. Formerly poor populations competing in monumental construction. Time appearing to accelerate. Purposeless violence.
The checklist also has to survive a control test of its own, and mostly it does not. Widespread ignorance, routine intoxication, commercial dishonesty, purposeless violence: these are near-universal features of civilisational decline, not signatures of this one. Late Rome would have matched most of the list, and so would several other eras. A diagnostic that matches every declining era detects decline in general, which is not nothing, but it identifies no particular era and therefore predicts nothing. The items with era-specific content are the ones worth keeping as hypothesis generators: formerly poor populations competing in monumental construction has a referent now that it could not have had in most centuries, and the collapse of the marriage structure connects to the one genuinely unprecedented datum below, voluntary sub-replacement fertility correlated with prosperity. The rest of the checklist I now treat as colour, not evidence.
Set that against what the cliodynamic and demographic literature independently reports, and the overlap is uncomfortable. Fertility across the developed world sits below replacement, with South Korea near 0.7, which halves each successive cohort. Partner markets exhibit exactly the defection dynamics the model predicts, with matching efficiency falling as search costs fall. Wealth concentration has returned to pre-1914 levels. Institutional trust is declining across nearly every developed democracy. Elite overproduction is measurable in credential inflation. Every one of these is documented by researchers with no religious commitment.
The game-theoretic reading is that these are not separate problems. They are the predicted consequences of a civilisation that optimised hard for individual freedom and did so by dismantling the shared superstructure that made cooperative equilibria self-enforcing. That trade produced enormous short-run returns. Axelrod’s finding is that defection strategies are highly profitable until the repeated game catches up with them.
The demographic case for a superstructure effect is the cleanest available test, because it is arithmetic rather than interpretation. Israel is the one high-income country sustaining fertility near 2.9, and it is also the one that retained a strong collective superstructure: shared national identity, near-universal service, religious tradition, and unambiguous external stakes. Correlation across a single case is not causation, and Israel differs from its comparison group in many ways at once. But it is the observation most awkward for the view that below-replacement fertility is a pure income effect, and the direction is the one the framework predicts.
9. The physics connection, and why I keep it short
The Quran describes God not as an entity located within reality but as the ground on which reality depends: as-Samad, that on which everything depends and which depends on nothing; “there is nothing like unto Him” (42:11); “God is the Light of the heavens and the earth” (24:35), the condition that makes seeing possible rather than an object within the visible. Creation is described as proceeding by precise measure, qadar (54:49).
This is closer to what information-theoretic approaches to physics describe than to the anthropomorphic figure that both literalists and their critics tend to assume. Wheeler’s “it from bit” proposes that physical reality is derived from information rather than the reverse. The apparent fine-tuning of the constants is at minimum a fact requiring explanation.
I keep this section short because it is the weakest limb, and stretching it would cost more credibility than it buys. The observer effect in quantum mechanics does not require consciousness, and decoherence accounts for measurement without invoking a mind. Fine-tuning has serious non-theistic responses, principally multiverse models and the observation-selection effect. Any argument that leans on the frontier of physics inherits the frontier’s instability, and the history of matching scripture to current physics is a history of embarrassment as the physics moves. The honest statement is a compatibility claim, not evidence: the Quranic description of God is consistent with where several branches of fundamental physics are pointing, and that is worth noting and nothing more.
10. What seven convergences buy you, and what they do not
Now the actual inferential question. Suppose Claim A holds across all seven areas. What follows?
The intuitive move is multiplicative. Each convergence is individually unlikely under chance, so their conjunction is very unlikely, so the design hypothesis wins. That reasoning has three holes, and I would rather open them myself.
The convergences are not independent. All seven describe the same underlying object: a system for sustaining cooperation in human groups. A text that solves the cooperation problem well will score on Axelrod, Ostrom, and Turchin more or less automatically, because those literatures are studying overlapping phenomena. The effective number of independent observations is smaller than seven. I do not know how much smaller, and I do not think anyone can quantify it without a formal model.
The selection problem is severe, and it is mine. I went looking for these matches knowing the answers. The Quran contains over six thousand verses. Any sufficiently large corpus of normative statements, searched by a motivated reader with a specific target, will yield matches. I have not systematically searched for prescriptions that formal analysis marks as suboptimal, which is precisely the search that would test the framework rather than decorate it. So that the search does not stay hypothetical, here is where it starts. The fixed inheritance shares of 4:11 to 12, evaluated against modern family structures and asset types. The provision in the 2:282 debt protocol that calls two women where one man would witness, evaluated against modern information environments. The severity of the hudud penalties, set against the graduated-sanctions logic Section 5 credits the text with elsewhere. And the verses a hostile reader maps to first defection against out-groups, 9:5 and 9:29, which strain the “nice” property of Section 3 unless the contextual readings restricting them to treaty-breaking belligerents carry the full load. I think some of these survive formal modelling and I am not confident all of them do. The framework gets credit for none of them until they are modelled and the results reported whichever way they land. Until then, Section 3 through Section 5 are a collection of confirmations, and confirmations are cheap.
