Trustless systems are infinite games—designed to run forever without capture. However, nearly every funding mechanism we have is intended for finite games: ventures with clear endpoints, exits, and returns. Venture capital needs liquidity within a decade. Token launches front-load extraction. Grants create dependency. The result: systems that start decentralized and drift toward concentration, or never decentralize at all.

This essay argues the mismatch isn't incidental—it's structural. The question isn't "who funds the trustless web?" The question is "what does tending look like?"

In September 2025, Nepal's government banned 26 social media platforms. Within days, over 100,000 protesters coordinated through Discord servers, using VPNs to bypass the ban. They organized demonstrations and coordinated strategy—all without central leadership. On September 9, Prime Minister Oli resigned.

No headquarters. No organizing committee. No institution. Just people aligning their behavior toward an outcome none could achieve alone.

That's coordination. And for most of human history, it required institutions to pull off at scale.

Institutions are durable structures that encode rules, concentrate enforcement capacity, and create shared expectations. They solve coordination problems by centralizing decision-making authority—someone has to decide which side of the road we drive on.

This centralization has a cost. Whoever controls the institution shapes its rules. They decide who gets included, what questions get asked, and which outcomes are legitimate. You get coordination, but you get capture too.

The tradeoff has always seemed unavoidable. Want coordination at scale? Accept institutional gatekeeping. Want freedom from gatekeepers? Accept smaller-scale coordination.

Nepal suggests the tradeoff may not be fixed.

Coordination has costs: communication, verification, enforcement, and trust-building. For most of history, these costs scaled with participants. Coordinating ten million people required a massive bureaucratic infrastructure.

The internet collapsed communication costs. This enabled new forms of coordination—open source software, Wikipedia, and social movements organized through hashtags. But communication isn't enough. You also need verification (is this person who they claim to be?), enforcement (will they do what they promised?), and trust (if things go wrong, is there recourse?). The internet left these to old institutions.

This is why the first two decades produced the "platform paradox." Technology promised decentralization; the outcome was unprecedented concentration. Facebook, Google, Amazon—each became a new institution, solving coordination problems the old way: by centralizing control.

Blockchain technology collapses verification and enforcement costs. Cryptographic proofs replace institutional trust. Smart contracts replace courts. Distributed consensus replaces central authorities. The cost of coordinating ten million people approaches the cost of coordinating ten.

Stablecoins moved $27.6 trillion in 2024—more than Visa and Mastercard combined. Not because they're better payment rails. Because they're coordination rails that don't require permission from institutions that control traditional finance.

The coordination monopoly is breaking. The question is what replaces it.

James Carse, in Finite and Infinite Games (1986), distinguished two types of play. Finite games are played to win—defined players, agreed rules, clear endpoints. When someone wins, the game is over.

Infinite games are played to continue playing. Players may change, rules may evolve, but the game goes on. There is no winning an infinite game—there is only continuing it or failing to.

The distinction runs deeper than it first appears. Finite games require boundaries: temporal, spatial, and membership. Infinite games resist boundaries. The moment you define them too precisely, you've converted them into finite games.

Building a trustless system is committing to an infinite game. You're not building toward an exit. You're building infrastructure designed to outlast its creators. The Ethereum blockchain is meant to run for decades, perhaps centuries. Bitcoin's core promise—a store of value independent of any government—requires indefinite continuation.

The Ethereum Foundation calls this "the Infinite Garden"—a metaphor drawn from Carse. Gardens require ongoing attention. They don't reach a final state. Tending is never complete.

But nearly every funding mechanism is designed for finite games.

Venture capital has a fund lifecycle. Limited partners expect returns within 7-10 years. This creates pressure for exits—acquisitions, IPOs, token launches—that convert infinite game systems into finite game outcomes. The VC doesn't want the garden to thrive forever; the VC wants to sell their stake at a higher price.

Token launches front-load value extraction. Early holders expect appreciation. This creates incentives for hype cycles, artificial scarcity, and governance that serves token value over system sustainability.

Foundation grants are finite by definition. When the grant ends, the work either finds new funding or stops. Corporate sponsorship ties development to sponsor interests that change.

