Joint exercises assume a crisis arrives on its own, is recognisable when it starts and ends on schedule. Compounded crises break all three, and readiness reporting can't see the gap.
That is the argument of my article for the RUSI Newsbrief, "What Joint Exercises Assume About How Crises Behave", published on 8 October 2026.
Why do the assumptions survive?
Each one solves a design problem. A single scenario can be scripted and marked against a model answer. A crisis already named lets exercise control steer the event. An end set by the clock fits the booked time. Together they describe a crisis that is bounded, legible and finite, which is the kind compounded conditions do not produce.
The same logic sends one common picture to every cell. A compounded crisis hands each cell a different fragment, and reconciling them is the work the uniform push removes. As a planner, I held to that design, and against the measures set for it, it worked. That is why the error was hard to see. Under a uniform push nobody could easily fail, so the exercise confirmed the plan.
What does the public record show?
The UK's Exercise CYGNUS in 2016 opened in week seven of a pandemic already identified, so the ambiguous early phase was never exercised. Its report said the exercise achieved its aim. The same report found the UK's preparedness and response not sufficient for a severe pandemic.
What would change it?
Four changes to the scoresheet, with no new resources. Open mid-event, with conflicting reports and no name for the crisis. Score how cells build a shared picture separately from how they execute the plan. End on a condition instead of a clock. Plant weak signals of a second, unrelated crisis in the cells least likely to own it, and track where each one travels.
The harder discipline is designing exercises that can fail.
Read the full piece at RUSI (RUSI members). For what this looks like with a leadership team, see the Crisis Leadership Exercise.
