Exploring Plural Voting as a Method for Citizen Engagement

Collaborative governance—or “co-governance”—offers a model for shifting power to ordinary people and rebuilding their trust in government. Co-governance models break down the boundaries between people inside and outside government, allowing community residents and elected officials to work together to design policy and share decision-making power. Cities around the world are experimenting with new forms of co-governance, from New York City’s participatory budgeting process to Paris’s adoption of a permanent citizens’ assembly. More than a one-off transaction or call for public input, successful models of co-governance empower everyday people to participate in the political process in an ongoing way. Co-governance has the potential to revitalize civic engagement, create more responsive and equitable structures for governing, and build channels for Black, brown, rural, and tribal communities to impact policy-making.

Still, co-governance models are not without challenges. The hierarchical and ineffective nature of our current governing structure is difficult to transform. Effective collaboration between communities and politicians requires building lasting relationships that overcome deep distrust in government. So far, effective models of co-governance tend to be local and community-specific—making it critical that we share stories of success and brainstorm ways to scale.

In this series, we share stories of co-governance in practice. For this interview, New America’s Hollie Russon Gilman and Sarah Jacob spoke with Paula Berman, Alex Randaccio, and Matt Prewitt from RadicalxChange (RxC). Founded by economist Glen Weyl in 2018, the RadicalxChange Foundation is a 501(c)(3) nonprofit dedicated to advancing the RxC movement, building community, and educating about democratic innovation. RxC connects people from all walks of life—ranging from social scientists and technologists to artists and activists. The RxC movement is ever-evolving and always welcomes new people and ideas to make our social world more diverse, equal, and free.

Could you please give us a brief introduction of your pilot?

We worked with the Office of Climate Preparedness and Disaster Recovery (CPO) in Colorado, a newly established agency created in response to several large disasters and a need for a more focused approach to climate adaptation, to be set on a coordinated roadmap for how the state of Colorado could prepare for climate impacts. As the office began its work, we saw an opportunity to develop a relatively short and replicable civic engagement model for the CPO, so that it could potentially be used again as the roadmap is updated every three years. Quality citizen engagement is a priority for the CPO, and so they agreed to work with us and explore this new way to take citizen inputs into account—in addition to a number of other engagement strategies they used throughout the process.

When developing the model, one of the key challenges for us was that popular citizen feedback formats, such as town halls, often convene groups that are not necessarily representative of the population. They also tend to lack the resources and facilitation required for good, meaningful deliberations to take place. Citizen assemblies, on the other hand, are the gold standard at achieving these ends, but can be extended and resource-intensive endeavors. So we wanted to develop a hybrid. A short and scalable model, but that convened a representative group of citizens selected by lottery. We worked through the process design with Rahmin Sarabi of The American Public Trust, who shared his experience with citizens’ assemblies and other deliberative formats.

We gathered the groups for three-and-a-half hour sessions, online, using world cafe style discussions; as well as a tool for online deliberations called Polis, which helped us understand the diverse perspectives our groups brought in and where they could be bridged; and finally plural voting, which enables participants to rank their priorities in a granular and nuanced way, helping us get to actionable outputs at the end.

How did you go about the recruitment process?

We took a comprehensive database of residential addresses in Colorado and randomly selected around 18,000 of them. Then we sent out invitation letters to these randomly selected addresses across the state, inviting them into a lottery process. We then used this tool called Panelot to draw a lottery from those who volunteered to join. We surveyed the people who responded and asked them to self-select into different demographic categories and then sent that demographic data into Panelot, along with statewide statistics on those demographics, so that it could randomly generate multiple representative panels that are like a cross-section of the state on those demographic categories.

Were you trying to oversample for traditionally underrepresented communities?

If there’s a demographic group that is 1 percent of the population but you have a panel of 20, then the algorithm is going to tell you that there should be zero people of that group in the panel so instead we rounded those up to one. Panelot implements the methodology that is described in a Nature paper that got a bunch of attention a few years ago in the citizens assembly world. The goal is to adopt a thoughtful, neutral, and optimal way of selecting these panels.

Can you walk us through the process with the CPO to determine the topics for citizens to deliberate on?

This work is part of a larger consultation for the CPO in which we’re partnering with Demos Helsinki. Demos interviewed multiple government agencies within the state of Colorado, to map their goals, the different climate preparedness programs that are currently in place, and identify some key growth opportunities for the state’s response.

As a product of these ongoing explorations, we co-developed with the CPO the questions that we wanted to ask. Those questions also shifted over time as the project evolved. For example, at the first session that we held with citizens, we asked them about what they thought the state should prioritize, and found that many of the suggestions were actually already being contemplated by the state in one way or another. So in the following sessions, we also asked where they’re getting their information and how the state could better communicate with citizens. It was an ongoing process that was shaped collaboratively with the CPO.

Can you walk us through the deliberation process?

The process begins with discussions. We divide the panelists into small groups of five or six each with its own facilitator. Citizens can then talk to each other and the facilitators are kind of helping the conversation along.

The first piece in this discussion is just to get them thinking and talking out loud about their feelings and lived experiences in the policy area. In the case of climate, their anxieties, fears, challenges, etc., and then their hopes and what they want to see, what a better future would look like, what services they want to see from the state, etc. We wanted them to become acquainted with one another and establish a human connection with their fellow citizens.

Everything from there on out is directed towards how we can use these tools like Polis and plural voting to [generate] a detailed snapshot of the perspectives in that representative group we’ve assembled. So we then go into a Polis conversation where we have everybody take 15 minutes or so to share their thoughts in response to a prompt, which was “What strikes you as very important for the State of Colorado to consider in its plans on climate preparedness?”

