The Bounded Economy

An economy designed so that value cannot become capital

A closed complementary currency that fuses a floor, a ceiling, demurrage, an effort curve, and one-human-one-account identity — with agent-based simulation evidence, validated against the econophysics of inequality, and positioned against the systems that have actually run.

Status: working draft. Source material: the design spec spec.md, the prior-art survey prior-art.md, the deployment analysis deployment.md, and the simulation the simulation (results in results.md).

Abstract

Inequality is the attractor of even "fair" economic rules: neutral multiplicative exchange reliably condenses wealth into a Pareto tail. We ask the inverse question — can an economy be designed so that wealth provably cannot condense, while still rewarding effort and circulating freely? We present the Bounded Economy: a closed, non-convertible complementary currency that fuses five primitives in a single unit — an unconditional floor (a universal stipend), a hard per-human income ceiling (≈20× the floor), demurrage (idle value decays), an earning curve with diminishing returns whose remainder overflows to a commons, and one-human-one-account identity. The cap binds income (a flow): earned income saturates toward N·S·F while a held balance (a stock) is just income accumulated under demurrage (≈ income/δ). We give the mechanics, a governance layer (collectives defined by a small public constitution), and a currency-board membrane that lets the closed economy trade with the outside fiat world. An agent-based simulation — validated against closed forms from the conserved-exchange (Chakraborti–Chakrabarti) literature — shows the design self-stabilizes to a bounded, tail-free income distribution: there is no Pareto tail for any talent input (income Gini ≈ 0.19 at a modest talent spread, the cap clipping the tail as heterogeneity rises), across a broad parameter basin, and that its fragility is concentrated not in the economic dials but in three places: personhood, consumption-flow enforcement, and the membrane boundary. The design is not clean-sheet: it is best read as "Circles UBI plus a ceiling plus a balance-of-payments membrane," carrying the anti-capital principle independently reached by Participatory Economics, and stress-tested against the three failure modes that ended the kibbutz.

1. Introduction

1.1 The problem: condensation is the default

The canonical agent-based models of wealth — Bouchaud–Mézard and the "yard-sale" family — establish a sobering result: under neutral, multiplicative, individually-fair exchange, wealth condenses. A Pareto tail and eventual oligarchy are not a market failure to be corrected; they are the attractor of the dynamics. Real economies add compounding returns, inheritance, and luck, all of which push the same direction. The policy tools we reach for (progressive taxation, transfers) are corrections applied after the condensing engine has run.

1.2 The proposition: bound the engine, don't just tax its output

This paper asks the inverse question. Rather than redistribute the output of a condensing economy, can we design a monetary system whose attractor is already bounded — where the rich literally cannot accumulate without limit, value cannot detach from persons into freely-tradable capital, and yet effort is rewarded and money circulates? "Bounded," here, is used in its technical sense: a bounded economy is the well-behaved case; unbounded accumulation is the pathology.

The Bounded Economy is a closed complementary currency built from five fused primitives:

1. Floor — every verified person receives a universal monthly stipend. Dignity is unconditional.
2. Ceiling — a hard cap on per-person income at ≈20× the floor — itself the product of a work premium and a success ratio (C = N·S·F; §2.1) — approached asymptotically by an earning curve so it is a horizon, not a wall.
3. Demurrage — idle balances decay, so money is a medium of exchange, not a store of value; standing must be continuously re-earned. (A held balance therefore rests at ≈ income/δ — the savings buffer, a stock distinct from the capped income flow.)
4. Earning curve + overflow — earned income rises with diminishing returns toward the ceiling; the remainder overflows to a commons. This is a continuous, bracket-free marginal income tax with no tax authority.
5. Identity — one human, one account, making the cap bind per person rather than per wallet.

