We live in the wealthiest civilization in human history, yet we struggle to articulate, much less pursue, a compelling vision of what that civilization should become. Humanity has no shortage of positive futures to pursue, such as a spacefaring civilization, a world without disease, and forms of governance that enable lives of meaning and dignity. The wonders of the modern world, from reusable rockets to nuclear power plants that can power entire cities, were built by people pursuing grand visions.
Our wealth can become a powerful instrument for building such futures. The history of transformative firms such as DeepMind, SpaceX, and Nvidia suggests that backing extreme outliers in intelligence, drive, and technical vision is one of the most effective ways to spur progress. The same is true in philanthropy. The Ford and Rockefeller Foundations’ support for agricultural scientist Norman Borlaug’s development of high-yield, disease-resistant dwarf wheat varieties is credited with helping prevent mass starvation on an enormous scale.
Yet turning wealth into great futures requires a fundamentally different logic from the one that created the wealth in the first place. Our problem is not a shortage of capital, but a failure of institutional design.
America’s wealthy are offered an endless array of vehicles for growing wealth, and a much narrower set of socially approved philanthropic options. Between these poles lies an entire realm of underexplored possibilities; projects that are neither profit-maximizing nor traditional charity, but could meaningfully contribute to building a greater future.
Without a clear framework, wealth drifts. It sits in passive assets or gets divided between philanthropy and profit-seeking, neither of which sustains a long-term purpose of its own. Capital left to itself becomes inert, and the effects are visible. Americans are richer than ever, but our public spaces, art, and ambitions rarely reflect that abundance.
What is missing is not money, but a system that can hold a steady goal over time while allowing the means of pursuing it to evolve. To achieve this purpose, we propose the creation of “permanent games.” A permanent game is a durable system of capital, rules, incentives, and selection pressures designed to pursue a civilizational goal across generations by funding competition, not institutions. It preserves processes, standards, and evaluation mechanisms while also allowing individual organizations to be created, transformed, and destroyed. Instead of trying to ensure that any particular institution endures, the game itself is designed to protect the integrity of a project across time.
To understand why such a novel structure is required, we must first examine the limits of our current approach to philanthropy.
Our Foundations are Crumbling
The dominant philanthropic structure, the nonprofit foundation, has proven incapable of advancing ambitious civilizational goals. The so-called “Big Three”—the Ford Foundation, the Rockefeller Foundation, and the Carnegie Corporation—were once powerful institutions that shaped much of the postwar world. The U.S. government outsourced substantial intellectual, diplomatic, and even cultural work to them. Notably, the Ivy-educated WASPs who ran these foundations shared a coherent worldview: opposition to radical politics, faith in the power of experts to solve seemingly intractable social problems, and a commitment to an American-led democratic world order.
For a time, the system worked. The foundations played a key role in creating the entire field of molecular biology, eradicating hookworm in the American South, and developing the policy frameworks behind the Marshall Plan, which rebuilt Western Europe following World War II. However, that system broke down during the crises of the 1970s. Vietnam, Watergate, stagflation, urban decay, and declining trust shattered the ideals of mid-century liberalism. At the same time, the 1969 Tax Reform Act fundamentally altered the operating environment for American philanthropy. It imposed mandatory annual payout requirements, sharply limited lobbying and political activity, created new disclosure rules, and empowered the IRS to supervise foundations with unprecedented scrutiny.
Together, these pressures made large foundations more bureaucratic, more cautious, and less willing to pursue the long-horizon experimental projects that had defined the previous era. Meanwhile, control of the great foundations shifted to people who shared neither their founders’ backgrounds nor their worldviews. No unified theory of progress replaced the old one. Today, some of America’s largest foundations operate according to forms of post-national progressivism that would have been unrecognizable, even repugnant, to their founders. The Ford Foundation, once a pillar of Cold War American internationalism, now funds grassroots activism that explicitly rejects American exceptionalism. The Rockefeller Foundation, which funded the Green Revolution, now prioritizes “equity” frameworks that would perplex John D. Rockefeller’s vision of scientific efficiency.
