Originally published 2020-09-07.

Left: a protester at a Black Lives Matter demonstration in Portland, Oregon (Caitlin Ochs/Reuters). Right: a demonstrator hit with pepper spray at a Portland protest (Marcio José Sánchez/AP).
Yes, the large majority of Black Lives Matter demonstrations in the summer of 2020 were peaceful. A dataset from the Armed Conflict Location & Event Data project (ACLED), compiled with Princeton University's Bridging Divides Initiative, recorded more than 7,750 BLM-linked demonstrations across the United States between 26 May and 22 August 2020, and found that over 93 percent involved no violence or destructive activity by demonstrators.
That figure was widely shared at the time, often as a rebuttal to images of burning buildings on cable news. It is a solid number, but it is also easy to misread. Here is what the data actually measured, what it did not, and what else it revealed about policing that summer.
Who ACLED is and how it counts protests
ACLED is a non-profit that collects data on political violence and protest around the world, including in war zones such as Syria and Yemen. In 2020 it launched the US Crisis Monitor with the Bridging Divides Initiative at the Princeton School of Public and International Affairs, applying the same methods it uses elsewhere to the United States.
Its unit of measurement is the event, not the person. Each demonstration in a given place on a given day is logged and coded, mostly from news reports and local sources, according to whether it was peaceful, met with intervention by authorities, or involved violence or destruction such as rioting, vandalism or clashes.
That choice explains both the strength and the limits of the "93 percent" figure. It tells you how many separate gatherings stayed peaceful. It does not tell you how many people were hurt, how much property was damaged, or whether a single violent night in one city caused more harm than hundreds of quiet marches elsewhere.
What the summer 2020 data found
The key figures from ACLED's September 2020 report and its accompanying press release:
- Scale. More than 7,750 BLM-linked demonstrations in over 2,440 locations, covering all 50 states and Washington, DC. BLM made up roughly 73 percent of all demonstrations recorded in the period.
- Peaceful share. More than 93 percent of BLM demonstrations involved no violence or destructive activity. Violent demonstrations happened in fewer than 220 locations.
- All protests combined. Of more than 10,600 demonstrations of every kind, about 95 percent were peaceful.
- Pandemic protests. Over 1,000 demonstrations were linked to COVID-19, in 47 states, including more than 330 about school reopenings, along with protests over evictions and by healthcare workers.
The report was released after the shooting of Jacob Blake in Kenosha, Wisconsin, on 23 August and the release in early September of video showing Daniel Prude's death after an arrest in Rochester, New York. Both events fuelled new protests, but they fall outside the data window.
Why many people believed the opposite
ACLED's director of research, Roudabeh Kishi, summed up the gap to The Guardian: "There have been some violent demonstrations, and those tend to get a lot of media coverage." Seen one by one, the overwhelming majority were peaceful.
That is a well-known feature of news coverage: unusual, dramatic scenes are filmed and replayed far more than orderly marches. The ACLED report cited research from the University of Washington linking the difference between perceived and actual violence to political views and media framing, and pointed to a poll in which 42 percent of respondents believed most BLM protesters were trying to incite violence or destroy property.
The perception was also shaped by real damage. Unrest in Minneapolis, Kenosha, Portland and other cities did cause serious destruction in some neighbourhoods, and several people were killed in incidents connected to the protests. Both facts are true at once: most events were peaceful, and a minority caused harm that was very visible and very costly to the people affected.
What the data said about policing and force
The report also looked at how authorities responded, and here the numbers surprised many readers.
- Authorities intervened in more than 9 percent of BLM-linked demonstrations, compared with about 3 percent of other demonstrations.
- Where they did intervene, they used force, such as tear gas, rubber bullets, pepper spray or batons, in more than 54 percent of cases. Overall, force was used at over 5 percent of BLM events, against under 1 percent of other demonstrations.
- For comparison, in July 2019 authorities intervened in fewer than 2 percent of demonstrations and used force at just three recorded events. In July 2020, force was used at 65 or more.
- ACLED logged more than 100 separate incidents of government violence against journalists, in at least 31 states and DC.
One of those journalists, freelance photographer Linda Tirado, was permanently blinded in one eye after being hit by a projectile in Minneapolis in May 2020.
The report also flagged a rising number of armed individuals and non-state groups, including militias, engaging demonstrators, and more than 360 counter-protests, about 12 percent of which turned violent.
Deaths and violence after the data window
Some of the most serious violence of 2020 happened just after the period ACLED analysed. On 25 August, during unrest in Kenosha, 17-year-old Kyle Rittenhouse shot and killed Joseph Rosenbaum and Anthony Huber and wounded Gaige Grosskreutz. A jury acquitted him of all charges in November 2021 after he argued self-defence. On 29 August in Portland, Aaron Danielson, a supporter of a right-wing group, was shot dead by a self-described anti-fascist.
During the summer, viral social-media posts also circulated claiming that BLM protests had caused dozens of deaths, thousands of injured officers and billions of dollars of damage in a single sweep of numbers. Such posts usually blended figures from different sources, or invented them, and were not backed by any single reliable count. A good rule of thumb: if a dramatic total has no named dataset or researcher behind it, treat it as unverified.
How to read "93 percent peaceful"
The most honest way to use the ACLED figure is as an answer to one specific question: if you picked a BLM demonstration at random from that summer, how likely was it to involve violence? The answer was less than 7 percent.
It is not a measure of total harm, and it does not cancel out the damage that did happen. What it does show is that the movement, as it appeared on thousands of streets in towns large and small, was overwhelmingly a peaceful one, and that authorities used force against it far more often than against other kinds of protest. ACLED itself called its numbers a snapshot, and that is the right spirit in which to read them.
Sources: ACLED and Bridging Divides Initiative (September 2020): Demonstrations and Political Violence in America, New Data for Summer 2020; ACLED press release: US Crisis Monitor releases full data for summer 2020; Wikipedia: Kenosha unrest shooting. Kishi quote as reported by The Guardian, 5 September 2020. Photos: Caitlin Ochs/Reuters; Marcio José Sánchez/AP.

COMMENTS