1.3 Contributions

• A five-way fusion that, to our knowledge, no running system carries in full (Section 5). The closest running system, Circles UBI, has three of the five; the closest theoretical model, Parecon, reaches the same anti-capital principle by a different route.
• Overflow as the back of an asymptotic earning curve — a marginal income tax that requires no brackets and no authority, funding a commons (Section 2).
• A currency-board membrane modeling the boundary between the closed economy and the open fiat world as a balance-of-payments interface, which we show exhibits classic currency-crisis dynamics (Dornbusch overshooting; self-fulfilling speculative runs) and the defenses against them (Section 4).
• Simulation evidence, validated against closed forms, that the design is robust across a broad parameter basin and that its fragility is isolated to identity, flow-enforcement, and the membrane (Section 4).
• An adoption path — a voluntary, complementary layer inside a host community — that routes around the coordination, power, and gravity constraints that defeated the precedents (Section 1.5).

1.4 Honest positioning

We do not claim clean-sheet originality. Every individual mechanism has prior art and most have running implementations. The contribution is the specific fusion and two novel mechanisms within it. The defensible one-line framing: Circles UBI + a hard ceiling + a balance-of-payments membrane, carrying Parecon's anti-capital principle, validated on the model class econophysics already studies, and stress-tested against the three points the kibbutz died of.

1.5 Feasibility was never the constraint — adoption was

A second clarification matters even more, because it is easy to mistake what a result like ours contributes. Bounded, egalitarian designs are an old and well-developed tradition — debt-jubilee, Plato's 4:1 ratio, Gesell's demurrage (run at Wörgl in 1932), the kibbutz, Participatory Economics, UBI, Georgism, limitarianism. People have always known a healthy design is possible. So this paper does not claim to have proved that; the historical record shows feasibility was rarely the binding constraint. Adoption was: Wörgl's working scrip was shut down by the central bank (incumbent power); the kibbutz eroded under the gravitational pull of the surrounding economy (brain drain, external debt); UBI is blocked by politics, not ignorance; Circles UBI works technically but lives or dies on local context (Berlin failed, Bali thrived).

Three forces defeat schemes that require replacing a whole economy: coordination failure (no one defects into the bounded world alone), incumbent power (its winners can block it), and gravity (it leaks its most productive people to the unbounded outside). The simulation result therefore plays a specific, limited role — a shield that makes this design un-dismissable on feasibility grounds — not a discovery. The genuinely new move is the adoption path: the Bounded Economy is a voluntary complementary layer inside a host community, not a replacement for the macro-economy, which routes around all three forces (you need only a sub-community to coordinate; you threaten no incumbent; the membrane lets the inside coexist with the outside). And the incentive to opt in must be a present, felt good — security and belonging, the ease of anguish — never abstract fairness, which loses to the lure of unbounded gain. The full argument is in adoption.md; the motivating application (Section 7) is built on exactly this — converting the uncertainty of retirement savings into the certainty of a guaranteed real life.

2. The model

2.1 The earning curve and overflow

Let the floor (monthly stipend) be F. The ceiling is not an arbitrary number but the product of two governable dials: a work premium N — the median worker earns ≈ N·F — and a success ratio S — the most successful are capped at S× the median. So the cap is C = N·S·F (baseline N = 5, S = 4 ⟹ C = 20F). N is the work-incentive dial; S is the inequality-tolerance dial; the headline "20×" is just their product, and decomposing it makes the two policy choices it bundles separately legible. A person's standing (worth) W grows with cumulative effort E with diminishing returns toward the cap:

dW/dE = k·(C − W)      ⟹      W(E) = C − (C − F)·e^(−k E)

W starts at F and approaches C asymptotically; the effort to climb the last stretch is effectively unbounded, so the cap is a horizon almost no one reaches (in simulation, <1% come within 10% of it). The same curve, read on each incoming payment rather than over a lifetime, is a keep rate: the share of a new unit of value a person retains is (room left to the cap) raised to a power — ≈100% at the floor, tapering to 0% at the cap. What is not kept is the overflow, which does not vanish — it flows to a commons. Near the floor almost nothing overflows; near the ceiling almost all of it does. The curve is a continuous, progressive, bracket-free wealth tax, and the keep rate at the median worker is the single most legible way to set its steepness.