The BIS documented this as the "decentralization illusion"—the tendency of blockchain systems to concentrate power despite decentralized rhetoric. Tim O'Reilly called blockchain "the most rapid recentralization of a decentralized technology I've seen in my lifetime."

Not because anyone intends it. Because finite game funding creates finite game incentives, and those incentives select for centralization.

You can't bootstrap an infinite game with infinite game resources. They don't exist yet. So you take VC money to build the thing that shouldn't need VC money. You accept the grant to create what should outlast foundations. You launch the token to fund the protocol that's supposed to resist capture.

This isn't hypocrisy. This is the trap. The question is whether you're building toward escape velocity or entrenching the dependency that got you started.

The mental model most projects operate under: raise capital, build product, achieve adoption, extract value. This is the startup playbook. It works brilliantly for finite games.

Applied to infinite games, the result is predictable. Early backers need liquidity. Governance concentrates on enabling faster decisions. Competitive pressure demands efficiency. Before long, the "decentralized" system looks like the centralized systems it was supposed to replace. This is the Web2.5 equilibrium—decentralized in rhetoric, centralized in practice.

Consider the language: "Funding." "Investment." "Returns." "Runway." This assumes finite game logic. Money in, more money out. Success is measured in multiples. Failure is measured in lost capital.

But infinite games don't have returns in this sense. The "return" on maintaining critical infrastructure is that the infrastructure continues to exist. The "return" on contributing to a protocol is that the protocol remains viable. No IRR. No exit multiple. Just continuation.

This doesn't compute in finite game accounting.

The mental model gap shows up everywhere. Foundations sit on treasuries, unsure whether to spend aggressively or preserve indefinitely. Projects promise progressive decentralization while their funding structures incentivize centralized control. Builders talk about "community ownership" while their tokens create class distinctions between early and late participants.

The old mental models aren't wrong. They're incomplete. They capture finite game dynamics perfectly. They miss infinite game dynamics entirely.

Sustaining an infinite game requires solving three problems simultaneously.

The resource problem: Where do resources come from? Not one-time fundraising but an ongoing flow. Not a single patron, but a distributed contribution. Not one sprinkler watering the garden, but rain from the entire sky.

The legitimacy problem: Why would anyone contribute? Contribution has to mean something—confer status, express identity, create belonging. Open source works partly because contributing to Linux signals competence and values. You become part of a community that matters.

The hardest shift isn't technical. It's conceptual: contributing to shared infrastructure as normal as paying rent, as meaningful as tending a garden.

The governance problem: How do decisions get made without creating concentration? Someone has to decide what gets built, how resources flow, and when rules change. Trustless systems can automate enforcement, but they can't automate judgment. Humans have to decide what rules should be before the protocol can enforce them.

If that power concentrates, the system isn't trustless anymore—it's just trust with extra steps.

Traditional institutions solve these through multiple mechanisms: tithing and endowments for churches, tuition and donations for universities, and taxation for states. They create a continuous flow. They make contributions meaningful through status and identity. They have governance that prevents permanent concentration.

Trustless systems need equivalent structures—without the concentrations of authority those institutions rely on.

If sustaining infinite games were impossible, we would have no examples. But we do.

The Catholic Church has persisted for two thousand years. Tithing creates continuous flow built into the practice of faith—not occasional fundraising but ongoing contribution regardless of any individual donor's enthusiasm. Endowments preserve principal while spending yield; time becomes an ally. Legitimacy as sacred duty weaves contribution into identity—giving wasn't philanthropy but obligation. Distributed infrastructure—parishes, dioceses, monasteries, each semi-autonomous—eliminates single points of failure.

Universities show a similar pattern. Oxford has persisted for over nine hundred years, teaching since 1096. Harvard was founded in 1636. Both predate the governments under which they now operate.

The endowment model creates perpetual funding from finite gifts—a donor gives once; the university benefits forever. Legitimacy ties contribution to status and meaning. Regeneration through tuition, grants, and donations creates ongoing revenue beyond the original endowment. Governance structures—boards, faculties, traditions—constrain any individual's control.

