Won’t a participation platform make polarization worse?

A participation platform like My Vote Forward could increase polarization if it rewarded people for picking sides in public, showed running scores of “us versus them,” or kept people inside their own group. Survey data show that division between American partisans grew sharply between 1994 and 2014, and cross-country research finds the United States had the largest rise in partisan hostility among twelve wealthy democracies studied. My Vote Forward is designed around priorities rather than team positions and keeps individual opinions private by default, but no evidence yet shows that this design reduces polarization; it is a hypothesis to be tested.

The objection in its strongest form

The objection to My Vote Forward on polarization runs like this. Every new channel for political expression in recent decades, from cable news to social media, was introduced with promises of better-informed citizens, and critics argue that many of them ended up sorting people into hostile camps. A platform that asks people to rank subjects and register opinions between elections adds one more arena where political identity is performed. Even if individual opinions are private, published group results will be read as a scoreboard: “our side ranks immigration first, their side ranks climate first.” Commentators will turn every release into a story about division. Activists will mobilize their side to move the ranking. The result, the critic says, is a permanent campaign with a new set of numbers to fight over.

A sharper version adds a structural point. Elections happen at intervals partly so that public life can return to governing between them. A tool designed to make participation continuous removes that pause. People who are already the most intense partisans are also the people most likely to participate in any voluntary political exercise, so the tool’s results will overweight the most polarized voices and present them as “the public.”

Why the objection matters

Polarization matters for My Vote Forward because the framework’s purpose is to make public priorities more visible and more usable by institutions. If the act of measuring priorities makes people more hostile to one another, the framework would damage the thing it claims to strengthen. The concern also matters because a nonpartisan civic project can be captured by partisan framing simply through how its results are reported. A project that cannot show its results resist that framing has no special claim to be neutral.

Defined terms

• Ideological polarization: the distance between groups in their policy positions, for example how far apart the typical Democrat and typical Republican are on a set of issues.
• Affective polarization: dislike and distrust of the opposing party and its supporters, as distinct from disagreement on policy.
• Sorting: the process by which people’s party identity lines up more closely with their ideology, even if the overall spread of opinion does not change.
• Deliberation: structured discussion of an issue with balanced information and a chance to hear other views. See deliberation.
• Priority: in the My Vote Forward framework, a judgment about which subjects deserve attention, separate from a position on what to do. See priorities.

Evidence

Polarization in the American public is measurable on both dimensions. Pew Research Center’s 2014 survey of 10,013 adults found that the share of Americans expressing consistently liberal or consistently conservative opinions doubled from 10% in 1994 to 21% in 2014.⁠¹ The same report found that very unfavorable views of the other party rose from 17% to 43% among Republicans and from 16% to 38% among Democrats over the same period.⁠¹ Those are partisan feelings, not policy positions, and they rose faster than the ideological measure.

The rise is not universal across democracies. Levi Boxell, Matthew Gentzkow, and Jesse Shapiro measured affective polarization in twelve OECD countries over four decades. In their baseline estimates, the United States experienced the largest increase; five countries experienced smaller increases and six experienced decreases.⁠² That finding matters for this objection because it suggests polarization is not an automatic consequence of new communication technology, which spread across all twelve countries, but depends on institutions and context.

Research on social media gives mixed results. A randomized experiment by Hunt Allcott and colleagues paid users to deactivate Facebook for four weeks before the 2018 United States midterm election. Deactivation “reduced both factual news knowledge and political polarization” and increased subjective well-being.⁠³ The trade is instructive: the same exposure that informed people also divided them. Other work cautions against assuming that exposure always changes minds. A study of the Russian Internet Research Agency’s Twitter campaign in 2016 found exposure was heavily concentrated, with 1% of users accounting for 70% of exposures, and found “no evidence of a meaningful relationship” between exposure and changes in attitudes, polarization, or voting behavior; the authors note their estimates are observational.⁠⁴

The most direct evidence that a participation format can reduce division comes from deliberation research. In “America in One Room,” more than 500 registered voters from across the country deliberated over a long weekend on five major national issues, with a control group answering the same questions. James Fishkin and coauthors report that deliberators showed “large, depolarizing changes in their policy attitudes” and “large decreases in affective polarization”.⁠⁵ That was an intensive, structured format lasting a full weekend. It does not show that a lightweight online tool would have the same effect.