2.2 Money supply and demurrage

Idle worth decays at rate δ. Demurrage does two jobs. First, it makes money a medium, not a store — to hold a high position you must keep contributing, so standing reflects recent contribution, not accumulated past. Second, it is the supply regulator: issuance (n·F per month, for n members) is balanced by decay, so the money supply self-stabilizes near n·F/δ with no central-bank discretion. (Throughout, uppercase N is the work-premium dial of §2.1; lowercase n is the head-count — they are different quantities.) Crucially, demurrage burns (controls supply) while overflow redistributes (funds the commons) — distinct jobs, so the two anti-accumulation mechanisms are complementary rather than redundant.

2.3 Why the fusion is necessary

Each primitive patches a hole the others would leave. A floor without a ceiling is UBI inside ordinary capitalism. A ceiling without identity is defeated by Sybil wallets. Demurrage alone (Gesell) makes money a poor store of value but lets wealth flee into land or durable assets — which is why Gesell paired Freigeld with Freiland (land reform), and why this design extends non-transferability to all stores of value. An effort curve without overflow throws away the value the most productive activity generates. The pieces are individually old and collectively unprecedented because their natural constituencies reject one another (UBI advocates dislike ceilings; sound-money advocates loathe demurrage). As a designed unit they are mutually reinforcing.

3. Design

3.1 The oracle is a bounded market

Who decides what effort is worth? Not a committee — people pay each other, and the limits live in the transfer rules. This is the only oracle that creates no central power center. It yields an emergent property: because the rich cannot over-accumulate and demurrage decays their position, demand routes toward providers who still have headroom. The ceiling does not merely bound wealth; it continuously redistributes opportunity.

3.2 Collectives and governance

Groups (economic, social, political) are defined by a small, public constitution: an admission rule, a decision rule, a distribution rule (with vesting), and clawback triggers — legible before anyone joins, because in a system premised on simple rules resisting capture, elaborate bylaws are where hidden power lives. Three principles govern them: influence is bounded, not flat (proportional voting is safe because the cap makes it bounded plutocracy, ≤20:1); partition, don't multiply (a person's finite capacity, effort, and voice are split across memberships, never duplicated); and delayed, clawback-able distribution keeps decision-makers exposed to the long-run consequences of their votes. The whole economy's constitution is itself just a political collective whose monetary dials move by published formula behind a supermajority-plus-delay.

3.3 The fiat membrane

A closed economy still needs external goods it cannot produce. The membrane is the balance-of-payments interface to the fiat world, and the design's hardest part — whoever runs it could become the sovereign the system abolished. The resolution is to shrink it to a rule-bound utility: a currency board that mints coin when exports earn fiat and burns it when imports spend fiat, owning nothing and clearing everything; rule-priced, plural, and demand-driven. Capital stays strictly external (investors are serviced in fiat from external revenue, never touching the coin), so the bounded economy never has to pay capitalist returns in an anti-accumulation currency.

3.4 Time-value, inheritance, and credit — the anti-capital core

Time-value of money cannot be abolished, only relocated: suppress it on the coin (demurrage) and it flees to durable assets, private credit, or the reserve. The design's answer is to collectivize time-value — the reserve captures it for the commons, the overflow waterfall redistributes it to the floor — and to block every route by which value becomes transferable, accumulable capital: accounts bound to persons, use-rights reverting to the commons at death, voice partitioned not sold, and credit claims made non-transferable (which is also the only thing that genuinely caps the return on lending — a tradable claim sold at a discount reconstructs interest regardless of any rate cap). Inheritance is permitted but impotent: the cap, demurrage, and overflow render it self-extinguishing — a head start, never a dynasty (Section 4.6).

3.5 A flow economy, not a capital-formation economy

Demurrage makes hoarding lumpy capital impossible by design. Large projects are funded in three tiers: broad collectives (scale from membership breadth, not concentration), the commons/reserve (public goods), and — for genuinely large assets — external fiat capital that amortizes into commons ownership (the community-land-trust pattern). The internal economy distributes flow fairly and delegates lumpy capital to the external layer.

