Monitoring and real-time decisions

Introduction

On 5 November 2021, at the Astroworld festival in Houston, ten people died of compression asphyxia in the crowd. The Houston Police Department’s 1,266-page report, released in 2023, assigns no blame, but its timeline contains one number every event professional should know: 37 minutes passed between the declaration of a mass-casualty event and the end of the show. The operations plan did not address crowd surges, and who had the authority to stop the show was unclear.

A crowd in trouble rarely announces itself in a way everyone recognises at once. By the time it is obvious, minutes matter, and a decision that has to be debated is a decision that comes too late. This lesson is about seeing early and deciding fast, which means doing most of the deciding before the event begins.

Seeing density from a control room

The quantity to watch is density, the subject of module 1’s first lesson, and the instrument is usually a CCTV screen and a trained eye. Eyes are poor instruments without calibration. Still warns that “Crowds rarely pack in regular formation” and that “Density can appear higher with lower CCTV angles”. Figure 1 shows why: from a low camera, heads further back hide the floor between people, and a crowd at 2 p/m² can look solid.

Three practices make judgement more reliable:

• Reference images. Images of known densities taken from the venue’s own cameras let an operator compare like with like: this is what 2 p/m² looks like from camera 14, this is 4. Their purpose is calibration, not decoration.
• Known areas. If the floor area in a camera’s view is measured in advance, an operator can count heads in a small patch and divide by its area rather than guessing. The next lesson builds this into a counting method.
• Counting flows. Counts at gates and on routes tell you how many are arriving and leaving, which density alone cannot. After 2006 the Jamarat Bridge introduced automated camera counting for exactly this reason.

Triggers and pre-agreed actions

A trigger is a pre-defined, measurable condition, such as a density in a named area, that obliges a named person to take a pre-agreed action. The point of a trigger is that the argument happens in advance, in daylight, with the plan on the table, not at 21:30 with a crowd in difficulty.

Density thresholds give the rungs. Typical UK event planning works at 2 p/m²; Still treats 5 p/m² as the upper limit for standing and viewing, labels 5 “Maximum” and 6 “High Risk” on his risk scale. For crowds that need to keep moving, the numbers are lower: Johansson and colleagues recommend keeping density below 3–4 p/m², and Still’s own pages put the critical range for moving crowds at 2–3 or 2–4 p/m², an inconsistency covered in module 2.

Three things make a ladder like this work. Each action is specific and already resourced: “hold entry” means a named gate team with a radio call-sign, not a hope. The decision-maker for each rung is named, with a deputy. And the trend matters as much as the level: an area at 3.5 p/m² and rising fast may need the amber action sooner than one sitting steadily at 4.

Decisions under pressure: three cases

Who can stop the show?

At Astroworld the authority to stop the show was unclear, and 37 minutes passed between the mass-casualty declaration and the end of the performance. A show-stop should never be improvised. It needs a written procedure, a named person with the authority to use it, an agreed way of telling the performer, and pre-scripted announcements, all rehearsed. The decision-maker must be able to act on a trigger without seeking further approval.

An action without flow control

At Hillsborough on 15 April 1989, exit Gate C was opened at about 14:52, and about 2,000 people entered. Nothing controlled where those people went once inside, and the 2016 inquests found that the central pens were already full before the gate opened. Any action that releases a crowd into an area must come with control of where it goes. Module 8 reads the case in full.

The anti-pattern

At the Kanjuruhan Stadium in Malang, Indonesia, on 1 October 2022, police fired tear gas into the stands. Police said 11 rounds; Indonesia’s human rights commission found about 45 inside the stadium. All the gates but one were locked, and 135 people died. The fact-finding team named the tear gas as the main cause. FIFA’s regulations already said “No firearms or ‘crowd control gas’ shall be carried or used”. Every element was the opposite of good practice: an action that created a sudden rush, exits that could not take it, and no information for the crowd.

Information is a control

Of all the levers in a control room, the public address system is the fastest. Module 3’s collective resilience lesson gives the evidence: crowds act on information, and honest, specific messages that tell people what is happening and what to do make the crowd a partner in the response. A message script for each trigger level belongs in the plan alongside the action.