The cycle that began with [YOU] on AI asks what the technology means for the human beings who live inside it. Ostrom’s framework asks the prior question: who decides? The governance of AI—the rules about what the systems may do, who may build them, how the resources underlying them may be used, who bears the costs and who captures the benefits—is the most consequential set of decisions humanity has ever faced, and it is being made almost entirely without the participation of the people most affected. Ostrom’s collective-choice principle names this as a structural failure: a commons is legitimate to the degree that those who live under its rules had a hand in making them, and by that standard the AI commons is a profoundly illegitimate one.
She stands in the cycle alongside Eliezer Yudkowsky—both urgently concerned with the governance of AI—but from opposite vantage points. Yudkowsky argues from the catastrophic risk of misaligned superintelligence toward the conclusion that only a coordinated global response can avert extinction. Ostrom argues from the empirical study of successful commons governance toward the conclusion that no single global authority can be adequate to a problem of genuine complexity, and that polycentric governance—many overlapping centers of decision at many scales—is both more robust and more legitimate. The tension between these positions is productive rather than merely contradictory: Yudkowsky names the stakes, Ostrom names the structure.
Her most important contribution to the cycle may be her refusal of the panacea. Late in her career, Ostrom became almost a one-woman crusade against the seductive idea that there exists, somewhere, a policy solution that will work everywhere. She titled a major paper “A Diagnostic Approach for Going Beyond Panaceas” and stated her conclusion flatly: ‘one-size-fits-all policies are not effective.’ The AI governance debate is saturated with panaceas. Ostrom’s life work is a standing demand to ask, before endorsing any of them: what kind of resource is this, at what scale does this problem live, and which institutional form is actually matched to it? This diagnostic discipline—refusing the clean solution in favor of the honest diagnosis—is the deepest thing she offers, and the thing the cycle most needs.
Her discovery that human beings are conditional cooperators—not the rational egoists of economic theory but beings capable of building and sustaining trust, of monitoring and sanctioning one another, of making and keeping collective rules—is the empirical foundation of the cycle’s humanistic orientation. The same optimizing agent that AI alignment researchers warn about—the rational egoist built into the foundations of reinforcement learning—is exactly the model of human nature that Ostrom spent her life disproving in the field. The machines we are building embody the false theory of human nature that the study of actual humans refuted. The deepest question her work leaves is whether, surrounded by systems that optimize without caring, we will remember that the best things about us were never the things the machine can do.
Ostrom was born in Los Angeles in 1933. As a girl she was steered away from trigonometry—the school’s policy held that girls who had not excelled in algebra and geometry simply did not take it—and the missing mathematics later disqualified her from doctoral study in economics at UCLA. The department encouraged her, at most, to take economics as a minor. She earned her PhD in political science instead, in 1965, and proceeded to spend the rest of her life dismantling one of economics’ most cherished pieces of conventional wisdom. When the prize committee called her the first woman to win the Nobel in Economics in 2009, they were also, without quite saying so, marking the return of someone the field had once shown the door.
Her doctoral work studied groundwater in Southern California, where competing public and private water producers, facing the collapse of an over-pumped basin, had managed to craft their own rules and save it—without nationalizing the water or auctioning it off. That early case planted a suspicion that would harden, over decades of fieldwork, into a thesis. Around the world, in thousands of documented cases, communities had governed shared resources sustainably for generations. None of them fit Hardin’s parable. All of them should, by his logic, have collapsed. They had not. The crucial distinction Ostrom pressed was that Hardin had described open access—a resource with no boundaries, no rules, no community—and called it the failure of governance. A genuine commons is the opposite: a resource governed by an identifiable community that has agreed on who may use it, how much they may take, who monitors, and what happens when someone cheats.
With her husband Vincent Ostrom she co-founded the Workshop in Political Theory and Policy Analysis at Indiana University—the Ostrom Workshop—which became the leading center for the empirical study of institutions. Her landmark Governing the Commons (1990) and decades of subsequent work developing the Institutional Analysis and Development framework established her as the preeminent scholar of how human beings govern their shared inheritance. She remained, to the end, an empiricist who distrusted every grand theory, including, characteristically, her own.
The tragedy is not inevitable. Hardin described open access and called it the commons. Ostrom documented the commons and found it was governed. The tragedy of the commons is real, but it is the tragedy of ungovernance, not an iron law of collective action. Whether a shared resource thrives or collapses depends on identifiable features of how its governance is arranged—features that can be studied, compared, and in some measure designed.
The eight design principles. From comparative fieldwork on hundreds of long-surviving commons, Ostrom distilled the regularities that distinguished durable institutions from failed ones: clear boundaries on users and resource; congruence between rules and local conditions; collective-choice arrangements that give affected users a genuine role in making and modifying rules; effective monitoring; graduated sanctions for rule violations; accessible conflict-resolution mechanisms; recognized rights to self-organize; and nested enterprises for resources embedded in larger systems. She insisted these were not a recipe but a diagnostic lens—patterns to recognize, not blueprints to implement.
Polycentric governance. The deepest thing Ostrom gave us is the concept of polycentricity—governance distributed across many overlapping centers of decision-making at many scales, none of them supreme. Her Nobel lecture was titled “Beyond Markets and States,” and the “beyond” pointed at exactly this: a system with many centers that check and learn from one another, more resilient and more adaptive than any single authority. Applied to AI, it argues against calls for a single global regulator and toward the deliberate strengthening of the many centers—national, professional, sectoral, communal—that are already governing AI at the scales they are suited to.
No panacea. Ostrom’s most important negative finding is that no single institutional form works everywhere. The specific rules that worked in one irrigation system would wreck another, because the conditions differed. The AI governance debate is saturated with panaceas—just make it open, just regulate it, just align it—each offered as the answer to a problem whose defining feature is its diversity, complexity, and dependence on context. Ostrom’s diagnostic approach asks different questions: what kind of resource is this, at what scale, with what community, facing what specific risks? The answers differ by case, and so do the appropriate institutions.
Trust as infrastructure. Ostrom’s empirical studies showed that human beings cooperate far more, and far better, than the rational-egoist model predicted. What made the difference was not formal enforcement alone but trust—built through communication, repeated interaction, mutual monitoring, and the slow accumulation of kept promises. The AI commons is failing, in large part, because trust is collapsing on every axis. Ostrom’s work suggests that no regulatory mechanism can substitute for this missing trust, because trust is what makes any mechanism function. The hard, unglamorous work of building trust is not adjacent to the governance problem. By her lights, it is the governance problem.