This points to the design's deepest structure: money's three conflicting jobs are best separated rather than forced onto one coin. People earn a single legible coin that splits, paycheck-style, into a spending medium (demurrages, cannot accumulate), a bounded security store (non-transferable, reverting — the legitimate home of saving, where "a dignified life" is saveable but "a transferable hoard" is impossible), and recognition (non-money standing). Governance then composes from the two non-spendable axes — committed stake and earned contribution — applied to different decision domains, so spending money can never buy power. A first engine for this (twopart.py) confirms accumulation is blocked on every axis at once and that the stake and contribution axes give bounded-but-distinct governance weightings; the full treatment is in the spec ("Separating money's three jobs"). This is the structural heart of the anti-capital claim, and the most important design extension beyond the single-coin model validated here.

4. Simulation and results

We built an agent-based simulation (sim.py) of the internal coin economy and stressed it; advanced dynamics (lifecycle, multi-collective, membrane attack) are in advanced.py. Full numbers and figures are in results.md.

4.1 Method and validation

The model belongs to the conserved-exchange / econophysics family (Chakraborti–Chakrabarti tradition), giving it closed forms to test against. validate.py checks four layers: the Gini estimator (correct to <0.0013 against analytic distributions); a true CC pairwise model (reproduces the econophysics closed form G(n), n = 1 + 3λ/(1−λ) to <0.004); and the engine's own fixed point (bounds off, wealth converges to exactly proportional to productivity, to machine precision — so the unbounded baseline Gini is the lognormal-productivity value, not the CC value), and the income-capped mean-field, whose closed form reproduces the simulation agent-by-agent (corr 0.99887 — exact below the security cap, §6). No modeling error. The cap binds income, and its job is to remove the tail: with a heavy-tailed (Pareto) talent input the unbounded income inherits the power-law tail while the bounded income is clipped at the ceiling (no Pareto tail for any input, §6). At a modest talent spread the income Gini sits at ≈0.19 (vs. ≈0.32 unbounded); the gap widens as talent heterogeneity rises and the cap bites. The simulation is implicitly stock-flow-consistent: money is conserved except explicit minting and burning.

4.2 Calibration — what each dial does (dials.png)

• The cap = N·S is two dials, not one: the work premium N (median ≈ N·F — the work-incentive dial) and the success ratio S (top capped at S× the median — the inequality-tolerance dial). Raising the cap sets the ceiling of spread but leaves the average unchanged — it redistributes position, not the pie.
• Keep rate (the earning curve, exponent α under the hood) is a pure equality dial — set as the share the median worker retains of each new unit (the rest overflowing to the commons); a lower keep rate tightens the distribution without touching supply. Pinning it to the median keeps its meaning fixed as N and S move.
• Demurrage δ is the supply / savings-buffer regulator (supply ≈ n·F/δ for n members; diverges as δ→0), with a non-obvious U-shaped effect on inequality: too little and the demand pool swells until everyone's income saturates at the cap; too much and the economy shrinks. δ also sets the wealth level (a held balance ≈ income/δ) — it does not set the income ceiling, which is the cap.
• Talent spread σ is the input inequality, and the cap clips its tail at any heterogeneity: a heavy spread (σ=1.4) that would give an unbounded income Gini ~0.66 with a Pareto-like tail is bounded to a tail-free income distribution (Gini ~0.30, top clipped at the cap). The system removes condensation regardless of the input — the thesis in one result — while the bulk inequality still tracks talent.

4.3 Robustness — a broad basin, not a knife-edge (stability.png, distribution.png)

Baseline income Gini is ≈0.19 (held-balance Gini ≈0.20), seed-stable, with supply identical every run — not artifacts. The δ × cap stability map (income Gini) shows a broad healthy basin: the only failure mode is incomes pinned at the cap, confined to weak demurrage (δ ≤ 0.03) and a very tight cap, and it fails gracefully (incomes equal at the ceiling, not a crash). The design is forgiving on its dials; tuning is not where the risk lives. The headline distribution.png shows, at a heavy talent spread (σ=1.0), the unbounded income tail (Gini 0.51, reaching far past the cap) clipped at the ceiling into a bounded income distribution (Gini 0.27).

