Data Commons Governance

In the AI Story

The framework responds to the training data question by demonstrating that neither privatization nor state regulation exhausts the available institutional options. The contributors to the training-data commons did not produce their contributions as property, and retroactively imposing property frameworks on data generated under different institutional assumptions creates distortions that may exceed the problem they address. But they did produce their contributions under implicit governance arrangements — norms of the open internet, terms of service, licensing frameworks — that can be examined and, where appropriate, updated through explicit community processes.

The practical challenges are significant. Contributors number in the millions and are distributed across every jurisdiction on earth. They have no pre-existing organizational structure, no shared identity, no institutional infrastructure through which to exercise collective governance. The data has already been incorporated into models that cannot easily disaggregate individual contributions. The corporations that hold the data have both the legal position and the economic incentive to resist governance arrangements that would constrain their use of it.

These are real obstacles but not unprecedented in Ostrom's empirical record. Communities with no pre-existing organizational structure, no shared identity, and no institutional infrastructure have developed effective governance arrangements in response to resource crises. The formation of self-governing institutions is itself an institutional achievement — requiring entrepreneurial individuals, catalytic events, and the recognition among potential community members that their individual interests are better served by collective action than by continued isolation.

Whether conditions for this kind of institutional formation exist in the training-data context is an empirical question that cannot be answered by theory alone. The framework's contribution is to identify what the institutional work would look like, not to prejudge whether it can be successfully undertaken.

Origin

The Mozilla-Ostrom Workshop collaboration emerged from parallel recognitions at both institutions that AI training data required governance frameworks the dominant discourse was not providing. Mozilla's public-interest technology orientation combined with the Ostrom Workshop's institutional analysis expertise produced the framework as a concrete operational resource for communities, policymakers, and researchers engaging with the data commons question.

Key Ideas

Four institutional features. Boundaries, congruent rules, collective choice, and monitoring structure effective data commons governance.

Beyond the false binary. Neither privatization nor state regulation alone addresses the core question of who makes the governance arrangements.

Practical but difficult. Scale, jurisdictional diversity, and power asymmetries make this harder than any commons Ostrom studied.

Institutional formation required. The framework identifies what must be built, not whether it can be built — an empirical question.