They Asked for Peace on a Thursday. On Sunday, Kenscoff Was Burning

Letter, massacre, hostages, arsenal: a negotiation by coercion less than four months before the election.

On 20 August, a letter circulated under the letterhead “Gwoup Rebèl 400 Mawozo.” In it, Wilson Joseph, alias Lanmò San Jou, raised a “white flag,” called for a truce, and asked for a dialogue under international facilitation. Three days later, during the night of Sunday 23 to Monday 24 August, Kenscoff was struck: 47 dead according to the United Nations Integrated Office in Haiti (BINUH), including at least five children, more than fifty people abducted, dozens of houses set on fire. A video then showed captives, while a man calling himself “Izo 2” addressed the director general of the Haitian National Police (PNH) by name and announced that he would kill them if any of his men were hit. The general election, the first in a decade, is set for 13 December.

Taken separately, these acts contradict one another. Why call for a truce, then strike? Why abduct dozens of people and threaten to execute them? They stop contradicting one another as soon as they are read as parts of a single sequence: a negotiation by coercion. The letter seeks an interlocutor. The massacre demonstrates a capacity to do harm. The hostage-taking constrains the response. They ask to talk while fixing in advance the balance of power under which any talks would take place.

Three levels of certainty must be kept apart. The assault, the deaths, the abductions, the order to investigate given to the Central Directorate of the Judicial Police (DCPJ), and the threat made on video are reported by corroborating public sources. The letter is circulating without having been authenticated by any institution. The videos of weapons laid out in rows, released afterward, establish neither their date, nor their location, nor the claimed American origin. The analysis therefore concerns the message that these documents put into circulation.

One difficulty must be stated at the outset. The letter is signed by 400 Mawozo. The assault is attributed to the coalition Viv Ansanm (“living together” in Creole). The threat comes from a man claiming allegiance to Johnson André, known as Izo, whose base is in Village de Dieu. Three signatures for a single sequence. It cannot be inferred that a single command ordered these acts. But the opposite does not hold either: Viv Ansanm is precisely the structure that has made these groups capable of acting together, and what defines a coalition is that it produces convergent effects without a single plan. The reading offered here concerns effects, not intentions.

The sequence, 20 to 26 August

Table 1. The sequence from 20 to 26 August 2026, and the degree of certainty attached to each item.

A letter less about peace than about status

The choice of words is central. The text does not speak like a gang. It speaks of “rebels,” of “territories,” of truce, of mediation, and of the law of war. A bandit is first and foremost a matter for the police. A terrorist is a threat to be isolated. The word “rebel” opens up a different frame, that of the ceasefire and the political settlement. Legally, this self-proclamation creates nothing: calling oneself a rebel confers neither belligerent status nor a right to a truce. What this vocabulary can produce is of another order: a shift in the category through which the group is read, enough to make it thinkable to treat the group as a counterpart, a standing that the law does not grant it.

The rest of the text extends this shift. The letter proposes a truce and a dialogue, but not disarmament. It asserts that the groups will remain armed, demands respect for the boundaries of the territories they control, and disputes the PNH’s authority in those areas. The truce looks less like a surrender than like a freezing of positions.

If the aim is to freeze the lines of control tomorrow, every position seized today becomes more valuable. Nothing proves that the attack of 23 August was decided with the letter of the 20th in mind. But in the logic of a territorial negotiation, conquering before asking for the freeze is not contradictory at all. Note also that no public response followed the letter for three days, and that a public offer that is ignored leaves its author under pressure to restore his credibility. That may be part of what played out between 20 and 23 August, though it cannot be established.

Kenscoff, moreover, is no ordinary municipality. Its heights command the approaches to Pétion-Ville and to the road to the South. BINUH and the Office of the High Commissioner for Human Rights had already noted this strategic value in their April 2025 report, which counted at least 262 dead in two months of offensives. The PNH statement of 25 August adds a decisive fact: Director General André Jonas Vladimir Paraison ordered specialized units reinforced and tasked the DCPJ with establishing whether the attack involved complicity or negligence. The police do not say that these facts are established. They acknowledge that the collapse of the security arrangements warrants an investigation, which is in itself revealing.

Hostages as a veto on the response

More than fifty people were carried off during the assault. They are neither just more victims nor simply potential sources of ransom. Their captivity works as a means of military pressure.