Nation-states sustain coordination at a massive scale. Taxation creates continuous distributed extraction—every transaction, every paycheck generates revenue automatically. Citizenship ties contribute to identity. Services create reciprocity.

What do these institutions share? Multiple resource streams. Continuous flow rather than discrete injection. Legitimacy makes a contribution meaningful. Regeneration. Governance preventing permanent concentration.

These aren't trustless systems. Churches require faith. Universities require credentials. States require a monopoly on violence. But they demonstrate that infinite games can be sustained.

Open source is the closest analogy. Linux has persisted for over thirty years, first released in September 1991. The infrastructure the internet depends on, maintained without central control.

But open source is a mixed case. Linux persists partly on volunteer labor, partly on corporate support from companies that depend on it. Nadia Eghbal documented the crisis: critical infrastructure maintained by exhausted volunteers who can't get paid without getting captured. Projects die or get acquired. The successes achieve corporate alignment or develop cult-like communities—neither transfers cleanly to trustless systems.

The historical pattern is clear: infinite games can be sustained. The mechanisms are known. The question is whether they can adapt to systems designed to resist concentrations of authority.

We cannot prescribe solutions. Anyone claiming to have solved the Infinite Game Problem is selling something. But we can identify constraints that any solution must satisfy.

Endowment. Preserve principal, spend yield. This is how universities achieve perpetuity from finite gifts. The corpus is sacred; only returns are spent. A $1 billion endowment at 5% annual return generates $50 million per year, forever.

The Ethereum Foundation holds reserves. But reserves are runway—they deplete. An endowment requires accepting that spending is constrained by returns, not ambitions. No existing crypto foundation operates with true endowment discipline.

Tithing. Continuous small contributions from many sources. Protocol Guild—projects pledge 1% of tokens to core developers—is the most promising experiment. Not charity but norm. The goal: contribute as automatically as tithing was for medieval Christians.

Early results are encouraging. Whether the norm spreads beyond early adopters remains uncertain. The test isn't whether committed projects contribute—it's whether uncommitted projects feel pressure to join.

Legitimacy. Make contribution high-status, visible, meaningful. This is cultural work, not mechanism design.

Open source succeeded partly because contributing to Linux meant something—it signaled competence, values, membership in a community that mattered. On-chain attestations and reputation systems attempt to make contributions legible. But legitimacy can't be engineered. It emerges from a genuine belief that contribution matters.

Regeneration. The system must generate some of what it needs. Transaction fees fund validators. Protocol revenue can fund development. But alignment between the value generated and the value directed to maintenance remains imperfect. Better mechanisms would tie protocol success more directly to infrastructure funding.

Distribution. Multiple funders, multiple decision-makers. Ethereum's multi-client ecosystem—five execution clients, four consensus clients—is expensive and inefficient. It's also why no single team's failure can compromise the network.

The same applies to funding: many small donors create resilience that one large donor cannot. Gitcoin's quadratic funding amplifies small contributions precisely to achieve this—a dollar from a thousand people creates more matched funding than a thousand dollars from one person.

Sunset. Temporary centralization must have a credible path to dissolution. Every project begins centralized—founders make decisions, early teams hold tokens. The question is whether centralization is a scaffolding or a foundation.

Most promise to decentralize "later." Later rarely arrives. When it's time to transfer power, those holding it discover reasons to delay. What mechanisms force dissolution when those in power resist? Time-locked contracts? Triggered governance? The problem is recursive: who enforces the sunset on the enforcers?

This principle remains largely unsolved.

No existing structure satisfies all six. Protocol Guild addresses tithing but not endowment. RetroPGF addresses legitimacy but struggles with sustainability. The Ethereum Foundation practices distribution but lacks sunset clarity.

Each experiment satisfies some constraints while struggling with others. This is not failure. This is the work.

Five questions that require experiments, not essays.

Can legitimacy scale? It works in small communities where reputation is legible. Everyone knows who contributes, who defects. Can digital systems create social density that makes reputation meaningful at scale? Legibility isn't legitimacy. Legitimacy requires that contribution matters—that people care about the status it confers. Unknown.

