Polis: A Breakthrough tool for broader, deeper consensus

Polis is an online survey platform where all the items being surveyed are created by the participants, whose responses to each other’s offerings are automatically organized in a way that reveals evolving points of consensus and division both for the participants to see and respond to, and for organizers to help deepen dialogue among stakeholders involved with the issue in question. I know of nothing quite like it and I see tremendous transformational potential not only in its sophisticated off-the-shelf version, but in innovations seeking to use it in creative ways.  In this article I go into some depth about all this, so it’s a bit long.  But I hope any time and attention you invest in reading it will be amply paid back in inspiration and, if you are a democratic theoretician or practitioner of democratic arts and public engagement, as a resource of incredible power. – Tom

Polis (sometimes written as pol.is, its website form) is an online participatory polling platform with oddly transformational potential. It offers all participants two simple ways to respond to a question from the pollsters: (1) they can submit their own short responses and/or (2) they can vote “agree”, “disagree” or “pass” on such statements submitted by other participants.

So far, this description seems typical of a number of online participatory polling platforms.  Polis is unique, however, in that it automatically sorts participants into 2-5 clusters, each of which shares a certain perspective that differs from the other clusters. 

It’s a bit like Amazon telling you that people who looked at the item you’re looking at, also looked at certain other items, as determined by an algorithm that has assigned you and those others to a group of seemingly like-minded shoppers. The Polis algorithm does that with your and others’ responses to the polling question.

I first stumbled on Polis several years ago while researching impressive democratic innovations in Taiwan. I’d never seen anything quite like it.

Most Polis exercises are set up – remarkably – so that the participants can view maps of the evolving “ecosystems” of different perspectives as those ecosystems shift in real time. The two mapping modes are graphic (using an x/y axis) and “bee swarm” diagrams (statements as dots laid out in a spectrum from most agreeable to most controversial.  You can see examples of these in the links immediately below.  

Participants can also view colored bar charts showing how many of them voted agree, disagree, or pass on each statement – both overall and in each of the different clusters.  

Clicking on a point representing a statement in either of the “maps” instantly reveals the statement and its bar graphs. 

Furthermore, participants and pollsters can see which key statements led the Polis algorithm to come up with each cluster. 

That’s a lot of information!  

As an aside, it’s an interesting example of “holopticism” because “the whole” population of Polis participants can view “the whole” evolving ecosystem of their responses.

EXAMPLE POLIS REPORTS – See a report for a big Polis exercise HERE and one for a small Polis exercise HERE

Polis reports provide incredible detail that evolves over the exercise as new statements are added and more votes are submitted. Sometimes one or more cluster’s definition and membership also shifts.

Most importantly – at least for me – a Polis report allows us to scan the agree/disagree bar charts to quickly note which statements have majority agreement especially across all the different clusters.  Those statements embody true consensus – that is, statements that widely different people agree on.  That’s a rare treasure!

When I first encountered Polis, I was blown away by this ability to reveal consensus without people even talking together!

However, the founders of Polis stress an important caveat:  As remarkable as Polis is, it is not designed for directly making decisions. And its power is wasted if it is used merely for crafting partisan messages.  It is intended to inform live human dialogue and deliberation in search of deeper understanding about where a particular population is at – collectively – regarding the question they’re responding to – and to then come up with strategies, policies of communications consistent with that and with the realities of the situation.  

For that purpose, the controversial statements – with lots of disagreement among responders – can be as valuable as the consensus statements in providing insights and highlighting challenges.  Similarly, statements that elicit supermajority “agree” votes (which is one kind of agreement) are as important as statements that elicit supermajority “disagree” votes (which is another kind of agreement).  And uncertainties – statements with lots of “pass” votes – are also meaningful.  

Sometimes demographic information (“metadata”) is solicited to enable organizers to learn what kinds of people liked or disliked what kinds of statements.

Polis produces so many levels of potential understanding that sometimes data scientists are hired to delve into the results for deeper insights than those provided directly by the fascinating automatically generated Polis reports, such as those you see in the samples linked above.  

But no matter how deep we may dig, the most potent and useful meaning will come from real people who care about the focus issue or question, talking amongst themselves about the Polis results and their own ideas, making co-creative sense of what the poll is telling them about what they might do next, and then going beyond that to continue learning together, often using Polis to feed their learning….