Modern foundations have a few structural problems. First, foundations often have vague mandates. This can result in a donor’s money ending up in the hands of people they profoundly disagree with. A foundation chartered to “improve education” or “promote human welfare” imposes no meaningful constraint on how future boards will interpret those aims. Vague mandates mean that there is no enforcement mechanism to keep foundations disciplined and focused.
There is also a talent problem. Foundation leadership is drawn from elites whose values often diverge from the founder’s, while succession mechanisms are typically weak, turning leadership selection into a popularity contest. The American sociologist Philip Selznick called this process “institutionalization”, a condition where the true purpose of the foundation shifts from its stated goal to institutional preservation. This effectively means that ideological drift is baked into the structure, because once an organization becomes an institution, its priority shifts from the founder’s mission to its own survival and legitimacy. Given enough time, deviation from the founder’s vision is an inevitable outcome, making the permanence of foundations itself a design flaw.
The deeper issue is that foundations attempt to preserve both capital and intent in a single structure, and over time, the former crowds out the latter. Bill Gates’s recent decision to spend down his foundation within twenty years, rather than let it outlive him, is best understood in this light. It reflects realism more than urgency. The only way to ensure his intentions are honored is to spend the money before he dies and the foundation inevitably becomes something else.
Ironically, this same logic animated Peter Thiel’s recent warning to Elon Musk that perpetual charitable vehicles often empower their founders’ ideological adversaries. When Musk was told that signing the Giving Pledge in 2012 would cause $1.4 billion of his wealth to go to Gates in the event of his passing, he responded, “What am I supposed to do, give it to my children?” To this Thiel replied, “You know, it would be much worse to give it to Bill Gates.”
If both institutions and heirs predictably drift from a founder’s values, then neither philanthropy nor inheritance offers a satisfying answer to the question: what is wealth for? Without an intentional worldview, a vision of how wealth might create a better world, and institutions capable of executing that vision, immense societal riches fail to translate into sustained human progress.
Gaming the System
To solve this problem, we return to the idea of the permanent game.
A permanent game has four core components: a fixed, measurable long-term objective; a gated pool of capital that cannot be permanently captured by any one organization; periodic competitive allocation across multiple approaches; and rotating evaluation by qualified experts.
Each permanent game is created to pursue a single, legible, long-horizon objective, defined narrowly enough to evaluate yet broadly enough to permit multiple approaches. For example, “Reduce the real cost of housing construction by 50%” or “Develop commercially viable fusion power.”
The game controls a gated endowment—say, $500 million to $5 billion—invested for longevity. Its payout would be conditional rather than guaranteed, with allocations reviewed at fixed multi-year intervals. This distinguishes these games from foundations where capital is attached to institutions, venture funds where capital is attached to exits, and federal funding where capital is attached to political goals.
Permanent games would create a lightweight mechanism for spinning up multiple temporary organizations to pursue the same objective. Creating temporary organizations is the core idea behind Focused Research Organizations (FROs), and one reason why DARPA has remained surprisingly effective over time.
At the outset, the game defines success in measurable terms. To make this concrete, consider a permanent game built around the de-extinction of animals, like the woolly mammoth. Let’s follow this game from its creation through a couple cycles of competition and evaluation.
The game funds five competing organizations, some nonprofit, some for-profit, some hybrid, each pursuing a distinct path. One focuses on editing elephant genomes to express mammoth traits. Another works on artificial gestation to get around the limits of elephant pregnancies. A third concentrates on computational genomics, trying to reconstruct traits in silico before moving to live animals. A fourth treats the problem as ecological, developing cold-adapted habitats and reintroduction models. A fifth combines everything in a single, vertically integrated effort.
