The Ratio

In 2025 in Burkina Faso, state forces and their armed auxiliaries killed 523 civilians. The Islamist insurgents they were fighting killed 339.

In Mali, state forces and their Russian partners killed 918 civilians. The insurgents killed 232.

The ratios — drawn from ACLED data — are 1.54 to 1 in Burkina Faso and 3.96 to 1 in Mali. In both countries, the counterinsurgency kills more civilians than the insurgency. In both countries, the governments that produce these numbers came to power through military coups justified on the grounds that elected governments had failed to provide security.

On April 2, Human Rights Watch published “None Can Run Away” — 316 pages documenting 57 incidents across eleven regions of Burkina Faso between January 2023 and August 2025. Four hundred and fifty people were interviewed: 380 victims and witnesses, 70 community leaders and experts. The documented dead: at least 1,837. Of those, 1,255 were killed by state forces and Volontaires pour la Défense de la Patrie — armed civilian auxiliaries recruited and deployed by the military government. Approximately 582 were killed by JNIM, the al-Qaeda-linked coalition that controls large sections of the country’s north and east.

The report’s subtitle: “War Crimes and Crimes Against Humanity in Burkina Faso by All Sides.” Both sides are named. The government: Captain Ibrahim Traore, president and supreme commander, and six military leaders — Brigadier Generals Coulibaly, Simpore, and Diallo; Colonels Nikiema, Kabre, and Nere — identified for command responsibility. JNIM: Iyad Ag Ghaly, Amadou Kouffa, Jafar Dicko, Ousmane Dicko, Abou Hanifa.

Both sides committed crimes against humanity. One side committed them at more than twice the rate. That side is the state.

On February 25, 2024, soldiers from the Rapid Intervention Battalion arrived at the villages of Nondin and Soro in Yatenga province, roughly thirty minutes after Islamist fighters had passed through following an attack on a military base. The soldiers went door to door, ordered civilians from their homes, demanded identity cards, separated residents. Then they opened fire.

In Nondin, 44 people were killed, including 20 children. In Soro, 179, including 36 children. Satellite imagery confirmed eight mass graves in Soro, three in Nondin. A witness: “They showed no mercy. They shot at everything that moved.”

The government’s high court prosecutor visited the sites and reported an “inability to locate bodies” — despite having received reports of approximately 170 deaths. The communications minister called the accusations “baseless.”

Six months later, on August 24, JNIM attacked Barsalogho in Sanmatenga province. Over 110 motorcycles. Fighters fired on hundreds of unarmed people. HRW confirmed at least 133 dead from geolocated video analysis. French intelligence assessed the toll at approximately 600.

The people JNIM fired at were digging a trench.

Soldiers stationed in Barsalogho had forced male residents to construct defensive fortifications — without payment, under threat of beating. JNIM attacked while the construction was ongoing. The civilians were in the kill zone because the military had ordered them there. A 52-year-old farmer: “Soldiers came to my door and ordered me to go dig the trench… they beat me with a rope.”

Traore issued no public statement for more than a week.

In 2006, Stathis Kalyvas published The Logic of Violence in Civil War, which established why counterinsurgent forces structurally produce higher civilian casualty rates than the insurgents they fight. The mechanism is informational.

Insurgents are embedded in the communities they operate among. They depend on the population for acquiescence, shelter, recruitment. Every civilian killed is a potential source turned against them. Their violence is targeted — not out of restraint, but because precision is how embedded forces survive.

State forces deployed into areas they do not control face the inverse problem. They cannot distinguish combatants from civilians. They lack local intelligence networks. Under pressure to produce results, operating without information, they treat the population as the threat. The prediction is not that states are morally worse. It is that states are less precise. Precision requires relationships. Relationships require time in territory.

The Volontaires pour la Défense de la Patrie were supposed to fix this. Recruit locals. Arm them. Deploy them in their own communities. Local knowledge, local precision.

What happened is documented in ICG Report 313 (December 2023): VDPs were recruited along ethnic lines, deployed without training or supervision, and used their authority to settle preexisting grievances — land disputes, grazing conflicts, ethnic tensions that preceded the insurgency by decades. The gap between the state’s stated intent and the VDPs’ behavior is not a failure of implementation. VDPs are cheap force multiplication. Their violence does not appear in military casualty reports. Responsibility diffuses.

The result goes beyond the ratio. HRW uses the words “ethnic cleansing.” VDPs and soldiers systematically attacked Fulani communities, forcibly displaced entire villages, and in documented instances issued explicit calls for extermination. A recorded VDP statement: “Nobody will escape! Look for the Fulani everywhere. We are going to kill all Fulani.”

Lieutenant Aziz Pacmogda, head of the presidential guard, to a kidnapped civil society activist in September 2023: “People say that this is discrimination, but we are going to kill them all!”

The counterinsurgency became the ethnic cleansing. The VDP model did not solve the information asymmetry. It weaponized it.

The pattern is not contained to one country.

In Mali, where the military has held power since 2020, Alexander Thurston documented the trajectory for the OECD-SWAC: before the coup, Malian state forces had never killed more than 100 civilians in a single year. In 2022, they killed 910. ACLED data from 2022 showed that 71 percent of Wagner Group engagement in political violence in Mali took the form of violence targeting civilians.

In Niger, where the military seized power in July 2023, HRW documented a drone strike on the village of Kokoloko in Tillaberi on January 6, 2026. Seventeen civilians were killed — eleven women, two men, four children aged five to ten. Between July 2023 and November 2024, ACLED recorded more than 600 civilians killed by drone strikes in approximately 30 Tillaberi localities.

Three coups. Three countries. All justified on security grounds. In all three, the state kills more civilians than the insurgency it claimed power to defeat.

Four hundred and fifty people spoke to Human Rights Watch. Every one of them is anonymous. HRW withheld all names due to witnesses’ extreme anxiety about retaliation. The anonymity tells you who they fear.

Samer — a pseudonym — returned to his village of Bouro in the Sahel region after a military operation in December 2023. “I found my family and my neighbor’s family massacred,” he said. “My wives and my children were dead.” Nineteen of his relatives had been killed during Operation Tchefari 2, which destroyed approximately sixteen villages near Djibo, killing over 400 civilians.

A 35-year-old witness: “My two daughters died on the spot… they slit their throats.”

A 50-year-old man in Solenzo during Operation Green Whirlwind 2, early 2025: “The military and the VDPs just prepared an operation to cleanse the area of all the Fulani.” The verified dead from that operation include 32 children aged one month to seventeen years. Thousands of Fulani families fled into Mali.

The scale: 2.3 million displaced — roughly ten percent of Burkina Faso’s population.

Philippe Bolopion, HRW’s executive director: “The scale of atrocities taking place in Burkina Faso is mind-boggling, as is the lack of global attention to this crisis.”

The second clause is the one that holds. The ratio was 1.54 to 1 before the report was published and it will be 1.54 to 1 after. The mechanism that produces it — information asymmetry, ethnic delegation, institutional impunity — is operating in villages where no one is counting, producing deaths that will not appear in the next report because the next report has not been commissioned. The documentation catches up to the reality. The enforcement never does.

Sources