From Participatory Experiments to Democratic Infrastructure

Carolyn Lukensmeyer’s case studies—Engaged California, Bowling Green’s BG2050, and the Fort Collins citizens’ assembly—establish convincing proof of concept. When conveners offered regular citizens opportunities to deliberate, they responded admirably. Tens of thousands of Californians weighed in thoughtfully on postfire recovery. Roughly one in ten people in Bowling Green contributed to the city’s fifty-year growth plan. When twenty of their neighbors were randomly selected to develop a major land-use plan, two-thirds of Fort Collins voters ratified it. Skeptics argue that ordinary Americans are either too disengaged or too overengaged (that is, polarized) to make constructive contributions to the policy process. Lukensmeyer’s examples and others like them show that such blanket claims are false.1

While we agree with Lukensmeyer’s critique of strong skeptics, we see the positive upshot of the cases differently. She believes these successes should encourage upward scaling, from cities to states and from states to a national pilot. We think that jump is less clearly warranted given the evidence. Each case illustrates ways participatory and deliberative innovations (PDIs) can enhance democracy. But each case also offers a distinct lesson about where PDIs tend to stall. California faces a sustainability question. Veteran reformers tell disheartening stories of making progress with one set of officials only to see it evaporate when the political winds or personnel change. The California case study does not consider what will happen when the current administration cycles out. Bowling Green faces an agenda-control problem. Nearly eight thousand residents generated two thousand ideas and more than one million votes. But no one has specified the path by which any of that even constrains, much less determines, the next zoning decision or capital budget. Fort Collins faced (and only partly solved) an accountability problem. The city council placed the citizens’ assembly’s recommendation on the ballot, and that institutional move gave deliberation a real role in the legislative process. Elected officials, however, tend to cede their authority only when they want to avoid accountability for difficult decisions.

Each process helped solve a specific local problem. But that is not the same as contributing to ongoing governance. Rather than further proliferating one-off experiments, we urge democratic reformers to prioritize building durable channels for PDI into institutions that have the authority to act on their recommendations. That is, now that we have successful examples of PDI events, reformers should turn their efforts toward building PDI as democratic infrastructure. 

In the short and medium run, we need to build PDI infrastructure within existing institutions of representative democracy, rather than running them in parallel or as replacements. When PDIs bypass or substitute for representative institutions instead of deepening them, innovations flourish only briefly or over a limited set of issues. They wither when personnel or interest configurations change. We should focus less on multiplying PDI episodes and more on weaving them into the routine workings of institutions that already exercise democratic authority. 

Lukensmeyer’s cases thus suggest that, for the foreseeable future, touting PDIs as alternatives to representative institutions often impedes our ability to transform episodes into infrastructure. Episodes can be successful while leaving the background practice untouched. Infrastructure changes the background. A citizens’ assembly that produces a binder of recommendations (or even decisions) is still an episode. A statutory or constitutional requirement that legislative or regulatory bodies respond on the record to an assembly’s recommendations is infrastructure. We see our proposal as both more modest and more ambitious than current reform practice. It is more modest in its aim of supporting representative democracy rather than replacing it. It is more ambitious in its aim to institutionalize PDIs rather than merely proliferate them.

Observers cite citizens’ apathy or ignorance as the main impediment to converting episodes into infrastructure. Lukensmeyer’s cases blunt that objection. Institutional uptake, not citizen competence, is the main limiting factor. Even when a PDI produces solid recommendations, the people running institutions with authority to act on them often cannot, will not, or do not believe they should. Most public hearings remain ceremonial. Constituent-communication­ systems reward volume, intensity, and organized mobilization rather than considered public reasoning. Under polarized conditions, legislators have stronger incentives to respond to mobilized factions than to reflective majorities. When a PDI ends with recommendations but no formal pathway, the failure does not lie with the citizens who took part; it lies with the institutions. Representative bodies were not designed with deliberative citizen input in mind. Retrofitting them should rise to the top of reformers’ to-do list.

Institutional uptake requires a motivating and organizing principle. We argue that the most promising is “recursivity.”2 The ideal of recursive representation is rooted in the beliefs that representation works better as dialogue than as monologue, and that the dialogue should be ongoing rather than episodic. Election campaigns bundle many issues into a single (often) binary choice, muddling mandates and obscuring accountability. The harried heat of campaigns is also not a good environment for constructive conversation. The long stretches between elections (where most actual governing happens) present a more generative climate for the dialogic cycles we have in mind. Citizens articulate their problems and goals, officials propose ways to address those problems and achieve those goals, and citizens then revise in light of that response. The loop runs again. 