What is minimum viable centralization? Every infinite game starts as finite. Someone writes the first code, makes the first decisions. Too much centralization, and founders never let go. Too little and the project never coheres. No framework exists for finding this point.

How do you make sunset clauses credible? Whitepapers are full of progressive decentralization roadmaps. Phase 3 rarely arrives. Technical enforcement faces a recursive problem: who enforces the sunset on the enforcers?

It may require social enforcement—communities refusing legitimacy to projects that miss transitions. But this returns us to the legitimacy question.

Can infinite game structures compete during transition? Finite structures are faster, concentrate resources, attract talent with equity, and make decisions quickly. If infinite structures can't survive bootstrapping competition, their long-term resilience never matters.

Bitcoin suggests it's possible. But Bitcoin had a first-mover advantage and a decade of obscurity to mature. Can new infinite structures compete against well-funded finite competitors in a crowded market?

What does success look like? Finite games have clear metrics: market share, returns, and exit value. Infinite games have continuation—but of what, at what scale, with what properties?

A protocol could continue indefinitely with three users and one maintainer. That's not success. We lack measures for operational continuity, contributor diversity, and regeneration rate. Without measurement, we can't distinguish structures that work from structures that haven't failed yet.

These questions will be answered by builders over decades. The questions themselves are the contribution. Asking what tending looks like is the first step toward learning how to tend.

The finite/infinite game distinction comes from James Carse, Finite and Infinite Games (Free Press, 1986). The Ethereum Foundation's "Infinite Garden" framing draws directly from Carse.

The Trustless Manifesto was published on November 11, 2025, by Vitalik Buterin, Yoav Weiss, and Marissa Posner. Protocol Guild's 1% pledge model was formalized in January 2024.

Nadia Eghbal's Working in Public (Stripe Press, 2020) and Roads and Bridges (Ford Foundation, 2016) document the open source maintenance crisis.

Stablecoin transfer volumes from CEX.IO research ($27.6 trillion in 2024). Nepal protest coordination from multiple news sources; Prime Minister Oli resigned on September 9, 2025.

Linux's first release: September 1991. Oxford has been teaching since 1096. Harvard was founded in 1636.

Trustless systems are infinite games—designed to run forever without capture. However, nearly every funding mechanism we have is intended for finite games: ventures with clear endpoints, exits, and returns. Venture capital needs liquidity within a decade. Token launches front-load extraction. Grants create dependency. The result: systems that start decentralized and drift toward concentration, or never decentralize at all.

This essay argues the mismatch isn't incidental—it's structural. The question isn't "who funds the trustless web?" The question is "what does tending look like?"

In September 2025, Nepal's government banned 26 social media platforms. Within days, over 100,000 protesters coordinated through Discord servers, using VPNs to bypass the ban. They organized demonstrations and coordinated strategy—all without central leadership. On September 9, Prime Minister Oli resigned.

No headquarters. No organizing committee. No institution. Just people aligning their behavior toward an outcome none could achieve alone.

That's coordination. And for most of human history, it required institutions to pull off at scale.

Institutions are durable structures that encode rules, concentrate enforcement capacity, and create shared expectations. They solve coordination problems by centralizing decision-making authority—someone has to decide which side of the road we drive on.

This centralization has a cost. Whoever controls the institution shapes its rules. They decide who gets included, what questions get asked, and which outcomes are legitimate. You get coordination, but you get capture too.

The tradeoff has always seemed unavoidable. Want coordination at scale? Accept institutional gatekeeping. Want freedom from gatekeepers? Accept smaller-scale coordination.

Nepal suggests the tradeoff may not be fixed.

Coordination has costs: communication, verification, enforcement, and trust-building. For most of history, these costs scaled with participants. Coordinating ten million people required a massive bureaucratic infrastructure.

The internet collapsed communication costs. This enabled new forms of coordination—open source software, Wikipedia, and social movements organized through hashtags. But communication isn't enough. You also need verification (is this person who they claim to be?), enforcement (will they do what they promised?), and trust (if things go wrong, is there recourse?). The internet left these to old institutions.

This is why the first two decades produced the "platform paradox." Technology promised decentralization; the outcome was unprecedented concentration. Facebook, Google, Amazon—each became a new institution, solving coordination problems the old way: by centralizing control.

