Magic Numbers Are No Shortcut to Strategy 

The past six months have given rise to the first signs of popular resistance to the second Trump administration. In early June, when immigration agents in Los Angeles escalated workplace raids, organizing by migrant rights groups exploded into mass resistance. Crowds gathered trying to protect community members, and soon afterwards demonstrators massed at the Federal Building and Courthouse, demanding a stop to the deportations. Protests spread, with some participants confronting riot police, blocking roads, and setting fire to Google’s driverless robotaxis. Trump called in the National Guard and then the Marines to put down the uprising.

The fiery images of LA reinvigorated old debates over protest strategy, often centered on questions of violence and nonviolence. Did the LA rebellion represent protest escalating to real resistance, or had it given Trump the pretext he needed to escalate repression? 

Days after the Marines descended on LA, coordinated “No Kings” rallies protested Trump’s birthday parade. Unlike the LA uprising, No Kings was generally calm, peaceful, and nation-wide. By participation numbers, it was one of the largest protests in United States history, with nearly 2% of the country’s population in the streets. As Pod Save America and many others observed with excitement, this came close to hitting the “magic number” for creating change. This month, protest numbers in the second national No Kings demonstration inched closer.

In their widely referenced book Why Civil Resistance Works: The Strategic Logic of Nonviolent Conflict (Columbia University Press, 2011), political scientist Erica Chenoweth and then-US State Department strategist Maria Stephan presented statistical analysis showing that nonviolent resistance was the most effective means of toppling governments. More than that, according to their research, every nonviolent movement that mobilized 3.5% of the population on a consistent basis succeeded in overthrowing the political regime they were resisting.

The “3.5% rule” garnered substantial media attention, and before long became something of a guiding beacon for many organizers in progressive and leftwing politics. I’ve been present in trainings, retreats, and meetings where someone has suggested literally doing the math for target populations to discover exactly how many people we’d need to turn out in order to win on an issue, according to the 3.5% rule.

The idea that strictly nonviolent protest is the best or only means of resisting tyranny provides us with a simple and relatively safe story of resistance, as does the “magic number” attached to it. In this story, the most accessible types of activism, involving relatively less transgressive, less illegal, and more funder-friendly tactics, happen to also be the most effective route to political change. More importantly, it’s backed by scientific research that claims to verify that it works, complete with an achievable numerical benchmark that we can build towards.

In reality, however, the “3.5% rule” rests on shaky evidence and bypasses crucial questions about movements’ politics and depth of organizing, as well as exactly how mobilizations exercise leverage. Ignoring these limitations can hinder movements’ ability to build grounded strategy at a time when we need it the most. Worse, the insistence on purely nonviolent protest can play into the regime’s narrative of “good” versus “bad” protesters and unnecessarily provoke bitter, splitting debates among movements. Uncomfortable as it may be, we can no longer afford the convenient untruth that real-world resistance to tyranny can escalate solely through peaceable demonstrations.

Strategic nonviolence

In previous eras of political struggle, arguments for nonviolence revolved around a combination of moral and spiritual considerations in addition to political ones. For its most famous avatars, Mohandas Gandhi and Martin Luther King, Jr., nonviolence was at once an expression of religious belief, a method of empowering individuals to action, and a route to political change.

Sometime after the international revolutionary moment in the 1960s and ’70s was crushed, organizing conversations shifted toward raw efficacy: what strategies and tactics actually work? Political scientist Gene Sharp developed the most influential argument for nonviolent action not as a moral good, but as a cold, hard strategy, claiming it was the best way to build power from below. But Sharp’s writing, and studies of strategic nonviolence that have built on it, focus entirely on the method, abstracting it from political principles and historical context. 

