The Jan. 6 attack on the U.S. Capitol has provoked debates about the power of white supremacist groups, the erosion of U.S. democracy, the dangers of social media, and the bias and failure of government security forces.

The attack surprised almost everyone, but it did not occur randomly. By examining the siege through the lens of studies of conflict around the world, we can understand the Capitol attack as part of a process of political violence. This process involves moving through three key conditions: 1) Violence becomes thinkable; 2) Violence becomes feasible; and 3) Restraints fail.

Violence becomes thinkable

First, for violence to escalate as it did on Jan. 6, a large enough number of people must come to see violence a legitimate means for achieving a political goal. This goal could be overturning an election outcome, sabotaging a peace deal or rectifying perceived injustices.

Violence becomes thinkable when someone or some group animates preexisting grievances through divisive narratives. If powerful political leaders legitimize and elevate these narratives, violence becomes a way of saving or defending group members from a perceived enemy. What makes this enemy so threatening that it requires a violent response? Typically, those engaging in violence perceive an enemy that threatens their material well-being and their privileged ethnic or racial status.

Over the four years preceding the Capitol attack, President Donald Trump elevated a set of narratives about saving the country from such threats as electoral fraud, the mainstream media, godlessness, the “radical left” and a corrupt government. Trump’s speech just before the Capitol attack invoked all these ideas. White supremacist claims were baked into these narratives, including Trump’s targeted allegations of electoral fraud in cities with large Black populations such as Detroit and Philadelphia, which he described as “the most corrupt political places.”

With Trump and other politicians elevating such narratives, violence became thinkable for some supporters. They saw it as a means to rectify a stolen election, reclaim power and promote their desire for a White and Christian homeland. These ideas surfaced in the dominant chant of the Capitol siege, “This is our house.”

We can find a striking parallel in Kenya. Following disputed elections in 2007, large-scale violence escalated when the opposition candidate Raila Odinga encouraged supporters to protest the results. For many, use of violence was only thinkable because longer-standing narratives, stoked by politicians, framed rivals as invaders who threatened the land and livelihood of followers. Politicians used coded language, referring to rivals as “spots” or “weeds” that needed to be removed. In a 2007 campaign speech, a parliamentary candidate — now deputy president — declared that the sitting president was “an enemy and anybody who campaigns for him is a witch.” Hate messaging also flowed through pamphlets and vernacular radio stations.

History provides numerous other examples of elites using divisive narratives to incite violence. For instance, in Bangladesh, conflicting narratives over the country’s 1971 war of independence and claims about the government’s intolerance of Islamist parties contributed to deadly election violence in 2014. Similarly, in Delhi, India, in February 2020 riots exploded after a leader from the Hindu nationalist Bharatiya Janata Party (BJP) warned police that his supporters would “hit the streets” if protesters against a controversial citizenship law did not quickly clear out from public squares.

Violence becomes feasible

For violence to escalate, it must not only be thinkable, but also feasible. Potential participants must plan and coordinate with one another, recruit others and acquire weapons and training. This concept of feasibility emerges from studies of civil war, which argue that insurgency is less about motive and more about financial and operational capacity to rebel. This ability to organize also requires a government that is unable or unwilling to de-escalate violence.

Since Jan. 6, journalists have pointed to the ways that social media, including Parler and Facebook, have created spaces for participants to organize and plan events such as the Capitol siege, and that some Trump allies may have helped enable the attack.

Participants coordinated in online forums where they discussed ride-shares, lodging, restaurants and which weapons to bring. Political violence has also become increasingly feasible in the United States as gun ownership rates rise and self-styled militia groups strengthen and proliferate.

This condition of feasibility has enabled violence to escalate in other contexts, as well. In Kenya, post-election violence in 2008 was feasible in part because politicians helped organize the violence. They hired and collaborated with armed groups, held local meetings and distributed weapons and fuel. Similarly, in India, studies have shown that political parties often coordinate with local criminals to carry out communal violence.

Restraints fail

In many situations, violence is both thinkable and feasible, but other factors restrain escalation. Police make arrests, the military intervenes, leaders call for moderation or fears of prosecution keep people at home.

The attacks on the Capitol were so alarming in part because a presumed restraint on potential violence — the Capitol Police — failed to stop the rioters. As many have noted, this failure contrasted starkly with the heavy police response to largely peaceful Black Lives Matter protesters during the summer of 2020. The breakdown of societal or institutional restraints on violence is a function of both capacity and political choice. But other important forces did help restrain further violence at the Capitol, including protesters who decried their fellows’ window-smashing rampages.

We can find several other examples of restraints failing and enabling the escalation of violence. During the Rwandan genocide, many highly influential churches that could have protected civilians actually facilitated violence against them. During Kenya’s post-election crisis, the police were overwhelmed by the violence and were often unable or unwilling to stop the attacks. Law enforcement also escalated electoral clashes, sexually assaulting civilians and shooting into crowds of protesters.

Similarly, there are many examples of police bias during communal riots in India, including evidence that law enforcement contains such violence only when it is in ruling politicians’ interests.

What does this mean?

To be sure, none of these conditions alone predicts violence. But large-scale violent events are more likely where we find all three conditions, as in the Capitol attack.

Those who seek to prevent future violence may wish to pay attention to the slow-moving shifts in political discourse that can transform violent talk into action. By doing so, they can more easily identify potential moments for intervention and de-escalation.

Kathleen Klaus (@KathleenKlaus) is an assistant professor of politics at the University of San Francisco and author of the book “Political Violence in Kenya: Land, Elections, and Claim-making” (Cambridge University Press, 2020).

Aditi Malik (@aditimalik_) is an assistant professor of political science at the College of the Holy Cross. She studies political violence, political parties, ethnic politics and social movements in sub-Saharan Africa and South Asia.