Peacekeeping After Certainty: Why the UN needs new decision-making structures built for uncertainty

A video appears to show a ceasefire violation. One party calls it proof of an attack; the other says the footage is fabricated. The mission currently cannot access the site, and the only available surveillance data belongs to one of the parties. Civilians may be at risk. Headquarters wants an assessment. It is clear that the mission has a mandate. But given the circumstances, the question arises what the mission can responsibly decide before all facts are settled.

This is no longer a hypothetical problem at the margins of peacekeeping. It goes to the institutional foundations of how peace operations observe events, establish responsibility and determine what they can do next in a moment of crisis.

The Secretary-General’s Review on the future of all forms of United Nations peace operations, released yesterday, is unusually clear about the environment in which future missions will operate. It describes AI-enabled systems, cyberattacks and increasingly accessible advanced capabilities; shrinking political space and strained host-state consent; financial pressure and widening gaps between mandates and means. It also acknowledges that missions will need to make decisions under uncertainty.

This diagnosis matters. And so does the report’s insistence that peace operations become more focused, adaptable and technologically capable. But recognizing a more contested and complex operational environment is not the same as changing how missions assess events and decide what to do.

The report asks what missions should do and how they should be equipped, but a mandate and the right capabilities are only part of the problem. What happens when a mission has to decide how to respond before it can establish what happened, who was responsible, or whether the evidence can be trusted?

From information gathering and analysis to decision-making

Decision-makers in peace operations have always had to work with somewhat incomplete information; nothing new about the fact that situational awareness can only ever be so comprehensive. However, what is changing now is the scale, speed and strategic usefulness of manufacturing uncertainty. Synthetic evidence can circulate before a mission can verify an incident. Attribution can be contested even when physical harm is obvious. Access to sensors, satellite imagery or communications infrastructure can depend on actors whose interests are not neutral.

Realistically, UN missions are often among the least technologically capable actors in the environments where they operate. Closing that gap would help, but financial constraints, host-state restrictions and competing Member State interests make technological parity quite unlikely. And even a better-equipped mission would still face manipulated evidence, contested attribution and decisions that potentially cannot wait lengthy verification assessments. The UN therefore needs decision-making structures that work under uncertainty regardless of how much technology it can realistically acquire.

A mission may be authorized to monitor a ceasefire and possess a robust operational posture, yet still struggle to decide whether a reported violation occurred, which party was responsible, what it should communicate publicly and which response options therefore remain legitimate and proportionate. Those are not separate technical questions. Together, they determine whether the mandate can de facto be exercised effectively, or at all.

Design for uncertainty

The Secretary-General's review calls for impartial fact-finding and a shared factual basis for action. It also highlights organizational adaptation, integrated analysis and leadership preparedness. The missing step is to translate those commitments into procedures that work even when the facts cannot be established quickly, independently or conclusively.

Make information requirements part of mission design.

A ceasefire-monitoring mandate is only as credible as the mission’s ability to assess alleged violations. Mission planning and mandate reviews should therefore examine what type of information different operations requires, whether they can access relevant locations independently, which technologies missions control/have access to and where they depend on parties to the conflict, or even third parties. If those conditions are absent, the limitation should be visible to mission leadership and the Security Council from the outset. A mandate cannot simply assume the verification capacity needed to implement it.

Set clear evidence requirements for different decisions.

Not every response requires the same level of certainty. A credible indication of danger to civilians may justify increased monitoring, protective deployment or urgent engagement even when attribution of responsibility remains unclear. Public attribution, support for punitive measures or actions carrying a risk of escalation should require stronger evidence. Missions need agreed procedures that distinguish between precautionary action, public assessment and escalatory response. The relevant question has to shift from aiming at establishing total certainty, which will be increasingly impossible, and move towards the question which decisions remain justified under uncertainty that might persist.

Separate facts, interpretations and attribution.

Mission reporting should distinguish what was directly observed, what was reported by others, what was independently verified and what remains inferred or disputed. Headquarters and the Security Council need a clear account of uncertainty, competing explanations and the consequences of waiting, not a polished narrative that disguises unresolved questions.

Technological dependencies belong in that assessment as well. If a mission’s only surveillance data comes from a conflict party, or a third party in general, that is not merely an equipment issue; it shapes the credibility of the evidence, the perception of impartiality and the range of defensible options.

Rehearse decisions under uncertainty

Paragraph 62(d) commits the United Nations to “developing scenario-based exercises to enhance leadership-level skills and decision-making”; the report also calls for joint exercises across partners. Used properly, these exercises can reveal a mission's crisis management strategy, what evidence is needed to deploy personnel, when Headquarters must be consulted, and whether civilians can be protected before responsibility for an attack is established, just to name a few examples. They can also show where different considerations pull the mission in conflicting directions. A military response may be operationally possible but politically unacceptable. Public attribution may risk escalation, while waiting for confirmation may leave civilians exposed. Those findings should lead to clearer agreements about how much evidence is needed before a mission can warn civilians, deploy personnel, issue a public assessment or escalate a decision to Headquarters. They should also inform reporting standards, decision authority and mission planning. That is the point of the exercises I design for senior UN and security leaders: to put decision-makers in situations their current procedures were not built for, and work out what needs to change.

The strategic question ahead

None of this replaces the need for political support, host-state cooperation, adequate resources or credible mandates. Nor can institutional redesign eliminate uncertainty from conflict. But missions that are expected to operate in technologically contested environments need more than new tools and broader awareness. They need decision-making structures that remain usable when evidence is incomplete, attribution is disputed and the available response options are already constrained.

The Secretary-General’s review identifies many of the right challenges and points toward several of the right instruments. The next step, however, is to ask not only what future peace operations should be mandated and equipped to do, but how they will decide what to do under conditions of uncertainty that may not be resolved.