The cycle is not just an abstract ideal. Over the past twenty years, we have run deliberative town halls (DTHs) that connect members of Congress with their constituents. DTHs are structured events in which officials discuss a single policy area with a random sample of their constituents, who receive simple but high-quality, nonpartisan background information in advance. The results are consistent and striking. Real two-way communication on difficult issues like immigration and health policy yields gains (for both citizens and officials) in knowledge quality, warranted trust, substantive persuasion, and other democratic goods.3 The demand matters as much as the supply. When offered the chance to deliberate with their representative, randomly selected constituents participated at rates several times higher than in conventional outreach. The participants were not the usual suspects either. Habitual nonvoters, the politically disengaged, and constituents with low prior trust in Congress took part at the same rate as everyone else, and their trust and political efficacy increased the most after the session. Good design, in other words, can reach the people who believe that representative democracy has been failing them. And elected officials are more apt to embrace opportunities to reach disaffected voters.

DTHs promise to provide especially valuable subsidies to representative governance, complementing smaller, more-intensive PDIs such as the citizens’ assembly (CA) in the Fort Collins case. CAs are mini-publics designed to elicit novel ideas and considered judgments in ways that interest-group politics cannot. They also ground those judgments in knowledge, reflection, and community in ways no opinion poll can match. Yet, because of their expense and complexity, they remain episodic and often run parallel to the institutions whose authority makes deliberation consequential. Perhaps more important, CAs face two challenges to their perceived legitimacy. First, conveners typically prestructure CAs quite substantially. Doing so may be necessary for CAs to produce their intended results within the allotted time. But observers and participants raise concerns that such prestructuring takes some options off the table and creates a favorable gradient toward others. Admirably, the Fort Collins experiment included an initial “mini-deliberation” stage to uncover latent community perspectives as input to the CA. We recommend including a more comprehensive DTH before the CA is convened, to advise, revise, and provide assent to the initial protocol proposed by the organizers. Doing so would lend democratic legitimacy to the agenda-setting stage.

The second challenge to CAs’ perceived legitimacy arises at the other end, after participants make their recommendations. Conveners typically use stratified random sampling to populate the assembly. CAs, however, require intensive commitment, so the number of participants is typically small and the acceptance rate (or compliance, in the terms of field experiments) is very low. That combination suggests that, despite stratified random sampling, strong selection on unobserved characteristics weakens the warrant for believing that the CA results represent the whole constituency well. For example, we may stratify to ensure that we include poor single mothers in the sample. Within that group, however, only those with a grandparent or other childcare support would be able to spend multiple weekends away from home. Yet disadvantaged people embedded in strong social networks are likely to experience policy problems and solutions in substantially different ways than people who are truly on their own. In general, more intensive deliberative formats face greater challenges around agenda-setting, participant selection, expense, and moderation.4 We have run successful DTHs with over seven thousand participants, and the lower barriers to entry in terms of time and travel mean that our selection pressures are much less severe. We propose running a DTH after CAs deliver their recommendations to create a broader democratic review and ratification stage. Doing so could greatly enhance the recommendations’ actual and perceived legitimacy with the general public, in addition to reassuring officials that they are robust and will be popular.

Imagine Lukensmeyer’s cases reconfigured as infrastructure. Fort Collins’s citizens’ assembly would be bookended by DTHs and extended from a one-off land-use decision to a recurring practice on the city’s most contested questions. Elected officials would participate at certain stages rather than merely receive output. We could pair Engaged California with a standing legislative commitment to deliberative consultation on disaster recovery, so the platform’s value would not depend entirely on the current governor’s appetite for participation. We could tie Bowling Green’s BG2050 vision to an annual deliberative review in which the county executive and council members report to participating residents on whether implementation is tracking the original vision. The actual cases stopped short not because the deliberative elements were poorly executed but because the representative elements were detached from the CAs.

Some worry that bringing representatives into deliberation risks politicizing or distorting the process. The risk is real. But it is a design challenge to be mitigated—and balanced against problems in the status quo—not a decisive reason for separation. The alternative is deliberation that remains normatively attractive but politically peripheral. Others worry that institutionalization will drain deliberation of its democratic energy. That risk is also real. But the alternative risks—fragility, ad hoc effort, dependence on champion leaders—are evident in Lukensmeyer’s three cases. The design task is to build durable institutions that preserve contestation, transparency, and public reason. The alternative to that work is not purity. It is marginality.

Many democracies have already begun converting deliberative and participatory experiments into more durable democratic infrastructure. Participatory budgeting has become routine or required throughout Latin America. Australia, Belgium, the European Union, France, Germany, Greece, Ireland, Portugal, Romania, Scotland, South Africa, and Switzerland have all built PDIs into formal government work, with many other nations considering such provisions. (A recent OECD report, Catching the Deliberative Wave, consolidates the growing global movement.)5

Lukensmeyer’s cases should both encourage and redirect us. The question is not whether to scale up deliberative experiments. It is whether the representative institutions that already structure American political life can be redesigned to make deliberation routine rather than occasional. The future of deliberative democracy depends less on proving (again) that citizens can reason together and more on building institutions that can receive, respond to, and learn from what they hear. The “responsiveness imperative” named in Our Common Purpose is the right frame for the work ahead.6 But the biggest roadblock is institutional responsiveness, not citizens’ willingness. We strengthen democracy when citizens can see their considered judgments reflected in the decisions that shape their lives. That is the work of building democratic infrastructure.