How comparable systems have handled it

Existing participation programs handle division in different ways, and the differences are informative.

• Deliberation before measurement. Deliberative polls such as America in One Room measure opinion after in-depth discussion over a long weekend, compared with a control group.⁠⁵ The design treats sustained exposure to other views as the remedy. It is expensive and reaches a small sample.
• Mapping agreement rather than conflict. vTaiwan, covered in the vTaiwan case, used the opinion-mapping tool Pol.is, which groups participants by how they respond to short statements, so that areas of agreement as well as disagreement become visible. The case page discusses what that did and did not achieve.
• Random selection rather than self-selection. Citizens’ assemblies, such as those in Ireland, draw members by lot so that the most intense voices do not dominate by showing up.

None of these is a tool that millions of people use continuously. The evidence for depolarizing effects comes from formats that are slower, smaller, and more structured than any mass online system.

Competing interpretations

One interpretation holds that polarization is driven mostly by elite behavior, party competition, and media business models, so a civic tool is neither the cause nor the cure. On this view the worry is overstated: a priorities tool is too small to matter either way.

A second interpretation holds that the format of participation shapes how people see each other. Asking “what matters most to you?” rather than “which side are you on?” may surface overlap that partisan framing hides; two people who disagree on immigration policy may both rank housing first. On this view a priorities-first design could reduce perceived division even if it does not change anyone’s position. That is a plausible hypothesis, not a demonstrated result.

A third interpretation is skeptical in the other direction. It holds that any public number about group preferences becomes a weapon, and that the framework’s separation of priorities from positions will collapse in practice because commentators will infer positions from priorities. The Facebook deactivation experiment’s finding that informative exposure also polarized people,⁠³ is consistent with this worry.

What the framework answers

My Vote Forward’s design responses to polarization are specific and limited.

• Priorities are not positions. The framework records which subjects people think deserve attention, separately from what they think should be done. A ranking of “housing” does not reveal a view on zoning. See priorities.
• Individual opinions are private by default. No one sees a feed of named people’s positions, and there is no public profile. Sharing is per item and opt-in. See opinions.
• No individual scoreboards. Releases report group results with methods and denominators, never leaderboards of people or groups.
• Overlap is reportable. A release can show where priorities are shared across groups, not only where they diverge, provided group sizes pass privacy review. See privacy.
• Releases are scheduled, not live. Results are published on a schedule as published releases, not as a live tracker that rewards moment-to-moment mobilization.

Hypothetical example. A community pilot publishes a release showing that participants who describe themselves as conservative and participants who describe themselves as liberal both place infrastructure among their top three subjects. The release reports the group sizes, the question wording, and the fact that participants were self-selected. It does not claim that the community “agrees” on infrastructure policy, because priorities say nothing about which projects people favor.

Where the objection still wins

The objection wins on two points. First, nothing in the framework prevents commentators, campaigns, or the operator’s own critics from presenting results as a partisan scoreboard. Design choices control what is published, not how it is retold. Second, the framework has no evidence that its design reduces polarization. The deliberation results it can point to come from an intensive deliberative format with a control group⁠⁵ which the demo does not resemble. Self-selected participation is also likely to overrepresent people with strong views, which is the core of the critic’s structural point. See what if most people never participate?.

Unresolved questions

• Does asking about priorities before positions change how participants perceive people on the other side? This could be tested with a randomized design.
• Should releases report results by self-described political group at all, given the risk that such breakdowns feed scoreboard framing?
• Can a lightweight online format borrow any of deliberation’s depolarizing features without its cost?
• How should the framework respond when a release is misreported as evidence of partisan conflict?

Related

• Polarization
• Misinformation
• Deliberation
• Priorities
• Opinions
• Can public priorities be manipulated by media?
• Hard questions