Blockchain technology collapses verification and enforcement costs. Cryptographic proofs replace institutional trust. Smart contracts replace courts. Distributed consensus replaces central authorities. The cost of coordinating ten million people approaches the cost of coordinating ten.

Stablecoins moved $27.6 trillion in 2024—more than Visa and Mastercard combined. Not because they're better payment rails. Because they're coordination rails that don't require permission from institutions that control traditional finance.

The coordination monopoly is breaking. The question is what replaces it.

James Carse, in Finite and Infinite Games (1986), distinguished two types of play. Finite games are played to win—defined players, agreed rules, clear endpoints. When someone wins, the game is over.

Infinite games are played to continue playing. Players may change, rules may evolve, but the game goes on. There is no winning an infinite game—there is only continuing it or failing to.

The distinction runs deeper than it first appears. Finite games require boundaries: temporal, spatial, and membership. Infinite games resist boundaries. The moment you define them too precisely, you've converted them into finite games.

Building a trustless system is committing to an infinite game. You're not building toward an exit. You're building infrastructure designed to outlast its creators. The Ethereum blockchain is meant to run for decades, perhaps centuries. Bitcoin's core promise—a store of value independent of any government—requires indefinite continuation.

The Ethereum Foundation calls this "the Infinite Garden"—a metaphor drawn from Carse. Gardens require ongoing attention. They don't reach a final state. Tending is never complete.

But nearly every funding mechanism is designed for finite games.

Venture capital has a fund lifecycle. Limited partners expect returns within 7-10 years. This creates pressure for exits—acquisitions, IPOs, token launches—that convert infinite game systems into finite game outcomes. The VC doesn't want the garden to thrive forever; the VC wants to sell their stake at a higher price.

Token launches front-load value extraction. Early holders expect appreciation. This creates incentives for hype cycles, artificial scarcity, and governance that serves token value over system sustainability.

Foundation grants are finite by definition. When the grant ends, the work either finds new funding or stops. Corporate sponsorship ties development to sponsor interests that change.

The BIS documented this as the "decentralization illusion"—the tendency of blockchain systems to concentrate power despite decentralized rhetoric. Tim O'Reilly called blockchain "the most rapid recentralization of a decentralized technology I've seen in my lifetime."

Not because anyone intends it. Because finite game funding creates finite game incentives, and those incentives select for centralization.

You can't bootstrap an infinite game with infinite game resources. They don't exist yet. So you take VC money to build the thing that shouldn't need VC money. You accept the grant to create what should outlast foundations. You launch the token to fund the protocol that's supposed to resist capture.

This isn't hypocrisy. This is the trap. The question is whether you're building toward escape velocity or entrenching the dependency that got you started.

The mental model most projects operate under: raise capital, build product, achieve adoption, extract value. This is the startup playbook. It works brilliantly for finite games.

Applied to infinite games, the result is predictable. Early backers need liquidity. Governance concentrates on enabling faster decisions. Competitive pressure demands efficiency. Before long, the "decentralized" system looks like the centralized systems it was supposed to replace. This is the Web2.5 equilibrium—decentralized in rhetoric, centralized in practice.

Consider the language: "Funding." "Investment." "Returns." "Runway." This assumes finite game logic. Money in, more money out. Success is measured in multiples. Failure is measured in lost capital.

But infinite games don't have returns in this sense. The "return" on maintaining critical infrastructure is that the infrastructure continues to exist. The "return" on contributing to a protocol is that the protocol remains viable. No IRR. No exit multiple. Just continuation.

This doesn't compute in finite game accounting.

The mental model gap shows up everywhere. Foundations sit on treasuries, unsure whether to spend aggressively or preserve indefinitely. Projects promise progressive decentralization while their funding structures incentivize centralized control. Builders talk about "community ownership" while their tokens create class distinctions between early and late participants.

The old mental models aren't wrong. They're incomplete. They capture finite game dynamics perfectly. They miss infinite game dynamics entirely.

Sustaining an infinite game requires solving three problems simultaneously.