There is a strong naturalistic rival, and it is not the one most apologetics attacks. The weak rival is coincidence, which is easy to dismiss. The strong rival is cultural evolution. On this account, religions are subject to selection: those whose norms sustained large-scale cooperation grew and persisted, those whose norms did not were outcompeted and vanished. We observe the survivors. Norenzayan’s work on moralising high gods, Henrich’s on cultural group selection, and Wilson’s on religion as an adaptive group-level system all develop this in detail, with evidence. Under that hypothesis, finding that a major surviving religion encodes game-theoretically sound cooperation norms is not surprising at all. It is the expected result. Any world in which we can ask this question is a world in which the surviving traditions encode workable cooperation solutions.
That rival explains Claim A completely. My framework does not currently beat it, and saying otherwise would be dishonest.
What might distinguish them is specificity and timing, and here I have to report a fact that weakens my own best discriminator. The Quran did not arrive into a normative vacuum. Interest prohibition already existed in Deuteronomy 23:19 to 20. Proportional retaliation is the lex talionis of Exodus 21, and the preference for remission over exaction has rabbinic and Christian precedent. Contract recording was standard Near Eastern legal practice long before the seventh century. The Quran itself says it confirms earlier revelation (5:48), so the tradition is untroubled by this, but my argument is not: a mechanism “arriving fully specified in a single generation” loses most of its force against cultural evolution if the specification was substantially inherited from norm systems that selection had already refined for centuries.
The discriminator survives only in narrowed form, at the joints where the Quranic version differs from its precursors. Deuteronomy prohibits interest within the in-group and permits it toward outsiders; the Quranic prohibition is general, and arrives paired with a mandated risk-sharing alternative and a verification protocol. The lex talionis states the cap; 2:194 states the cap and immediately couples it to restraint, with 42:40 marking forgiveness as the preferred branch, which is the generous strategy rather than merely the proportional one. Whether those deltas are the kind of refinement cultural evolution routinely produces in one step, or something that needs more explaining, is the actual open question, and it is a much narrower question than the one I was asking before I checked the precursors. That is where the research has to go. It has not gone there yet.
11. Falsification: what would sink this
A framework that cannot fail is not a framework. Here is what failure looks like.
| Test | What would refute the framework |
|---|---|
| Control scriptures | Run the identical protocol, blind and pre-registered, against the Bible, the Bhagavad Gita, the Analects, and the Stoics. If they score comparably, the finding is about religious and philosophical texts in general, not about this one. |
| Adversarial search for suboptimality | A systematic hunt for Quranic prescriptions that formal modelling marks as suboptimal for their environment. Section 10 names the starting candidates. A serious failure here damages Claim B directly. |
| Formal modelling | Specify the riba prohibition as an actual game: strategy spaces, payoff functions, equilibrium analysis under stated assumptions. If profit and loss sharing does not outperform interest-based contracting under reasonable parameters, the flagship case falls. |
| Out-of-sample prediction | Derive a prediction from the framework about a cooperation phenomenon not yet studied, then check it. Postdiction is cheap. Prediction is not. |
| Independence audit | Estimate how much the seven convergences actually overlap. If they collapse to two or three effectively independent observations, the cumulative case weakens sharply. |
The programme is then staged. Near term: formally model two or three prescriptions, riba first, against historical data. Medium term: the Ostrom mapping and the Axelrod comparison, each as a standalone piece, with the control-scripture protocol run properly and its negative results published. Long term: the attribute cascade and the physics compatibility claim, which need more mathematical maturity than I currently have.
The framework becomes strong not by accumulating further confirmations but by surviving attacks. I would rather it fail on a control-scripture test than persist as an unfalsifiable comfort.
12. What I have deliberately left out, and why
Three standard arguments are missing, and their absence is a choice.
The scientific miracle genre. The claims about embryology, mountain roots, and cosmology are absent because they are the weakest link in the popular case and they contaminate everything they touch. They depend on translation choices made after the science was known, they map ambiguous language onto specific findings retroactively, and they are the first thing an informed critic dismantles. The game-theoretic case does not need them and is materially stronger without them.