They do that through a simple interface where there’s an input bar to type in your comments and then submit them. After submitting comments, they go into a stack. It looks visually like a stack of index cards so that you can only see the top card at any given point in time. When you’re looking at a whole list, you can only see one comment. But the stack contains all of the comments that everyone has submitted. So in addition to putting your comments into the stack you can also react to other people’s comments by hitting either “agree” or “disagree” or “pass.” Very efficient, just a quick binary data on people’s feelings or reactions.

All of these comments and evaluations are then fed into a report that is automatically generated. The report uses that data to show what are the different clusters of viewpoints that are present in the group, and the specific points of divergence or convergence between the clusters. (Here’s an example of a report for a different conversation, about what’s needed to bring about a thriving multiracial democracy.)

This mapping tool is really interesting. Seeing the clusters helps catalyze deliberations to a degree of clarity and nuance that would otherwise take hours, if not days, to achieve. And one additional element that enables that is that people can submit their inputs anonymously. They can express their unvarnished views and evaluations in a way that they might not feel comfortable to do in the small group discussions. So this tool enables you to rapidly set up future discussions by getting an accurate map of the perspectives and lines of division and consensus in any given topic. But I want to emphasize again: We see this as an assistive tool to be used ideally, in tandem with, face-to-face deliberation.

The third and final step is the plural voting piece, the goal of which is to clearly rank priorities. In contrast with Polis, where you can only agree or disagree with different statements, plural voting also lets participants express the magnitude of their preference. For example, how much do you care about Proposal A or Proposal B?

It does that by working with a credit system that captures people’s preferences with very high resolution, and this usually leads to more moderate expressions of preference than traditional voting models. The most illustrative example is Likert surveys online, when people ask you, “how do you feel about this?” And the options go from “very strongly disagree” to “very strongly agree.” You know, studies find that and our anecdotal experience tells us that when you frame the options like that they tend to choose one of the extremes. The polarization is almost built into the mechanism. Plural voting is a way to solve that way of posing questions. It captures people’s responses with very high resolution and the results usually look like a bell curve, tending towards the center, the more moderate options, instead of the extremes as in Likert surveys.

The way plural voting works is you’re looking at a list of options and you have a fixed budget of voice credits that you can place into the different options to cast votes for them. If I care about a proposal I can cast a vote towards that. If I really care about it, I can cast two or three or four votes depending on the strength of my preference. The catch being that the number of voice credits that it costs to cast each additional vote increases exponentially: One vote costs one credit, two costs four, three costs nine, four costs sixteen, and so on. So for a very strong preference the increase will be quite dramatic, which means I will have less credits left for other options. It gets you to be very thoughtful about how you express your preferences, and you see the tradeoffs pretty clearly in the interface—so it’s also something that can be done in only a few minutes and feels intuitive.

At this point we had participants from many different age groups and backgrounds using these tools, and the response has been very positive. This is encouraging because those are methodologies that can help address deep seated problems in our democracies.

How was the participation and the follow-up?

The CPO has committed to sharing a letter describing how those inputs were considered or influenced the outcomes in the roadmap once it is complete . And then we’re also conducting the process with other groups of stakeholders, like subject matter experts and scientists. What has been most striking for us is that this process both brought in a lot of novel information for the CPO, and was also a meaningful experience for the participants. Our main challenge was how to provide a meaningful experience in such a compressed time, and what we found was that, yes, participants largely had really positive experiences.

And I would say that one element that has contributed to this is that despite using these technical tools, we made sure that everyone was brought on board with a thoughtful concierge process, where participants from all ages and educational backgrounds were walked through the steps before the sessions, as needed. There was also a high participant-to-staff ratio. Panelists had good tech support throughout the entire session, making sure that we could create a feeling of togetherness and everyone being on equal footing to participate in the process and having all of their needs assisted for.

So while these innovative technologies can really go a long way in improving citizen engagement processes, the key to making them work is to bring in careful facilitation and concierge and tech support to make sure that everyone can use the tools and participate fully in the conversation. This was vital to the process, and we really need to shout out to our partners at Healthy Democracy and Civic Canopy, who provided that. And hopefully these are skills that many public servants can be trained in, one day. Even though these were short, three-and-a-half hour sessions taking place online, we had very positive responses. Some people were emotional about the experience and expressed that they felt heard and felt like the State of Colorado cared about them.

How scalable is this, and what do you envision the future of this sort of integrative process of using polis and plural voting together being?

We are already replicating this model in a different context, and our hope is that it can be adopted in many more scenarios. Hopefully this is something that can help institutions have productive exchanges with citizens and different groups of stakeholders, on a more frequent and continuous basis. These tools are all open-source, which means public servants can use them for free, and all they’d need is a bit of facilitation training. Taiwanese public servants are pioneers on this front, making use of various free citizen engagement tools and having a special role designated for this kind of facilitation called participation officers. It’s a powerful example to learn from.

And lastly, we are already thinking about how this kind of process can be integrated with AI models. There are many pathways to explore: For instance, input from small group discussions could be directly processed by an AI that would transform them into Polis statements and feed them into the Polis platform. This would help make the process more smooth for the participants, and replicable in many different contexts. AI models could also be used as a bridge between Polis and plural voting, plotting out a report with the different possible compromises that could be achieved through a vote, based on the social clusters captured by Polis. There are many elements that would need to be carefully thought about and iterated on, but this is a rich design space that could have a positive impact on civic participation. That said, it’s always good to emphasize that the key to using these tools effectively is having the kind of thoughtful, careful facilitation that helps bring everyone into the fold. These tools have been adopted by many groups focused on AI and blockchain technologies, and that’s great—but if only tech savvy people are using them, the results won’t be nearly as transformative. Our work with the CPO demonstrated that with the right facilitation they can be made accessible to a much wider range of participants.