The threat is worded more broadly than it appears. The man on camera warns the PNH director general against using drones in the area, and announces that he will kill all the captives if any of his men is hit. So this is not a veto on one particular capability. It is a veto on any operational result, whatever the instrument. The choice of target is worth noting: since March 2025, a task force run by the Office of the Prime Minister has been using armed drones against the groups in the metropolitan region. It is the clearest tactical advantage the state has. What cannot be neutralized militarily can still be made politically costly to use.

The trap works both ways, as Romain Le Cour of the Global Initiative Against Transnational Organized Crime has noted: if the state holds back, it has failed to protect; if it acts and captives die, those deaths will be laid at its door. That is what distinguishes this hostage-taking from an ordinary kidnapping for ransom. Captives are no longer valued for the money they bring in, but for the action they prevent.

What coercion is trying to buy

What these groups seek is not territory. It is to convert a position of military strength into political gain. Territory is not legitimacy. A gun does not confer an amnesty. A roadblock does not earn a seat at the table. To turn armed control into status, another actor has to accept the conversion. The groups can conquer, tax, and abduct. They cannot grant themselves a political way out.

Hence the choice of addressees. The letter bypasses the Haitian state and is addressed to the Organization of American States (OAS), CARICOM, BINUH, the International Committee of the Red Cross (ICRC), and human rights organizations. The Haitian state cannot grant status without contradicting itself. Nor does an international third party grant it: the OAS, BINUH, and the ICRC know how to talk to an armed group without recognizing anything, and the ICRC has built its practice on that distinction. What repeated contact can establish is of another order: not recognition, but normalization. A group that is received, whose demands are passed on, whose exchanges are facilitated, gradually becomes an interlocutor. This de facto legitimation is never granted; it accretes.

This conversion now has a date. On 13 December, Haiti is due to hold its first general elections in a decade. An election is the moment when power is legally redistributed, when guarantees are negotiated, when candidacies are shielded, and when the question of amnesty stops being theoretical. In December 2025, the International Crisis Group noted the coalition’s wish to place allies in the administration and to obtain a broad amnesty. Whether the letter was designed with this deadline in mind cannot be established, and the argument does not depend on it. It is enough to observe that it arrives in the window when normalization, even informal, would be worth the most.

What to do about the hostages

Three questions arise at once. Should the state launch an assault? No. Should these people be abandoned? No. Should everything be put on hold? Not that either. The answer comes in three stages: locate and contain, impose a single channel, keep the final option conditional.

The assault first, because it is the deceptively obvious option. On 5 May 2003, in Colombia, the army approached the camp where the FARC were holding Governor Guillermo Gaviria and former minister Gilberto Echeverri: the guards executed both men and eight captive soldiers before the first contact. Conversely, the two successes everyone cites required what Kenscoff does not offer. In Lima, Operation Chavín de Huántar freed 71 of the 72 hostages held in the Japanese ambassador’s residence, but after a 126-day siege, at a single site, mapped through months of eavesdropping and tunneling; Operation Jaque, which brought out fifteen FARC hostages in 2008 without a shot being fired, crowned years of infiltrating the guerrillas’ communications. These precedents yield no mechanical rule. They trace an operational constant: the less the captives are located and grouped together, and the less the adversary’s setup has been penetrated, the more an immediate intervention adds to the uncertainty and to the risk to them. The Kenscoff captives are scattered, unlocated, on hostile ground. The assault is therefore ruled out not by the kidnapper’s threat but by the state of intelligence, which points to the task of the first days.

First stage: locate and contain. No option will ever exist without knowing where the people are, and nothing in the kidnapper’s condition forbids looking for them: observing hits no one, and the means that the threat sought to freeze remain available. Then contain, rather than go in, because the real danger of the first days has a name: Chibok. In 2014, the 276 schoolgirls abducted by Boko Haram were split into small groups and dispersed within days; some were never found. The risk is not so much that the captives stay in Kenscoff as that they are dispersed toward the metropolitan area, where they would become impossible to find and could be cashed in one by one. If containment holds, time changes sides: fifty people are costly to feed, to guard, and to keep quiet, and a group holding them loses the mobility that is its strength.