What the cap is for — tail-insurance, not flattening. At a modest talent spread the bounded and unbounded income distributions nearly overlap (the cap is then a slack horizon). This is a feature, not a failure: the design does not flatten the normal professional income range; it removes the pathologies the default cannot escape. Four of these are invisible in a single income snapshot — (1) the attractor over time (the default condenses into a Pareto tail; this design cannot, §6); (2) the floor (no destitution); (3) what wealth can become — demurrage + non-transferable use-rights + the per-person cap forbid value becoming hereditary capital; and (4) the hard ceiling that makes oligarchic capture structurally impossible. The default can only promise to tax the tail after it forms; the Bounded Economy cannot form the tail in the first place.

4.4 Adversarial — fragility is in identity and flow (sybil.png)

• Sybil/identity is load-bearing. Any identity leakage breaks the per-human cap immediately, and inequality is worst at partial penetration — 50% capture yields a Gini of 0.505, worse than the US — because the realistic failure (some game identity, most don't) mixes fat cheaters with thin honest players.
• Coercion-smurfing defeats the holding cap entirely: a boss renting 100 floor-dwellers commands 101× the mean consumption while never holding more than the cap. Only a per-person consumption (flow) cap defends. It is load-bearing, not optional.

4.5 Governance and membrane stress (attack.png)

• Collective capture. Against entryists, a supermajority threshold alone is a false comfort — if breached it makes a raid worse for loyal members (more attackers needed → harder dilution). The defenses that work are delay + exit (a winning vote nets a drained shell as loyal members leave with vested stakes) and tenure-weighting (a fast raid is impossible). Open admission is safe only with the full stack.
• The membrane survives an export shock only with active management. A floating clearing rate auto-stabilizes the reserve by price (rule-based, no administrator) but overshoots — a Dornbusch-style ~80% devaluation. And strategic importers front-running the devaluation can convert a survivable shock into a self-fulfilling run (the classic currency crisis); an adequate buffer or a circuit breaker each prevents it. The membrane must be designed against runs, not just shocks: a managed float + buffer + breaker.

4.6 Generational dynamics (lifecycle.png)

Tagging the founding rich and running births/deaths across generations: with use-rights reverting to the commons at death, a founder's advantage is purely earned within their own life and resets completely at death. Even 100% inheritance decays to a ~1.05× premium (and 50% ≈ 100%), because the cap and demurrage bound what can transmit regardless of the inheritance rate. Inheritance is permitted but impotent — the generational reset is structural, not legislated.

5. Prior art and positioning

The design is a fusion that no other model occupies. Scoring seven defining properties across the closest systems, the Bounded Economy is the only row that is full-house:

• Circles UBI is the closest running cousin (floor + demurrage + personhood, same equalization goal). Its Berlin-vs-Bali failure pattern — it worked only where a deep internal catalog of real goods existed — is direct field validation of our "a closed currency is worthless unless it buys things people want."
• Parecon is the theoretical twin: effort-based, non-transferable, non-accumulable credits — the anti-capital principle reached independently, via central planning rather than a bounded market. Convergent design is a strength, not a derivation.
• The kibbutz is the empirical precedent: closure and equality, but it lacked an effort-market (free-rider strain), lacked demurrage, and carried external debt (the 1980s bankruptcies). The Bounded Economy is the kibbutz with the three things it died of patched. Abramitzky's econometrics independently confirm our three flagged failure modes — exit/brain-drain, capital-account fragility, generational consent.
• Sardex proves a closed, zero-interest mutual-credit circuit works at regional scale — empirical validation of the membrane/credit layer.
• Georgism + social-wealth fund is the policy analog: its land-value tax is the real-world Freiland the design invokes.

What is genuinely unprecedented: (1) the full five-way fusion; (2) overflow as the back of an asymptotic curve = a bracket-free marginal income tax; (3) a currency-board membrane with a floating clearing rate on a complementary currency, exhibiting balance-of-payments dynamics.

