Who gets to decide whether a new technology is safe enough for society? We typically assume that technical questions demand technical answers—that only scientists, engineers, and domain specialists possess the competence to weigh in. But a fascinating institutional experiment, originating in Denmark in the 1980s, challenges this assumption at its root.

Consensus conferences invite randomly selected citizens—people with no specialized training—to interrogate panels of experts, deliberate among themselves, and produce substantive policy recommendations on complex scientific and technological issues. The results have been remarkably sophisticated, often rivaling or exceeding the nuance of expert-only advisory panels.

This raises a profound epistemological question: how can non-experts contribute meaningfully to technical deliberation without first becoming experts themselves? The answer reveals something important about the nature of knowledge, the limits of specialization, and why democratic participation in science isn't just politically desirable—it may be epistemically necessary.

The consensus conference model, pioneered by the Danish Board of Technology, follows a carefully structured process. A group of ten to twenty citizens, selected to reflect demographic diversity rather than subject-matter expertise, receives preparatory materials on a contested technical topic—genetically modified organisms, nuclear waste storage, surveillance technologies, or similar issues where science intersects with public values.

Over several days, these citizen panelists hear presentations from experts they themselves have helped select, cross-examine those experts in public sessions, and then deliberate privately before producing a consensus document. The output isn't a vote or an opinion poll. It's a reasoned, written assessment that weighs competing technical claims against social priorities and ethical considerations.

What distinguishes this model from public consultation or focus groups is its epistemic seriousness. Citizens aren't asked what they feel about a technology—they're asked to evaluate evidence, identify uncertainties, and render judgment. The Danish experience with consensus conferences on topics ranging from gene therapy to electronic surveillance demonstrated that lay panels consistently identified relevant considerations that expert committees had overlooked or deprioritized, particularly around implementation risks and distributional consequences.

The institutional design matters enormously here. The preparatory phase, the adversarial questioning format, the requirement to produce written reasoning—these structural features transform ordinary citizens into something Thomas Kuhn might have recognized as a distinct epistemic community, one organized not around shared paradigmatic commitments but around shared responsibility for a consequential judgment.

Takeaway

Expertise is not solely a property of individuals—it can be a property of well-designed processes. The right institutional architecture can generate collective competence that no single participant possesses alone.

The most common objection to citizen panels is straightforward: if people lack technical knowledge, their participation is merely symbolic—a democratic ritual that contributes nothing to the quality of the resulting decisions. This objection assumes that the only valuable input to technical deliberation is more technical knowledge. But that assumption misunderstands how complex sociotechnical decisions actually work.

Helen Longino's work on the social dimensions of scientific objectivity offers a useful framework here. Longino argues that objectivity in science depends not on individual detachment but on transformative criticism—the presence of diverse perspectives that can identify background assumptions invisible to those who share them. Expert communities, precisely because of their shared training and professional incentives, develop systematic blind spots. Lay citizens bring what philosopher Harry Collins calls "interactional expertise"—not the ability to do the science, but the ability to engage critically with scientists' claims once given adequate context.

Empirical studies of consensus conferences bear this out. Citizen panels on biotechnology in Norway and the Netherlands consistently pressed experts on assumptions about consumer behavior, risk tolerance, and long-term ecological uncertainty that specialist advisory bodies had treated as settled. The citizens weren't contributing new data—they were contributing new questions, drawn from life experiences and value frameworks that technical training tends to screen out.

This reframes what legitimate epistemic participation means. You don't need to replicate an expert's knowledge to challenge the framing of a problem, to notice whose interests are absent from an analysis, or to insist on transparency about what remains uncertain. These are epistemic contributions, not merely political ones, because they improve the quality and completeness of the reasoning process itself.

Takeaway

The value of non-expert participation isn't that laypeople know what experts know—it's that they see what experts have learned not to notice. Diverse cognitive vantage points are an epistemic resource, not an epistemic liability.

If consensus conferences work so well, why haven't they become standard practice? The obstacles are partly logistical, partly political, and partly epistemic in their own right. The Danish model relies on small groups deliberating over extended periods—a format that resists easy scaling. Running a single conference is resource-intensive; institutionalizing them across multiple policy domains and levels of governance multiplies the cost and organizational complexity dramatically.

There is also the problem of capture and co-optation. When consensus conferences produce recommendations that challenge powerful interests, those interests have strong incentives to undermine the model itself—by questioning the panel's representativeness, by flooding the expert witness list with sympathetic voices, or by simply ignoring the results. The Danish experience was relatively insulated from these pressures because the Board of Technology operated with significant institutional independence. Attempts to replicate the model in countries with weaker traditions of deliberative governance have often produced diluted versions that function more as public relations exercises.

Perhaps the deepest challenge is epistemic: as issues grow more technically complex and politically polarized, the preparatory phase becomes both more crucial and more contested. Who designs the background materials? Which experts are deemed credible? These framing decisions shape deliberation before it begins, and they require their own layer of accountability. Some scholars, including James Fishkin, have proposed combining citizen panels with random selection of expert witnesses and independent oversight of preparatory materials—building checks and balances into the epistemic infrastructure itself.

None of these challenges are insurmountable, but they underscore a fundamental insight: democratizing expertise is not a one-time institutional fix. It requires ongoing maintenance of the conditions that make genuine deliberation possible—independence from political pressure, adequate resources, transparent processes, and a culture that takes lay judgment seriously rather than treating it as a concession to democratic ideals.

Takeaway

Democratic knowledge institutions are not self-sustaining. Like any epistemic practice worth preserving, they require deliberate investment in the structural conditions that make honest inquiry possible—and vigilance against the forces that erode them.

Consensus conferences reveal that the boundary between expert and non-expert is not a wall but a membrane—permeable under the right institutional conditions. Knowledge production has always been social, but we rarely design our institutions to reflect that reality.

The lesson extends beyond technology assessment. Wherever specialized knowledge meets public consequences—in medicine, education, environmental policy, artificial intelligence—the question is not whether non-experts should participate, but how to structure their participation so it genuinely improves collective understanding.

Building those structures is itself an act of epistemic responsibility. The alternative—leaving complex decisions entirely to specialists—isn't neutral. It's a choice about whose questions count, and it carries its own risks of systematic error.