Thirty horses jumped the first fence at Aintree on 3 April 1993. They ran the full four miles and thirty fences of the Grand National. Seven of them finished. Esha Ness crossed the line first, in what would have been the second-fastest time in Grand National history.

None of it counted. The race had already been declared void. The field had set off on a false start, and the recall had failed to reach them. Not because nobody tried. Because the flag didn't unfurl.

TWO FALSE STARTS. ONE FLAG. NOTHING.

Around fifteen animal rights protesters had made it onto the course near the first fence, forcing a delay before the race could begin. When the field finally lined up, the starting tape tangled in the horses on the first attempt. False start. The field was recalled without incident.

On the second attempt, the tape caught around Richard Dunwoody's neck. Starter Keith Brown called another false start and waved his recall flag. The flag, wet from the rain, simply hung there. It gave no signal. It communicated nothing.

Nine jockeys caught something: a waving official, an instinct, a glimpse. They pulled up. Thirty did not. Many who continued later said they assumed the officials along the course were further protesters. The crowd noise was enormous. The flag, to anyone not positioned directly in front of it, was invisible as a signal.

The thirty who ran did so in good faith. They ran a clean race. They had no way of knowing it would be erased.

THE FIX REVEALED THE FLAW

The inquiry that followed found Keith Brown, the starter, partly responsible; Ken Evans, the recall official further down the course, was also held to account. Brown did not return as starter after the inquiry. The reforms that followed were revealing: two officials with fluorescent yellow flags and radio contact with the starter were added, along with a third official positioned with a vehicle if physical interception was needed.

The system was rebuilt not because the concept of a recall mechanism was wrong. But because the one that existed had never been tested under the conditions it was supposed to handle.

NOT NEGLIGENCE. UNTESTED CONFIDENCE.

It is a specific kind of failure, and a familiar one. It happens in organisations that are careful, that have thought about risk, that have built contingency into their plans. The failure mode is not negligence. It is untested confidence.

The 1993 Grand National had a recall system: a flag, an official, a procedure. Everyone knew the procedure. The problem was that nobody had verified whether it would work in the wet, in the noise, in the specific chaos of a Grand National false start. On the day it mattered, the backup was a prop.

The reforms are instructive precisely because they reveal the original gap. Radio contact. Fluorescent flags visible at distance. A vehicle. These are not improvements to a working system. They are the difference between a process that looks capable of handling failure and one that actually is.

Most organisations have recall flags of one kind or another. Disaster recovery plans filed and seldom opened. Escalation procedures written into handbooks and never rehearsed. The question is not whether the process exists. It is whether anyone has confirmed it works when conditions are worst: when the tape is wet, the crowd is loud, and the outcome depends on a signal landing cleanly.

Esha Ness ran a brilliant race. Jenny Pitman had prepared a horse that was ready to win. The day was lost not to a competitor, not to bad luck, not to any failure of the thirty who ran.

It was lost to a flag no one had checked.

TRY IT YOURSELF

🚩 Pick one recovery process your organisation relies on. Ask three questions:

🕰️ When was it last tested? Not reviewed or documented — actually run under pressure.

🌧️ Under what conditions? Calm, controlled tests don't count. Wet flags look fine until race day.

🔍 What would you add if it failed today? Whatever comes to mind is your real gap.

If you can't answer all three, the flag hasn't been checked.

FURTHER READING

📚 Normal Accidents: Living with High-Risk Technologies by Charles Perrow
In complex, tightly coupled systems, failures are built in. Adding redundancy often increases risk rather than reducing it, because it assumes the backup will actually perform when activated.

🗣️ Atul Gawande: The Checklist Manifesto on Moonshots Podcast
Gawande's argument that checklists work not by telling experts what to do, but by enforcing a moment of explicit verification: confirming that what you believe has been done has actually been done.

🧑‍💻 Grand National Declared Void by RTÉ Archives
Archival footage and contemporaneous reporting from the day itself: the moment the flag failed, and Jenny Pitman's reaction at the finish line.

SOME FINAL WISE WORDS

"Whatever could go wrong that day did."

Keith Brown, starter of the 1993 Grand National

Until next time

Business lessons from the world of sport

Keep Reading