To me, the implications of all this for dialogue, deliberation, decision-making, choice creating, consensus-building, conflict resolution, and all other realms that address our collective affairs are profound and overwhelmingly positive.


Although the software and platform are free to use, those who plan to use them should understand some important behind-the-scenes aspects that are not immediately obvious:

1. PURPOSE:  The unique nature and purpose of a Polis exercise needs to be understood by both its organizers and its participants.  Its logic is different from the logic at work in a normal poll, survey or focus group.  We aren’t just “getting feedback” on statements we generate, nor are we looking to manipulate anybody, and we certainly aren’t highlighting any particular statements in the poll to attract special attention. What we’re doing is evoking the views and creativity of the participants, themselves, and thus we need to make it clear that their contributions are welcome, important and primary. If either the organizers or the participants don’t get that, the results can be very frustrating for all involved.

2. SEEDING:  To launch a Polis exercise, the empty field of statements needs to be “seeded” with statements provided or collected by the organizers, so there’s something there at the start for early participants to vote on.  These “seed statements” can be random brainstorms or something that organizers want “feedback” on (despite 1, above) or statements strategically designed, for example, to represent a broad range of perspectives (including those normally marginalized or taboo) to encourage participants to say whatever they want.

3.  MODERATION:  The above doesn’t mean that any kind of statement is accepted.  A statement may be irrelevant garbage that would simply distract or turn off other participants without adding any real value, e.g., trolling.  Or it may be an obvious duplicate of another statement, thus diluting or distorting the vote, or simply overwhelming other participants trying to make their way through dozens or hundreds of statements.  

    SO…  a Polis exercise is moderated.  The more participants and submissions we expect – the greater the volume and speed of submissions – the more moderators we will need.  And moderators need to be at least briefed and, ideally, trained a bit on what to look for and how to handle it.  Their work can make a tremendous difference both in the experience of the participants and in the value of the Polis outcomes.

           HOWEVER, poor or ill-intentioned moderation opens the door to distortion of the results.  Therefore, for transparency’s sake, a public/shared record should be kept of all the deleted statements including an indication of why they were removed from the poll.

4.  FRAMING:  Although the briefing for participants on how to use the interface is fairly standard, there’s a certain art to framing the activity, especially creating a question (a) that covers the ground you need covered, (b) that evokes participants’ engagement in making thoughtful contributions, and (c) that results in the KINDS of statements you need.  A bit of pre-testing may help develop something compelling that does the job.

5.  RECRUITMENT:  Each Polis exercise involves recruiting the right participants.  Sometimes it’s anyone who responds to a more or less targeted announcement or advertisement or news coverage about the exercise. Sometimes you may want to select particular types of people – perhaps the way professional surveyors gather a stratified sample of randomly selected citizens. Some Polis organizers go out into the community with tablets to engage people on the street or in some particular public venue in order to reach those who otherwise might not even hear about the opportunity, enticing them and getting them started as participants, perhaps with a bit of coaching or help.  Sometimes you may simply need to recruit participants from an organization or network you’re working with.  There are many approaches, but be intentional about your “participant audience” and creative about the best ways to engage them. If needed to sustain their participation, you may entice them with some incentives, like special opportunities, token payments, or visions of impact.

6. EVOLUTION:  If you have a sizable, adequately diverse and/or very engaged pool of participants, Polis can produce useful results in a few days or weeks.  But occasionally organizers have extended Polis exercises for a month or more, generating phenomena of particular interest to those of us in the dialogue and deliberation field.  

     As we extend the duration of a Polis exercise, the nature of participant statements tends to move from complaints and generalizations to actionable solutions, especially with some encouragement.  

     Probably the most cited example of this was the use of Polis in Taiwan to deal with controversy around Uber – see https://www.tomatleeblog.com/archives/175327886 and scroll down to “EXAMPLE 2 – THE UBER POL.IS EXERCISE”.  The Polis participants were Uber drivers and riders and taxi drivers and riders. It was incentivized by organizers saying that statements supported by 80% of all groups would be the basis for face-to-face deliberations among a group representing all stakeholders in the issue (including Uber representatives). 