In the first cycle, some of these approaches show early promise while others stall. The genome editing group may demonstrate partial trait expression but struggle with viability. The artificial gestation team may fail entirely. The ecological group may show that existing habitats cannot support the target species without further engineering. This is not failure so much as discovery; the purpose of the game is not to fund a winner immediately, but to surface information about which paths are viable.
Every five years, a council reviews the results. Some groups lose funding, others are wound down or recapitalized, and new teams are formed to try different approaches. The council itself turns over each cycle, drawing from a standing pool of qualified experts whose eligibility is defined in the founding charter in specific, field-relevant terms the council cannot revise. A smaller standing body of trustees, rotating on a longer cycle, holds sole authority to update those criteria as fields evolve. Individual evaluators serve fixed terms with mandatory cooling-off periods, so no single cohort can entrench its own standards of judgment.
Yet even this structure has limits. Over a generation or more, the trustees become interpreters of a founding document rather than executors of a living intent, which is the same drift the permanent game was designed to prevent. For that reason, the charter should build in a constitutional moment at the twenty-five or fifty year mark, where an external body can review the game’s structure and, if necessary, dissolve it entirely. If the game itself begins to prioritize its own survival over its stated goal, it has become the thing it was designed to replace.
In the next cycle, new entrants incorporate what was learned. A new team might combine genome editing with advances in synthetic biology. Another might abandon mammoth traits entirely in favor of engineering cold-tolerant elephants directly. Over time, the composition of approaches shift, but the direction of the game does not.
Colossal Biosciences, founded in 2021, offers a ready-made test case. The company is already pursuing mammoth de-extinction with private capital, but its structure illustrates the problem permanent games are designed to solve. A single organization, accountable to investors who need an exit, pursues a timeline set by funding cycles rather than biology.
A permanent game built around de-extinction would not replace Colossal, it would integrate it. Colossal could compete for game funding alongside university labs, nonprofit research groups, and rival startups, each evaluated based on scientific milestones rather than commercial ones. The founding consortium might include backers who have already invested in Colossal, alongside university endowments willing to lock capital on a twenty-year horizon. The goal would be written in specific biological terms: a viable cold-adapted proboscidean reproduced to second generation in a functioning habitat. Colossal might win or it might be overtaken by a team that doesn’t exist yet. Either outcome would mean the game is working.
The term “permanent game” invites an obvious question: what happens when the objective is achieved? Two designs are possible. In the first, the game includes a sunset clause. When the objective is demonstrably met, remaining capital is distributed as a final prize, the competing organizations are dissolved, and the game closes. Here, “permanent” describes the durability of the process rather than of any particular institution. The game runs only as long as the problem remains unsolved.
In the second design, the goal is written so that it can never be fully achieved, only advanced. Reducing the real cost of housing construction has no final state. You can always go further, addressing new material constraints and creating new methods of construction. Games of this kind resemble scientific disciplines more than engineering projects. Neither design is inherently superior; the preferable choice depends on whether the objective is a discrete threshold or a continuous frontier.
The permanent game design borrows elements from existing successful innovation systems, including prize competitions. For example, the 2004 DARPA Grand Challenge offered a $1 million prize for teams that could build a fully autonomous vehicle capable of navigating a 150-mile desert course in the Mojave. All entrants failed, but it sparked a wave of research into autonomous vehicles across the country. Stanford team leader Sebastian Thrun would later help launch Google’s self-driving-car effort, while other participants went on to lead Aurora, Uber ATG, and other autonomous-vehicle startups.
DAOs are perhaps the most structurally interesting precedent here. Their core insight is that governance rules can be encoded to outlast any individual participant, which is precisely the problem permanent games are trying to solve. In principle, DAO governance need not be tied to token holdings; voting power could be assigned based on credentials, past contributions, or other non-financial metrics. However in practice, these systems tend to collapse back toward financial weighting or static proxies for expertise. Credentials are hard to verify and quickly become outdated. Contribution-based systems reward activity more than judgment. More importantly, once encoded, these rules are difficult to revise without conflict or capture.