6. Limitations and open questions

6.1 Scope and coverage

The Bounded Economy models the monetary and distributional layer of an economy — in the kinetic-exchange / stock-flow-consistent tradition — not a general or growth model. Stated plainly:

Two kinds of omission must not be conflated. Some are the thesis, not gaps: no internal capital market, no accumulation, bounded zero-interest credit — excluded by design, because the system exists to forbid value becoming capital. The rest are genuine boundaries, and the chief one is this: productivity is an input, not an output. The model assumes each agent's productivity and derives how the resulting money distributes; it does not show whether bounded, demurraged incentives sustain that productivity — effort, output, innovation. The exit/retention result answers one half (whether skilled people stay) but not whether capped reward elicits effort in the first place. The model's claims are therefore monetary-distributional — "wealth cannot condense; the system is monetarily stable" — never claims about productivity, efficiency, or growth. Extending the model toward an endogenous production side is the most important open direction (§6.3).

A second axis — whether value can simply flee the coin bound into goods held as a store of value — is now modeled (Stage 2, engine goods.py): it can, severely (max total wealth runs to ~75× the cap via a non-depreciating good), unless scarce durable assets are held as use-rights — demurrage alone only dampens the leak proportionally, never closing it. This validates the design's use-rights / Freiland choice as load-bearing rather than decorative. The remaining genuine gap on this axis is endogenous price formation (and the asset-bubble dynamic) — the one place the design still hand-waves allocation.

6.2 Where the fragility lives

The simulation is a mechanism-isolation model, not a forecasting one — no firms, banks, business cycle, or expectations; it cannot speak to growth or employment. Absolute Gini values are soft (no compounding, luck, or inheritance in the base model) — the standard, correct caveat for this model class, which understates real inequality. Directional findings are robust.

The genuinely hard, partly-unsolved problems are exactly where the simulation says the fragility lives:

• Personhood. A Sybil-resistant, privacy-preserving, one-human-one-account system is the field's hardest open problem; the entire egalitarian property rests on it. A permissionless public-blockchain version is not viable (it cannot deliver personhood or enclosure); the design needs a bounded community with an identity authority — which is why the deployment substrate is a permissioned ledger, not an open coin.
• The membrane boundary — runs, overshooting, and a permanent (not temporary) export collapse, which forces lasting austerity.
• The exit constraint. The system holds only if it is voluntarily more attractive than the outside for the people it most needs — empirically the binding constraint on the kibbutz. Measured directly (engine exit.py): against an uncapped, winner-take-all outside option (external ∝ talent^γ), retention is bought by the community premium h — the pay cut a member willingly takes to stay (security, community, meaning) — not by the pay ceiling. A member leaves only when the outside still wins after the haircut, external·(1−h) > internal. Phase maps over (h, γ) and over (cap, h) make the asymmetry visual: the most productive decile is held by a high premium or a gentle outside, and never by loosening the cap — at a winner-take-all γ=2 it takes h ≳ 60% to keep the top tier, while the broad middle is retained at a modest premium (results.md §8, maps.png). The retention lever is desirability, not pay.

Open threads include the Circles Berlin-vs-Bali post-mortem (catalog depth), Sardex's clearing mechanics, the Kolkata-index benchmark, and whether the asymptotic ceiling answers limitarianism's objection that a single fixed cap is untenable.

6.3 Toward an endogenous production side

Making productivity an output rather than an input is the model's most important extension, and the existing machinery already contains the ingredients. The earning curve W(E) = C − (C−F)·e^(−kE) is, read differently, a marginal-reward schedule: dW/dE = k(C−W) falls toward zero as standing nears the cap. So the natural step is to let each agent choose effort e to maximize reward minus a convex effort cost — turning assumed productivity into elicited output. A staged path:

1. Endogenous effort (highest value, tractable). Each agent supplies effort until marginal bounded reward equals marginal effort cost. This directly answers "does the bounded incentive sustain effort?" Two forces pull opposite ways: the cap, demurrage, and overflow dampen top effort (a high earner's marginal reward is small), while the floor reduces ruin-risk and may broaden effort (more people can afford to contribute and to try). The central, testable hypothesis: the bounded economy reshapes the effort/output distribution — flatter at the top, broader at the base — rather than necessarily lowering the aggregate. Whether the total rises or falls vs. an unbounded economy is genuinely open, and is exactly what this extension would measure.
2. A production function + the internal catalog. Let output q = A·p·e be the real goods the coin buys, closing the effort → output → consumption loop and letting the model report output per capita and whether supply meets the floor's demand (the Circles "deep enough catalog" question, endogenized).
3. Innovation and growth — the risk-taking tension. Capped upside dampens the lottery-ticket motive for moonshots; the guaranteed floor lowers the cost of failure and may raise broad entrepreneurship (downside insurance is empirically associated with more business formation). Modeling technology A that grows with R&D effort would let the design finally speak to growth — the hardest, most-demanded question. Honest hypothesis: fewer moonshots, more people daring to try — net effect unknown.
4. Prices, allocation, general equilibrium. Endogenous goods prices and market-clearing would let the model speak to allocative efficiency. Most ambitious; likely a separate paper.

The literature cuts both ways and should be engaged: optimal-taxation work (Mirrlees; Saez's elasticity of taxable income) on how little top effort sometimes responds to high effective marginal rates; the entrepreneurship-and-safety-net evidence that downside insurance increases risk-taking; and the macro finding (Ostry et al.) that redistribution is not generally growth-harming. Stage 1 alone would move the productivity claim from assumed to derived — the single highest-value next piece of modeling.

7. Deployment: a system you choose into

The design is voluntary and commons-oriented by nature, so it should be released as something people opt into, not a token to speculate on. The substrate is a permissioned ledger (a plain database suffices at first; a chain earns its keep only for credible commitment, transparency, and cross-node portability). It splits into two independently testable halves: a reward-and-governance half best piloted in an open-source software community (visible contributions, fork-as-exit culture, a real funding pool — using peer allocation, never algorithmic scoring, the lesson of SourceCred's failure), and a closed-provisioning half best piloted in a services community. The framing must be explicitly non-speculative — non-transferable, non-convertible, no public token, no fundraise — or the serious audience dismisses it and the wrong audience arrives.

A motivating application is any community that provisions real things its members need, where the bounded economy appears as a contribution-funded mutual-aid commons. There the deepest value proposition is concrete: it converts the uncertainty of an individual's savings into the certainty of a guaranteed real claim on what the community provides, with the collective bearing risk an individual cannot — a defined-benefit promise denominated in quality-of-life rather than dollars.

8. Conclusion

Inequality is the attractor of fair rules; this paper shows, in simulation and against the relevant literature, that an economy can instead be designed with a bounded attractor — rewarding effort, circulating freely, and structurally forbidding value from becoming capital. The economic mechanics are robust and forgiving across a broad parameter basin. The hard problems are not the dials but the boundaries — personhood, flow-enforcement, and the membrane — which is precisely where the closest historical precedent broke, and precisely where engineering and governance effort should concentrate. The honest claim is not utopia but a buildable, testable, community-scale complementary currency: Circles UBI with a ceiling and a membrane, carrying Parecon's anti-capital principle.

References and sources

Grounding and citations are collected in prior-art.md. Key sources: Bouchaud & Mézard (wealth condensation); the yard-sale/oligarchy model; Chakraborti & Chakrabarti (kinetic exchange) and the Gini/Kolkata saving-propensity literature; Gesell, The Natural Economic Order (demurrage); Circles UBI whitepaper and its Frontiers in Blockchain (2024) and FRIBIS (2025) evaluations; Albert & Hahnel (Participatory Economics); Dini et al., LSE (2016) on Sardex; Robeyns, Why Limitarianism? (2022); Abramitzky, The Limits of Equality / The Mystery of the Kibbutz; Caiani–Godin–Stiglitz (stock-flow-consistent macro). Simulation, validation, and figures: the simulation.