Two people smiling as they look at a laptop together at a table

Cultural Evolution / Governance / Institutional Learning

Reducing Political Hostility Through Cooperation

A cooperative online quiz game that pairs Democrats and Republicans as teammates

0US participants across five experiments

0Hour of play in a single session

0Months that some effects lasted after one game

Executive Summary

Democrats and Republicans who spent an hour as teammates with someone from the other party reported warmer feelings toward that party, and some effects lasted up to four months. The pairs played Tango, a cooperative online quiz game. Lucas Woodley, Evan DeFilippis and colleagues in Joshua Greene's lab at Harvard ran these experiments online with US participants. Anyone can now join weekly public Tango games, and by mid-2025 the team was piloting the game in Israel.

Who's behind this

Joshua Greene is the Alfred and Rebecca Lin Professor of Civil Discourse and Professor of Psychology at Harvard University. He is an experimental psychologist, neuroscientist and philosopher who studies moral judgment and decision-making. His book Moral Tribes: Emotion, Reason, and the Gap Between Us and Them (2013) asks how groups with different values can get along. He later said the book was "very philosophical and offered little practical guidance for real-world problem-solving." His lab now builds and tests practical tools for reducing conflict between groups.

Greene co-created Tango with members of his lab. The 2025 Nature Human Behaviour paper on the five experiments is by Lucas Woodley and Evan DeFilippis (joint first authors), Shankar Ravi and Greene. The public game at letstango.org was developed in partnership with the Global Development Incubator. The research was funded by Beyond Conflict, Schmidt Futures, Open Philanthropy, the Templeton World Charity Foundation, the John Templeton Foundation and Georgia State University. A National Science Foundation fellowship supported Woodley.

Key terms

Why partisan hostility is hard to reduce

Democrats and Republicans increasingly dislike and distrust each other. Greene's team wanted to ease that hostility online, including among people unlikely to sign up for cross-party dialogue.

A widening gap
From 1978 to 2020, the gap between how warmly Americans feel toward their own party and toward the other party grew by 29 points on a 0 to 100 scale. The figure comes from American National Election Studies data cited by the Tango authors.
The motivational challenge
Structured dialogue in person can reduce partisan hostility for months. It is hard to run at scale. Online interventions are easier to scale, but many people, especially strong partisans, aren't motivated to take part. Psychologists Alexander Landry and Eran Halperin call this the "motivational challenge."

How the game works

Partners chat in real time and agree on each answer together. In the experiments, both earned a bonus of 5 to 15 cents for each correct answer. The authors hypothesize that players warm to the other party when they see their partner as someone who helps them succeed. The hypothesis depends on perceived help, not on whether the partner actually raises the pair's score. The studies were not designed to fully isolate this mechanism.

Stage 1: General knowledge

Easy questions are meant to build goodwill between partners and toward the game.

Stage 2: Culture

Each question is easier for one party's supporters. Republicans are more likely to know the name of the family on Duck Dynasty. Democrats are more likely to know the name of the lead character in Stranger Things.

Stage 3: Politics

Questions affirm or challenge each side's beliefs. Republicans are more likely to know that relatively few US gun deaths involve assault-style weapons. Democrats are more likely to know that US immigrants commit relatively few crimes. These questions are meant to build trust and respect that extend from the partner to the other party as a whole.

The Tango experiments

Greene's team ran five randomized experiments with Tango between 2018 and 2024 and preregistered four of them, recording their hypotheses and analysis plans in advance. All participants were in the US and identified mainly as Democrats or Republicans. They were recruited through the online survey platforms CloudResearch and Prolific. The main comparison was between cross-party and same-party pairs.

Playing with someone from the other party raised self-reported warmth toward that party by about 7 to 12 points on a 0 to 100 scale, compared with playing with a same-party partner. The authors estimate that a gain of about 9 points rolls back about 15 years of rising dislike between the parties, at least momentarily. Players also divided $100 between an unknown Democrat and an unknown Republican, and a lottery decided which allocations were paid out. Players with a partner from the other party allocated about $8 to $13 more to the other party's member than players with a same-party partner did.

Two experiments followed players for four months. The full game, with its political questions, still increased giving to the other party four months later in both experiments, and still increased warmth in one. The authors note that the repeated follow-up surveys may have reinforced the effects. Players also became more willing to say the other party can make valid points, and the effect persisted at the one-week follow-up. Effects on support for undemocratic candidates were inconsistent across experiments.

Which parts of the game mattered

Only the two-player game, played with a live partner from the other party, had statistically significant effects on warmth and giving toward the other party. A chat with no quiz and a one-player version of the game did not.

Row	What players did	Result, compared with a same-party partner
Chat only (Experiment 1)	A five-minute get-to-know-you chat with a partner from the other party, with no quiz	No significant difference in warmth toward the other party
One-player game (Experiment 4)	The quiz played alone, with answers from an earlier player of the other party	No significant difference in the main measures of warmth or giving
Two-player game (Experiments 1 to 3, and 5)	The quiz played in real time with a partner from the other party, with or without political questions	Warmer feelings toward the other party in all four experiments, and more giving to its members in the three that measured giving

What the results suggest

The authors think games like Tango might reach people with no interest in cross-party dialogue. None of the players had been recruited for such an interest, and players from both parties rated the game highly enjoyable. The authors caution that the participants were experienced survey takers who may differ from the wider public, and that all the experiments were in the US. Greene values constructive dialogue. He says it depends on "that basic sense of mutual respect and openness," of thinking "this person is on my team."

Next steps for Tango

Greene and Scott Warren of the Global Development Incubator led a Templeton World Charity Foundation project (2023 to 2026) to test the game with new populations.

1. Other countries
The project set out to test the game in India and Israel. By mid-2025 a pilot was underway in Israel, and the team was writing questions for India and Northern Ireland.
2. The public
The project also planned to test the game in everyday online settings in the US, including with people who hold extreme political views. Anyone can join the weekly public games at letstango.org. Each lasts 15 minutes, compared with an hour in the experiments.
3. Campuses and workplaces
By mid-2025 the team had run Tango with thousands of undergraduates on US campuses and had finished a first trial with employees of a Fortune 500 company. The researchers also want to test whether repeated play makes the effects last longer or become larger.

Published by Center for Human Progress (US) Inc., a 501(c)(3) organization