The resource problem: Where do resources come from? Not one-time fundraising but an ongoing flow. Not a single patron, but a distributed contribution. Not one sprinkler watering the garden, but rain from the entire sky.

The legitimacy problem: Why would anyone contribute? Contribution has to mean something—confer status, express identity, create belonging. Open source works partly because contributing to Linux signals competence and values. You become part of a community that matters.

The hardest shift isn't technical. It's conceptual: contributing to shared infrastructure as normal as paying rent, as meaningful as tending a garden.

The governance problem: How do decisions get made without creating concentration? Someone has to decide what gets built, how resources flow, and when rules change. Trustless systems can automate enforcement, but they can't automate judgment. Humans have to decide what rules should be before the protocol can enforce them.

If that power concentrates, the system isn't trustless anymore—it's just trust with extra steps.

Traditional institutions solve these through multiple mechanisms: tithing and endowments for churches, tuition and donations for universities, and taxation for states. They create a continuous flow. They make contributions meaningful through status and identity. They have governance that prevents permanent concentration.

Trustless systems need equivalent structures—without the concentrations of authority those institutions rely on.

If sustaining infinite games were impossible, we would have no examples. But we do.

The Catholic Church has persisted for two thousand years. Tithing creates continuous flow built into the practice of faith—not occasional fundraising but ongoing contribution regardless of any individual donor's enthusiasm. Endowments preserve principal while spending yield; time becomes an ally. Legitimacy as sacred duty weaves contribution into identity—giving wasn't philanthropy but obligation. Distributed infrastructure—parishes, dioceses, monasteries, each semi-autonomous—eliminates single points of failure.

Universities show a similar pattern. Oxford has persisted for over nine hundred years, teaching since 1096. Harvard was founded in 1636. Both predate the governments under which they now operate.

The endowment model creates perpetual funding from finite gifts—a donor gives once; the university benefits forever. Legitimacy ties contribution to status and meaning. Regeneration through tuition, grants, and donations creates ongoing revenue beyond the original endowment. Governance structures—boards, faculties, traditions—constrain any individual's control.

Nation-states sustain coordination at a massive scale. Taxation creates continuous distributed extraction—every transaction, every paycheck generates revenue automatically. Citizenship ties contribute to identity. Services create reciprocity.

What do these institutions share? Multiple resource streams. Continuous flow rather than discrete injection. Legitimacy makes a contribution meaningful. Regeneration. Governance preventing permanent concentration.

These aren't trustless systems. Churches require faith. Universities require credentials. States require a monopoly on violence. But they demonstrate that infinite games can be sustained.

Open source is the closest analogy. Linux has persisted for over thirty years, first released in September 1991. The infrastructure the internet depends on, maintained without central control.

But open source is a mixed case. Linux persists partly on volunteer labor, partly on corporate support from companies that depend on it. Nadia Eghbal documented the crisis: critical infrastructure maintained by exhausted volunteers who can't get paid without getting captured. Projects die or get acquired. The successes achieve corporate alignment or develop cult-like communities—neither transfers cleanly to trustless systems.

The historical pattern is clear: infinite games can be sustained. The mechanisms are known. The question is whether they can adapt to systems designed to resist concentrations of authority.

We cannot prescribe solutions. Anyone claiming to have solved the Infinite Game Problem is selling something. But we can identify constraints that any solution must satisfy.

Endowment. Preserve principal, spend yield. This is how universities achieve perpetuity from finite gifts. The corpus is sacred; only returns are spent. A $1 billion endowment at 5% annual return generates $50 million per year, forever.

The Ethereum Foundation holds reserves. But reserves are runway—they deplete. An endowment requires accepting that spending is constrained by returns, not ambitions. No existing crypto foundation operates with true endowment discipline.

Tithing. Continuous small contributions from many sources. Protocol Guild—projects pledge 1% of tokens to core developers—is the most promising experiment. Not charity but norm. The goal: contribute as automatically as tithing was for medieval Christians.

Early results are encouraging. Whether the norm spreads beyond early adopters remains uncertain. The test isn't whether committed projects contribute—it's whether uncommitted projects feel pressure to join.

Legitimacy. Make contribution high-status, visible, meaningful. This is cultural work, not mechanism design.

