Will There Be Violence?

Americans expect political violence far more than they say they would commit it. The fight they can see is over who they are. The contest that will decide the 2026 midterms is happening elsewhere — in federal courts, county election offices and grand jury subpoenas — and it runs…

Will there be violence? We are asked it every day now. The honest answer is that violence is possible, and there is no way to know for sure. Even the agencies that monitor violence and extremism cannot say with certainty that a particular person or group will act within a given window, and they are sometimes wrong when they try. But sit with the question, because it tells you something. It is late. With the election days away and more than 27 million Americans already through the polls by October 25th, 1 the question says less about the country than about what all of us have been watching, and what we have and haven’t done, on the way here. So this piece is a different kind of answer. Not a forecast. A look beneath the impulse to ask, and where the risk actually sits, which isn't where most people are looking. Tension is not the problem You have needs, and an agenda for your own growth. You cannot fulfill it alone, so you join or form groups: a church, a trade, a party, a neighborhood association, a company. Those groups exist to meet their members' needs. Every group ends up with leaders, elected or self-appointed, whose job is to pursue the group’s agenda. A society holds many such groups at once, some pursuing agendas and others reacting to them. That produces tension. Tension is not a malfunction. It is how a society learns. The people who govern a system have two jobs. The first is to turn that tension into learning: policy, rules, budgets, services, and opportunities that let the largest number of groups pursue their agendas successfully inside the system. The second is to manage conflict while doing so. When leaders do both, the system’s capacity for cooperation grows. When they fail at either — when groups find they cannot meet their members’ needs inside the system — those groups’ leaders turn to coercion and force. Cooperation falls. Some groups go further, into militancy. A few go into extremism. You can watch this happen. In the framework of the political scientist Monty G. Marshall, who directs the Center for Systemic Peace, a system whose political tensions go unresolved moves through recognizable stages: conventional politics, then contentious politics, then factionalism organized around issues, then factionalism organized around identity, then militancy, then open warfare. 2 Socially, you can measure the ratio of groups that use cooperation to those that use coercion and force. Where the country is On our read, the United States is squarely in identity factionalism, with the question of whether it sits in the gap between that stage and the next. Pressure is not destiny. But the gap is real, and the November midterm elections sit inside it. Part of that read is other people’s. Marshall’s Polity Project, a scorecard that has rated every government for decades, has coded American political competition as factional since 2016. It scores the country at zero on its scale, which in its own words means the United States “is no longer considered a democracy.” 3 That coding is dated January 20th, 2025, a special reading by Marshall, yet the project has been unfunded since 2020. Three funded indices that update every year all put the United States materially higher. 4 We carry the Polity Project’s zero score with those limits attached. Part of the read is also ours. On ideology, the issues have fused into two identities. We counted one sentence on X: that Democrats will destroy or burn down the country. Thirteen distinct people wrote it in their own words on October 11th. Forty-one did a week later, on October 18th. Its mirror image — that the administration is fascist or authoritarian — was written by fourteen people both times. 5 The second count fell on the peak of the riots in France; set aside every post that mentions France and the rise is still a doubling. 5 Then we read the posts. Of the twenty-seven that do not mention France, fifteen name no policy at all. 6 That is identity talking, not a complaint about anything. Nor did it come from the top. The vocabulary is on the campaign trail — annihilate, decimate, mutilate, from the President in Dalton, Georgia, on October 16th — but the sentence itself is in none of the fifteen campaign documents we read, and it was circulating eight days before the President used its framing at a rally. 7 8 It did not come down. It came up. On who runs elections, the sorting has not gone the way you might expect. Of the fifty states sent a federal demand for their complete voter rolls on July 7th, sixteen handed them over and thirty had to be sued. The sixteen are mostly Republican-governed. 9 But Idaho’s Secretary of State refused in writing. Utah’s Lieutenant Governor refused and is under federal investigation. Wyoming’s Governor asked for a state investigation of federal monitors. 10 Wisconsin’s election commission issued guidance on federal agents at the polls with two Republicans in the majority, and on October 21st a Republican candidate for Governor of Wisconsin said on a debate stage that he would certify the 2028 election regardless of who won — the only unconditional commitment to certify we found from either side. 11 The national associations of state election officials issued a joint bipartisan statement on October 14th. 12 The parties have sorted. The officials have not. Now set this beside every headline you read. Going into the 2024 Presidential election, the share of Americans who agreed that “true American patriots may have to resort to violence in order to save our country” rose from 15 percent in 2021 to 23 percent in a survey fielded in August 2023. 13 Yet when another team asked about specific acts, in surveys through January 2024, fewer than 4 percent supported assault or arson against political opponents, and about 2 percent supported politically motivated murder. 14 The country did not tip into militancy after the election. This year, a survey of the same ground — fielded May 15th to June 15th — was designed by its authors to catch a midterm-year rise. 15 In its previous wave, a year earlier, support for political violence had risen. 16 This time it found the opposite. Fifty-one of fifty-three measures of anticipating, justifying or being willing to commit political violence fell. Willingness never exceeded three percent. The share likely to be armed fell from 9.1 to 6.7 percent. The one measure that rose significantly was the expectation of civil war, from 7.1 to 8.0 percent. 15 Read those two together. More people expect violence. Fewer would commit it. The 2026 survey is four months old and cannot describe October. What a newer one measures is a different fear: asked between October 9th and 13th, seven in ten Americans said they were not confident the President would accept the result if his party lost, half of them not at all. 17 It is the newest reading on that question there is; by Election Day it will be three weeks old, and no newer one looks likely before the vote. That is fear that the system will not hold, not an appetite for violence. The freshest national poll, fielded October 23rd–26th, found 48 percent of registered voters expect the federal government to deploy agents or troops at polling places in Democratic areas. 18 Yet a separate survey fielded October 6th–14th found confidence that votes are counted as voters intend had risen in both parties since July — from 66 to 73 percent among Democrats and from 58 to 67 percent among Republicans. 19 More people trust the count, and half expect interference with it. Both are true. And the expectation keeps rising. Searches for “civil unrest” rose 241 percent in the three months to October 19th, slowly rather than in spikes. 20…