This shows in Chenoweth and Stephan’s research, which is based on a dataset called Nonviolent and Violent Conflicts and Outcomes (NAVCO), comprising data on hundreds of maximalist campaigns dating back to 1900. (“Maximalist” here means extralegal attempts to overthrow a government, secede from a territory, or oust a foreign occupation—not just turning out voters for the next scheduled election) 

Chenoweth and Stephan’s research is most famous for the claim that nonviolent campaigns are about twice as likely to succeed in overthrowing a government as violent campaigns. Statistically, however, the relationship between success and nonviolence in their data actually hinges on the size of the movement, suggesting that the effectiveness of predominantly nonviolent movements is driven by their ability to amass higher levels of participation. Even though it was tacked onto the argument after Why Civil Resistance was published, the 3.5% idea is an extension of the book’s central claim: nonviolent movements are larger, and larger movements have better chances of winning.

The “rule” that isn’t

The 3.5% number is a statistical happenstance, not a rule. First, many of the participation numbers in NAVCO are based on rough estimates that amount to guesswork, especially as we go back in time, and they struggle to account for types of participation that do not involve mass protests and public actions. Second, the “magic number” narrative mistakes correlation for causation. If there is a historical numerical benchmark for movement participation, it is an incidental byproduct of effective grassroots organizing combined with the right conjunctural moments—not a cheat code that short-circuits authoritarian governments as soon as the correct number of people hit the streets.

In the NAVCO dataset, the eleventh largest uprising was Palestine’s in 1990. The “peak year” of the First Intifada is classified as “primarily nonviolent” and as failing to achieve its objective. According to Chenoweth and Stephan’s estimation, that movement mobilized 3.45% of the Palestinian population that year. The uprising ultimately ended with the 1993 Oslo Accords, but taken literally, the 3.5% rule would imply that had the Palestinian resistance mobilized 152,000 people in 1990 instead of 150,000, they would have been successful in overthrowing Israeli occupation altogether. Clearly, movement histories don’t work that way.

Today’s urban-based civil revolts are just as capable of overthrowing political regimes as last century’s armed guerrilla struggles, if not more so. Mass protests are a major factor, but numbers do not tell the whole story. Who protesters are and what they are doing matters at least as much as how many there are, combined with a variety of other social, economic, and geo-political factors. Chenoweth themself has acknowledged that the 3.5% number is a “descriptive statistic” and “not necessarily a prescriptive one,” but they continue to call it the “3.5% rule” and have done little to publicly discourage its prescriptive use.

To say that larger movements are likely to be more powerful is a truism. But we need to be clear that if the next No Kings protests manage to mobilize 12 million Americans—3.5% of the US population—nothing will automatically happen. Mass participation in demonstrations no doubt plays a role in creating radical change from below, but strategizing toward people power hinges on the depth of organizing bases, the political orientation of the movement, and the leverage mobilizations can create—not on raw protest numbers.

Mixed tactics in real-world resistance

Chenoweth and Stephan’s research offered empirical validation for the argument that maintaining nonviolent discipline is necessary for protests to work. The basic argument goes something like this: Nonviolent action can mobilize mass refusal to be governed, which creates powerful material leverage against oppressive governments, while violent repression of nonviolent protests is more likely to backfire on authorities.

The authors reached their conclusion by dividing the hundreds of cases in the NAVCO dataset into two categories based on the predominant means of struggle—primarily nonviolent and primarily violent—and comparing them to see which category succeeded more frequently. The NAVCO dataset doesn’t compare violent tactics to nonviolence; it compares “primarily violent” armed campaigns to “primarily nonviolent” unarmed campaigns. The difference is important. Nearly all of the “primarily nonviolent” uprisings include actions like riots, sabotage, vandalism, and physical fights with police and government-backed gangs. These forms of resistance occurred right alongside the kinds of protests and civil disobedience that we would readily call nonviolent. In other words, real-world civil resistance involves a mix of tactics.

The anti-ICE civil resistance in LA this past summer was not openly attempting to overthrow the government, and so would not be counted in NAVCO. However, based on the categories in that dataset, the fact that some protesters burned vehicles and threw rocks and teargas canisters back at police would not disqualify it as “primarily nonviolent.” (Even the assassination of Charlie Kirk, which was certainly violent, would not register in Chenoweth and Stephan’s data.) In NAVCO terms, crossing over into “primarily violent” would mean movements collectively taking up firearms, reorganizing into an army, and literally going to war against the US government, causing at least a thousand battle-related casualties in the process. 