Open source succeeded partly because contributing to Linux meant something—it signaled competence, values, membership in a community that mattered. On-chain attestations and reputation systems attempt to make contributions legible. But legitimacy can't be engineered. It emerges from a genuine belief that contribution matters.

Regeneration. The system must generate some of what it needs. Transaction fees fund validators. Protocol revenue can fund development. But alignment between the value generated and the value directed to maintenance remains imperfect. Better mechanisms would tie protocol success more directly to infrastructure funding.

Distribution. Multiple funders, multiple decision-makers. Ethereum's multi-client ecosystem—five execution clients, four consensus clients—is expensive and inefficient. It's also why no single team's failure can compromise the network.

The same applies to funding: many small donors create resilience that one large donor cannot. Gitcoin's quadratic funding amplifies small contributions precisely to achieve this—a dollar from a thousand people creates more matched funding than a thousand dollars from one person.

Sunset. Temporary centralization must have a credible path to dissolution. Every project begins centralized—founders make decisions, early teams hold tokens. The question is whether centralization is a scaffolding or a foundation.

Most promise to decentralize "later." Later rarely arrives. When it's time to transfer power, those holding it discover reasons to delay. What mechanisms force dissolution when those in power resist? Time-locked contracts? Triggered governance? The problem is recursive: who enforces the sunset on the enforcers?

This principle remains largely unsolved.

No existing structure satisfies all six. Protocol Guild addresses tithing but not endowment. RetroPGF addresses legitimacy but struggles with sustainability. The Ethereum Foundation practices distribution but lacks sunset clarity.

Each experiment satisfies some constraints while struggling with others. This is not failure. This is the work.

Five questions that require experiments, not essays.

Can legitimacy scale? It works in small communities where reputation is legible. Everyone knows who contributes, who defects. Can digital systems create social density that makes reputation meaningful at scale? Legibility isn't legitimacy. Legitimacy requires that contribution matters—that people care about the status it confers. Unknown.

What is minimum viable centralization? Every infinite game starts as finite. Someone writes the first code, makes the first decisions. Too much centralization, and founders never let go. Too little and the project never coheres. No framework exists for finding this point.

How do you make sunset clauses credible? Whitepapers are full of progressive decentralization roadmaps. Phase 3 rarely arrives. Technical enforcement faces a recursive problem: who enforces the sunset on the enforcers?

It may require social enforcement—communities refusing legitimacy to projects that miss transitions. But this returns us to the legitimacy question.

Can infinite game structures compete during transition? Finite structures are faster, concentrate resources, attract talent with equity, and make decisions quickly. If infinite structures can't survive bootstrapping competition, their long-term resilience never matters.

Bitcoin suggests it's possible. But Bitcoin had a first-mover advantage and a decade of obscurity to mature. Can new infinite structures compete against well-funded finite competitors in a crowded market?

What does success look like? Finite games have clear metrics: market share, returns, and exit value. Infinite games have continuation—but of what, at what scale, with what properties?

A protocol could continue indefinitely with three users and one maintainer. That's not success. We lack measures for operational continuity, contributor diversity, and regeneration rate. Without measurement, we can't distinguish structures that work from structures that haven't failed yet.

These questions will be answered by builders over decades. The questions themselves are the contribution. Asking what tending looks like is the first step toward learning how to tend.

The finite/infinite game distinction comes from James Carse, Finite and Infinite Games (Free Press, 1986). The Ethereum Foundation's "Infinite Garden" framing draws directly from Carse.

The Trustless Manifesto was published on November 11, 2025, by Vitalik Buterin, Yoav Weiss, and Marissa Posner. Protocol Guild's 1% pledge model was formalized in January 2024.

Nadia Eghbal's Working in Public (Stripe Press, 2020) and Roads and Bridges (Ford Foundation, 2016) document the open source maintenance crisis.

Stablecoin transfer volumes from CEX.IO research ($27.6 trillion in 2024). Nepal protest coordination from multiple news sources; Prime Minister Oli resigned on September 9, 2025.

Linux's first release: September 1991. Oxford has been teaching since 1096. Harvard was founded in 1636.