Minipublics

In the AI Story

The minipublic format addresses limitations of both representative democracy and direct democracy. Representative mechanisms suffer from principal-agent problems and party-system distortions that can produce outcomes diverging from considered public preference. Direct mechanisms like referenda suffer from the absence of deliberation, producing outcomes shaped by campaign dynamics and aggregating uninformed preferences. Minipublics preserve the deliberative quality that legislative processes were originally intended to provide while ensuring the representative breadth that sortition uniquely produces.

Application to AI governance requires specific adaptations. The speed of technological change demands standing bodies with continuous engagement rather than ad hoc assemblies convened for specific decisions. The technical complexity requires information-provision systems that translate rather than simplify. The global scope requires connections between local minipublics and higher-level governance forums — the design challenges Fung addresses in his AI Impact Assembly and National AI Deliberation Platform proposals.

The format's structural resilience to algorithmic subversion is its most important feature for AI governance specifically. Face-to-face deliberation with balanced information and skilled facilitation operates outside the informational environment AI degrades. Recommendation algorithms have no purchase, persuasion technologies have no target, synthetic content has no channel. In an environment where other democratic mechanisms are being systematically compromised, the minipublic format becomes not merely one option among several but the form most capable of producing legitimate governance.

Critics from the democratic elitism tradition argue that minipublics cannot substitute for elected representation because participants lack electoral accountability. Fung's response is that minipublics are not proposed as substitutes but as complements — institutional additions that address failure modes of existing democratic mechanisms without replacing them. The goal is a deliberative system in which multiple mechanisms operate together, each addressing what the others miss.

Origin

The minipublic concept crystallized through the work of Robert Dahl, James Fishkin, and Ned Crosby in the 1970s–1990s. Fishkin's development of deliberative polling provided the methodological foundation; Crosby's Citizens Jury format developed the binding-recommendation variant. Fung's contribution was theoretical integration: situating minipublics within the broader framework of empowered participatory governance and specifying the conditions under which they produce legitimate governance outcomes.

The 2012–2018 Irish citizens' assemblies represented the format's most sustained application to consequential national governance. The success of the assemblies in producing constitutional amendments on issues that parliamentary politics had failed to resolve for decades established minipublics as a viable institutional innovation at national scale.

Key Ideas

Random selection plus deliberation. The minipublic combines sortition — eliminating self-selection bias — with structured deliberation that transforms opinion into considered judgment.

Non-specialists can engage with complexity. The evidence across multiple domains shows that citizens given adequate information and structured deliberation can reason meaningfully about technically complex questions.

Structural resilience to algorithmic subversion. Face-to-face deliberation with balanced information is uniquely robust against the mechanisms through which AI degrades other democratic forms.

Complement rather than substitute. Minipublics work best as additions to existing democratic institutions, not replacements — part of a deliberative system in which multiple mechanisms operate together.