The problem is, when this research is presented to the public, a sleight of hand occurs where “primarily nonviolent” becomes “nonviolent.” The findings then get repackaged and used to argue that any minor instance of protester violence will hurt a movement’s chances for success, usually paired with selectively retold histories of movements that erase parts of the story that don’t fit the nonviolence narrative.

For example, during a recent interview with MSNBC’s Chris Hayes, Chenoweth pointed to the Serbian uprising that ousted Slobodan Milošević in 2000—a movement that brought masses of people into the streets to topple a dictator following a stolen election—as a key example of nonviolent success. Chenoweth left out the part where protesters smashed through police lines with a bulldozer, sacked the headquarters of the national television station, and set the Yugoslav parliament building on fire.

Likewise, protesters who attacked and burned police stations across Tunisia and Egypt in January 2011 may have been surprised to learn that, according to the top experts, they were part of one of the most impressive success stories of nonviolent resistance in modern history. The Gen-Z–led anti-corruption riots in Nepal last month that forced the prime minister’s resignation, burning down government buildings and luxury hotels? Primarily nonviolent by NAVCO standards.

A fuller examination of rights-based American movements most commonly understood as nonviolent reveals a more nuanced picture as well. For example, in her book America on Fire historian Elizabeth Hinton discusses the important role of fiery urban rebellions in Black-led struggles against white supremacy. Charles E. Cobb, a former member of the Student Nonviolent Coordinating Committee, writes about how the nonviolent Civil Rights Movement was undergirded by a culture of armed community defense. The long movement for LGBTQ rights has been punctuated with moments like the Stonewall Riots and White Night Riots.

Using Chenoweth and Stephan’s work to criticize protesters who smash windows, confront police, or vandalize Teslas is misapplying the research, plain and simple. Aside from the “maximalist” criteria, which would exclude nearly all US protests, by NAVCO standards literally all social movement actions in this country since the Civil War would qualify as “primarily nonviolent.”

Embrace the messiness

When I first proposed this research project back in graduate school—unpacking NAVCO and conducting mixed-methods research to explore the nuances of protester violence—my advisor warned me I might find things that no one wanted to hear. “We would all prefer it if Chenoweth was right,” he told me.

Maybe that’s part of the problem. The argument for strategic nonviolence isn’t just something many people believe, it’s something many of us want to believe. It would indeed be comforting to know that there are simple, accessible, comparatively safe guidelines for movements, scientifically proven to be effective at creating transformative change and coincidentally in line with the norms of polite society. However, the very impulse that leads organizers to take the results of empirical studies seriously is the same one that should push us to reckon with their limitations. Put simply, protest numbers alone tell us little about the power of a movement.

Embracing the messiness means rejecting the binary of “good protester/ bad protester,” and the implicit acceptance of certain types of violence while stigmatizing others. To be very clear, when authorities feel their power is at all threatened, they are likely to accuse protests of being violent no matter what the protesters are actually doing. That absolutely does not mean organizers should abandon tactical discipline in each and every action; what it does mean is that we are on the regime’s rhetorical terrain when we demand strict nonviolence in general. There are no commonly accepted definitions of violence and nonviolence, and it would take another article, at least, to begin to explore the ways political violence manifests in the US. But we can begin by refusing the tropes that smear protesters like those in LA who set fire to robotic taxis that double as surveillance vehicles, but leave the children shackled in detention centers out of the picture.

The legacy of nonviolence philosophy and practice offers us vital lessons and confers a wealth of tactical considerations that are highly useful in many circumstances. Incorporating this body of knowledge in the struggle for a liberated society is important, but it is a dead end to imagine that building people-power in dire times can be clean, simple, and pure. Real-life resistance looks like both No Kings and the LA rebellion—and more.

Featured illustration by Kimmie Dearest