Proximity Democracy

In the AI Story

The developer in Lagos, celebrated in The Orange Pill for her collapsed imagination-to-artifact ratio, can now build things previously accessible only to well-resourced teams. This expansion of creative capability is real and morally significant. But the expansion of capability is not the same as expansion of governance, and the confusion between the two is one of The Orange Pill's most consequential elisions. The developer can build; she cannot decide the terms under which building is possible. The models she uses were trained by American and European companies according to criteria she had no input into, reflecting priorities and cultural assumptions she may not share. Content policies constraining what models will and will not produce were written by teams in San Francisco and London, informed by those jurisdictions' legal frameworks and cultural sensibilities, and applied globally without adjustment for contexts in which models are actually used.

Pricing structures determining whether she can afford the most capable models are set by companies whose cost calculations and competitive strategies she has no mechanism to influence. Terms of service she accepts to use the platform are non-negotiable contracts of adhesion, written by corporate lawyers, designed to protect company interests, presented as binary choice: accept or do not use the tool. This is not a democratic relationship—it is a service relationship. The developer is a user, not a citizen. Her capability has expanded; her governance has not. The distance between decision-makers and the developer is epistemic—those making decisions do not understand the cultural context in which she builds, the specific needs of her community, the particular problems her products solve, the local knowledge that should inform how AI tools are designed for her context.

The problem has historical precedents Rosanvallon's work illuminates. Colonial governance was often well-intentioned in its self-conception—the colonial administrator believed (sometimes sincerely) that infrastructure, education, legal systems imposed on colonized peoples would improve their lives. The improvements were sometimes real. Roads were built, schools opened, legal frameworks replaced arbitrary local authority with codified rules. But governance was illegitimate, in Rosanvallon's terms, not because outcomes were always bad but because process excluded the governed from meaningful participation. Roads were built according to colonial power's priorities, not local community's. Schools taught colonial curriculum. Legal frameworks encoded colonial values. The analogy is imperfect—AI companies are not colonial powers, using Claude Code is not living under colonial rule—but the structural pattern is recognizable: decisions profoundly shaping conditions of creative and economic life for people globally are made by people in a handful of cities, without institutional mechanisms through which affected populations can participate in decision-making.

The democratic response is not to restrict access—the developer should have the tools, and any governance reducing access in the name of protection would be paternalistic in precisely the way proximity democracy is designed to prevent. The democratic response is to create governance structures giving the developer genuine voice in decisions shaping her creative environment. Possibilities include: regional advisory bodies with genuine influence over model deployment decisions (not corporate diversity committees performing inclusion theater but institutions with authority to modify content policies, training priorities, pricing structures for specific regional contexts); participatory design processes including affected communities from earliest stages (when design space is still open, not after architecture has been fixed); and open governance frameworks for the most consequential platforms, making decision-making for content policies, training data selection, deployment priorities transparent and participatory, modeled on internet standards bodies and open-source communities.

Origin

The concept descends from the republican principle of subsidiarity—that decisions should be made at the lowest level capable of addressing them effectively—and from Tocqueville's observation that American democracy's vitality depended on local participation more than on national institutions. Rosanvallon developed proximity legitimacy as one of three pillars (alongside impartiality and reflexivity) in Democratic Legitimacy (2011), drawing on his study of how French local governance evolved from Napoleonic centralization through successive waves of decentralization. The concept addresses the tension between the scale at which modern problems present themselves (often global) and the scale at which democratic participation is most meaningful (local, face-to-face, grounded in shared conditions).

The AI application is Rosanvallon's extension of his framework into the domain of platform governance. A 2025 report he co-authored with Yann Algan for the Global AI Summit in Paris proposed 'creating a citizen intermediary body to oversee the use of AI' and 'drawing inspiration from the Swiss model of the citizen army for a democratic oversight committee for algorithms.' The proposals recognize that global platforms governing billions require governance mechanisms that somehow combine the proximity that legitimacy demands with the scale at which the platforms operate—a tension for which no fully satisfactory institutional solution yet exists but which must be addressed if AI governance is to possess democratic legitimacy rather than merely technical competence.

Key Ideas

Governance close to the governed. Decisions made by those who share conditions of life with the affected carry legitimacy that distant governance cannot replicate—intuitive democratic principle now challenged by AI platforms governing globally while accountable locally to no one.

Capability expansion without governance expansion. The developer in Lagos can build what previously required well-resourced teams; she cannot decide terms under which building is possible—models trained, content policies written, pricing structures set by people who do not share her conditions.

Epistemic distance, not just geographic. Decision-makers do not understand cultural context, community needs, particular problems, local knowledge that should inform tool design—gap no amount of corporate goodwill can close because closing requires deep, continuous, reciprocal engagement global platforms structurally cannot provide.

Colonial governance pattern. Well-intentioned imposition of infrastructure, education, legal systems on populations excluded from meaningful participation—improvements sometimes real, governance illegitimate because process excluded the governed, an imperfect but recognizable analogy to platform governance.

Regional governance mechanisms required. Advisory bodies with genuine influence over deployment decisions, participatory design processes from earliest stages, open governance frameworks for consequential platforms—each facing practical obstacles, each necessary for democratic legitimacy that competence alone cannot supply.