Second stage: a single channel. There is no need to look abroad for a precedent: 400 Mawozo wrote it themselves with the seventeen missionaries abducted in October 2021. Two released in late November, three in early December, the escape of the last twelve in mid-December, a partial payment reported in the press. This precedent offers no model for an exit, as the final escape shows. What it establishes is capabilities: holding a group in captivity for weeks, staggering the outcome, negotiating over money, moving the captives. The state therefore knows what the duration will look like; its task is to organize that time to its own advantage. Not by centralizing any ransoms: since May 2025, Viv Ansanm has been designated a foreign terrorist organization by the United States, and any transaction exposed to the American financial system becomes legally perilous, for intermediaries and banks alike. What the state can centralize is everything else: the reporting of every contact, the conduct of exchanges, support for families, the exploitation of intelligence. Colombia gave this principle institutional form with the GAULA, which bring intelligence, investigation, operations, and prosecutors together in a single circuit. The fall in kidnappings, from more than 3,500 in 2000 to a few hundred in 2008, owed much to the military turnaround of those years, but that circuit played its part, and it is the circuit that can be transferred: every contact reported, every person released and debriefed in time, yields a piece of the adversary’s plan.

That leaves the humanitarian third party. The state does not hire the ICRC like a contractor; it can request its involvement, which follows its own principles and requires the consent of the parties. At Chibok, the eighty-two releases of May 2017 came about through Swiss facilitation, with the ICRC acting as neutral intermediary for the handover, and not through the Nigerian army. Proof of life, medical evacuations, priority for minors and the elderly: that is the realistic scope of such a request.

Third stage: the final option. It is neither promised nor ruled out: it is conditional. Jaque and Chavín de Huántar show what the condition looks like: a precise location, captives held together, an adversary’s setup that has been penetrated. If containment and the single channel one day produce these three elements, the option exists. If they never materialize, the operation never takes place, and that is not weakness: no force that has succeeded has acted otherwise.

What the state says, and what it leaves unsaid

On its operational posture, nothing. Announcing a suspension of drone operations publicly rewards the blackmail; announcing that it will press on regardless condemns the captives. On substance, a single, cold statement: no recognized territory, no status, no amnesty, however many captives there are. And no promise that everyone will come back unharmed. The 2021 precedent lasted two months and ended in an escape; a government that had promised a quick resolution would have been proved wrong every week, and a state that is proved wrong ends up conceding anything.

If the state nonetheless chooses to talk about more than the releases, the conditions are set beforehand, not during: release of the captives as a precondition, no recognition of territories, no general halt to operations as the price of entry, a limited duration, and third-party verification. The Colombian precedent does not validate each of these points, but it establishes the essential one: the dialogue opened in 2012 required neither territorial sanctuaries nor a prior suspension of security operations. And if a lull comes, it should not be judged by the body count: El Salvador’s 2012 truce reduced homicides while extortion thrived. The indicators that matter are kidnappings, tolls, reopened roads, accessible markets, and population returns.

The question is political

The choice is not between negotiating and never negotiating. States sometimes talk to their adversaries. The real question is who writes the rules, and when. A truce that leaves extortion intact, a zone frozen for the benefit of whoever conquered it, a mediation obtained under threat: these are not ways out of the crisis. They are conversions of violence into power.

Kenscoff imposes a simple line, and the calendar makes it urgent. The state can talk if it deems it necessary, but it cannot teach armed groups that killing, abducting, and conquering more earns a better seat at the table. The danger is not that a gang asks to enter politics. The danger would be for the state, less than four months before an election, to teach it that massacre is the shortest way in.

A note on method

The toll of the 23–24 August attack, the threat against the captives, the PNH statement of 25 August, and the background on drone operations come from news agency dispatches and from BINUH. Data on earlier offensives against Kenscoff come from the joint report by BINUH and the Office of the High Commissioner for Human Rights of April 2025. The “Deklarasyon Drapo Blan” of 20 August 2026 and the weapons videos are analyzed as documents circulating in the Haitian public sphere, not as authenticated evidence. The precedents drawn on are publicly documented: the execution of the Urrao hostages in Colombia (2003), the Chavín de Huántar (Lima, 1997) and Jaque (Colombia, 2008) operations, the Chibok abductions (2014) and their resolution through Swiss facilitation and ICRC intermediation (2016 and 2017), the abduction of Christian Aid Ministries missionaries by 400 Mawozo (2021), public data on the GAULA and on trends in kidnapping in Colombia, the designation of Viv Ansanm by the US Treasury (May 2025), the Colombian negotiation process opened in 2012, and the International Crisis Group’s analysis of the Salvadoran truce.