Let’s return to the de-extinction example and imagine a DAO used to allocate funding across competing teams. Voting power is assigned to credentialed biologists and engineers. Early on, this works. But as new techniques emerge, the relevant expertise shifts and the credentialing system begins to select for incumbents trained in older approaches. Updating the criteria requires approval from the existing voter base, which has little incentive to dilute its own authority. The system does not fail outright, but it hardens. Governance remains intact, but its judgments become less aligned with the frontier of the field. Permanent games require both durability and the capacity for judgment to evolve. DAOs reliably provide the former, but not the latter.
Until recently, a gated endowment model roughly described the governance of the MRC Laboratory of Molecular Biology (LMB) in Cambridge. The LMB has produced twelve Nobel Prizes over its 78-year history—an average of one every six years—with a yearly budget of around $50 million and only a few hundred employees. Every few years, a committee met to review the laboratory’s work and evaluate its scientific output. Groups that failed to produce were shut down, while those making breakthroughs received more resources. The LMB kept its governing structure deliberately small and technical. Scientists judged other scientists’ work, preventing the drift toward professional managers and grant administrators that bureaucratizes most research institutions.
Gated endowments impose something like market discipline, creating a genuine threat if the organization strays or becomes ineffective. Without the cushion of perpetual funding, failure has consequences. Resources run out and the institution winds down. That pressure sharpens priorities and keeps attention on results.
In the de-extinction game, a team that spends five years pursuing an unworkable approach does not persist indefinitely. It is shut down and capital returns to the game. At the same time, permanent games give organizations room to experiment and take risks, since they are judged on outcomes rather than forced to seek approval for every project or program.
Readers might argue that permanent games just sound like market competition with extra steps, but the de-extinction example demonstrates the gap in market efficacy. No private investor would fund multiple parallel efforts over decades with no clear exit, especially when most of the benefits accrue to science and society rather than the firm itself. Permanent games supply the patient capital and tolerance for failure that neither venture capital nor traditional philanthropy provides.
There are still possible failure modes. The permanent game solves the problem of drift, not the issue of origin. Whoever sets the initial goal and selects the first council is making a values-based choice, and no procedural mechanism can push this decision into neutrality.
Permanent games could be captured by evaluators, misspecified in their goals, or stagnate if progress proves harder than expected. The advantage of this model is not perfection, but adaptability. By rotating decision-makers, forcing periodic evaluation, and dissolving failing organizations, they make it harder for any single error to persist indefinitely.
Cooperative Competition
Many principals possess the capital necessary to spark a project, but lack the time required to assume long-term responsibility. Worse, coordination between these actors is rare. Institutions need stability to become load-bearing, and ecosystems require interlocking structures to support long-term planning. This creates a collective action problem where no single principal can build a complete ecosystem alone, yet the trust and shared vision required for collective governance are in short supply. The principals most capable of funding such systems are often the least willing to cede the control required to make them work.
If permanent games are engines of competition, why call for cooperation? The answer lies in the difference between competing within a game versus playing by yourself. Friendly competition united around a common, clearly defined goal creates the necessary conditions for advancement and achievement. Without that shared definition there is no competition, only the pursuit of incommensurate futures.
The wealth to dramatically drive progress exists, and more is coming. A new class of economic elites will be created by the liquidity events of SpaceX, Anthropic, OpenAI, and others. The founders of Anthropic have already committed to donating at least 80% of their wealth, while the engineers and early backers of SpaceX have spent a decade trading immediate gains for a shared civilizational mission.
Permanent games correct a specific failure of contemporary philanthropy: they fix the goal, rotate the referees, eliminate the failing teams, and keep the capital in play. Permanent games are not just a new way to spend wealth. They are a way to convert private fortunes into systems that can pursue progress across generations. It is up to the holders of our great societal wealth, both large and small, to build them.