     Since Polis participants could track how the statements added up at any given time, some of them started crafting statements that might attract more support and thus actually influence the final deliberations. And that’s what happened. 

     (This phenomenon is parallel to a pattern I’ve been told manifests in some extended Open Space “unconferences”, where during the first day or two participants tend to talk about the issue or situation. The next day or two they tend to talk about solutions and approaches to it. Later in the week they tend to come up with breakthrough ideas and/or start organizing action.)


Some of us are so interested in the potential of Polis that we are envisioning and testing new ways to use it.  Its natural simplicity and depth invite and enable productive experimentation.

One obvious (to me) approach would be to integrate it with various conversational processes.  For example, to delve into a topic, World Cafes (mixing participants among small group dialogues) could be convened before, during, or after a Polis exercise. World Cafe “stirs the pot” on an issue, generating a level of implicit and evolving collective intelligence. Participants from such a conversation could then join a Polis exercise (which could include other people, as well).  Both Cafes and Polis offer the capacity to engage large numbers of people, to shift and expand their thinking, and to harvest energies and insights from their engagement. In fact, Polis can be used to harvest ideas from a program that involves many different approaches to community engagement.  I’m particularly interested in using Polis results to give a deliberative citizens assembly (or other “mini-public”) a sense of what the broader public (or stakeholders and experts) think about the issue they are considering.  And a citizen council could share its draft findings and recommendations with a Polis pool for feedback, or perhaps share their final statement with a question like “What is possible now?” to help evoke action on the issue.

One colleague is testing what happens when people who are participating in a Polis exercise are invited to join a pool from which participants in a citizen assembly or other mini-public will be selected. These participants can be solicited right on the Polis user interface page by referring them to a website where they can fill out an application form providing organizers with whatever demographic, contact and special needs information are needed for the selection process.

Another approach would be to increase Polis’ already significant iterative dynamics (e.g., the ways participants respond to each other’s statements) by running a series of Polis exercises where the results of an early exercise inspire new framings, questions or seed statements for a subsequent exercise.  This process can help delve deeper into the focus issue … or help formulate statements that will nudge responses towards greater or more useful consensus…. or for some other purpose. Iteration – feeding the output of one process into the next process – can increase the speed and nature of evolution towards greater understanding and possibilities.

Finally, I envision testing what happens when we run parallel Polis exercises tapping the same or significantly different pools of participants with the same question(s).  To what extent do the results come out similar or different?  This could be combined with the interactive approach noted above, with results from one set of parallel exercises influencing how the next iteration of those exercises is framed and launched, even with the diverse exercises influencing each other.  I believe such parallel processing could surface all sorts of unexpected and useful dynamics, insights and possibilities.

I suspect this barely scratches the surface of what can be tried to enhance the transformational potential of this unusual tool….


Polis is a work of love and world-service by the folks who created it in its nonprofit, open source form.  Thankfully, they are developing increasingly clear and useful instructions for its use by ordinary people, community organizers, process designers, activists, programmers and others. They are also working to increase the platform’s already significant safeguards against manipulation and abuse.  I believe Polis promises to be one of the most potent and flexible forms of public engagement available to those working in the realms of deliberative and wise democracy.  I see it helping us deliver ever greater benefits, in terms of its insightful reports, its focus on subsequent quality dialogue and deliberation, and its potential to generate widespread engagement and sense of agency. And I’d love to see it used increasingly in a community of practice whose creative members share their experiences, their lessons learned, and their creative efforts to optimize the wisdom-generating potential of this remarkable resource.


The Computational Democracy Project

CompDem offers access to (a) pol.is itself and (b) extensive resources, freely available in their knowledge base which, together, are all an organization needs in order to implement and derive value from the Polis methodology.

What a Polis participant sees

An interesting description for beginners

Hacking Ideology: Polis and vTaiwan (a brief intro with good embedded links)

Polis in Taiwan – presentation by Colin Megill, one of the founders of pol.is 

The internet doesn’t have to be bad for democracy 

The evolution of Polis (how it evolved into its recent functionality) 

A Polis experiment in the US (The pol.is exercise was a success, although the follow-up left much to be desired….) 

Leave a Reply

Your email address will not be published. Required fields are marked *


This site uses Akismet to reduce spam. Learn how your comment data is processed.