The illiteracy argument. The meaning of ummi is genuinely disputed, and an argument whose force depends on a contested lexical reading cannot bear weight.
The perfect preservation claim. The strong form, that no variants exist, does not survive contact with the manuscript evidence, including the Sanaa palimpsest, or with the recognised qiraat tradition. The defensible claim is narrower and still historically unusual: the consonantal skeleton is stable across manuscript traditions and geography to a degree that has no clear parallel among comparably old texts. I make the narrow claim and not the strong one.
What remains: the literary challenge (2:23), which is an aesthetic judgement and I will not pretend otherwise; the internal consistency claim (4:82), which is checkable and which the text itself invites you to attack; the Byzantine prediction (30:2 to 4); and the structural convergences that are the actual subject of this paper.
13. Limitations
Beyond the specific weaknesses flagged in place, four general ones.
I am arguing for a conclusion I already hold. Every methodological safeguard in this paper exists because I do not trust my own judgement about which convergences are real.
No formal model exists yet. Every claim in Sections 3 to 5 is structural correspondence argued in prose. Until at least one prescription is written as an actual game with a proved result, Claim B rests on analogy.
“Optimal” was doing too much work in my earlier notes and I have narrowed it throughout. Optimality in game theory is always relative to an environment, a strategy set, and a payoff function. Change any of the three and the ranking changes.
Section 8 is the section most likely to be read as ideology. The demographic and institutional data are real and independently sourced, but the interpretive frame is mine, and a reader could accept every number in it and reject the reading entirely.
14. Bottom line
A text from 7th century Arabia specifies a cooperation strategy that is nice, proportional, forgiving, transparent, and credibly deterrent, which is the region of strategy space that sixty years of formal research has repeatedly returned to. It prohibits precisely the risk-asymmetric contract that concentrates capital, and mandates the symmetric alternative, with a verification protocol attached. It states four or five of the eight design principles a Nobel laureate would identify in 2009 as necessary for durable common-pool governance, with the remainder supplied by the tradition that grew around it. It names hyperbolic discounting and social proof as errors twelve centuries before they were formalised. Its ritual schedule generates a small-world network topology. And it frames existence as a test whose information structure is exactly what a test capable of separating types would require.
That is a real pattern and it deserves a real explanation. Cultural evolution is the strongest one on offer, and this paper does not defeat it. What the pattern does not deserve is either of the two cheap responses: dismissal without examining the correspondences, or triumphalism that treats correspondence as proof.
The framework earns its keep by surviving attack, not by accumulating agreement. If the control-scripture test comes back showing the Analects scores as well, I will publish that too.
The Quran’s own invitation is empirical: “Travel through the land and observe how creation began” (29:20). This is what taking that seriously looks like when the terrain is formal rather than geographic. The work is now to attack it properly.
References
Axelrod, R. (1984). The Evolution of Cooperation. Basic Books.
Henrich, J. (2016). The Secret of Our Success: How Culture Is Driving Human Evolution, Domesticating Our Species, and Making Us Smarter. Princeton University Press.
Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
Kahneman, D., and Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263 to 291.
Norenzayan, A. (2013). Big Gods: How Religion Transformed Cooperation and Conflict. Princeton University Press.
Nowak, M., and Sigmund, K. (1992). Tit for tat in heterogeneous populations. Nature, 355, 250 to 253.
Nowak, M., and Sigmund, K. (1993). A strategy of win-stay, lose-shift that outperforms tit-for-tat in the Prisoner’s Dilemma game. Nature, 364, 56 to 58.
Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press.
Piketty, T. (2013). Capital in the Twenty-First Century. Harvard University Press.
Spence, M. (1973). Job market signaling. Quarterly Journal of Economics, 87(3), 355 to 374.
Taleb, N. N. (2012). Antifragile: Things That Gain from Disorder. Random House.
Turchin, P. (2016). Ages of Discord: A Structural-Demographic Analysis of American History. Beresta Books.
Watts, D. J., and Strogatz, S. H. (1998). Collective dynamics of small-world networks. Nature, 393, 440 to 442.
Wheeler, J. A. (1990). Information, physics, quantum: The search for links. In W. Zurek (Ed.), Complexity, Entropy, and the Physics of Information. Addison-Wesley.
Wilson, D. S. (2002). Darwin’s Cathedral: Evolution, Religion, and the Nature of Society. University of Chicago Press.
The framework was prompted by the geopolitical lectures of Jiang Xueqin, whose game-theoretic analysis of civilisational cycles repeatedly arrives at conclusions this paper examines from the other direction.