The Drills Everybody Complained About

Rick Rescorla was Morgan Stanley's director of security. The firm was the largest tenant in the World Trade Center, occupying twenty-two floors of the south tower with nearly 2,700 employees.

Rescorla had lived through the 1993 bombing and had watched that evacuation go badly. He became convinced the building would be attacked again, and when he could not persuade the firm to move he did the only other thing available to him. He built an evacuation plan and made everybody rehearse it.

The plan was specific rather than general. Two abreast in the stairwell. A lane kept clear so first responders could climb while employees descended. Upper floors first. And drills twice a year, for everyone, from the top floors down.

People complained. Morgan Stanley's own account of it is unsentimental on this point: the rehearsals provoked the odd complaint, and surely, some said, they did not have to be so frequent. These were traders and bankers being pulled off desks to walk down twenty-two floors of stairwell, twice a year, for a building emergency that had not happened in eight years.

On the morning of 11 September 2001, the Port Authority announced over the public address system that occupants of the south tower should remain calm and stay at their desks. Rescorla overrode it. He picked up a bullhorn and started the evacuation he had drilled: before the second aircraft struck his building.

2,687 people got out.

Thirteen Morgan Stanley employees and consultants did not. Rescorla was among them. He was last seen on the tenth floor, climbing, sweeping for stragglers.

It was not the plan. Plans are common and cheap. Almost every large tenant in that building had an evacuation plan on file somewhere.

It was that the plan had been run, repeatedly, by the people who would have to execute it: until descending twenty-two floors two abreast was something their bodies already knew how to do. When the announcement told them to stay put, they had a stronger instruction available: the one they had practiced.

Let us be straight about the likelihood. You are not going to evacuate a skyscraper. The specific event is essentially irrelevant to a private household, and any article that tried to make you feel otherwise would be selling fear rather than preparation.

But the structure of the failure scales all the way down, and you can test it yourself in about four seconds.

Does your household have a plan for a fire at two in the morning? Most people say yes. Now: when did anyone last walk it? Not discuss it: walk it, at night, with everyone who lives there.

If the honest answer is never, you are in exactly the position Morgan Stanley's neighbors were in on the morning of 11 September. Not planless. Unpracticed. The two look identical on paper and behave nothing alike under pressure.

Not preparedness in the abstract. Something narrower and more useful: the answer to a question you cannot currently answer.

Does everyone in this household know two ways out of the room they sleep in? Does the person who would need help have someone assigned, and a backup? Does everyone know where to go once they are outside, and would they wait there?

You either know those answers or you do not, and twenty minutes converts a guess into a fact. That is worth having whether or not anything ever happens: and unlike most preparedness work, it costs nothing but the willingness to be slightly ridiculous in your own house on a Tuesday evening.

First, pick the meeting point and say it out loud. A tree, a mailbox, the end of the driveway. Somewhere everyone can name. This takes one minute and it is the single most common missing piece, because families assume it is obvious and it is not.

Second, walk two ways out of every bedroom. Not discuss. Walk. Open the second window and find out whether it actually opens, because a painted-shut window is a discovery you want to make now rather than at two in the morning.

Third, assign the dependents, and assign a backup. Who gets the child who sleeps heavily, the grandparent with limited mobility, the dog? And if that person is travelling, who does it instead? A plan that only works when everybody is home is not a plan.

Fourth, run it at night. Daylight and darkness are not the same exercise. If you only ever do it once, do it at night.

The order matters. There is no point rehearsing a route to a meeting point nobody has agreed on, and no point assigning dependents before you know whether the second exit opens.

Here is the transferable lesson, and it is bigger than fire.

Rescorla's people did not survive because they had a plan. They survived because the plan had been run, by them, until it was automatic: and because when an official instruction contradicted it, they had something practiced to fall back on.

The next disruption your household faces will not be the one you planned for. It will be a different hazard, at a different hour, with a different person unavailable. What transfers is not the specific route. It is the fact that this household has done the exercise of coordinating under pressure at least once, and therefore knows something about how it behaves when it does.

So the question worth carrying out of this article is not whether you have a plan. Nearly everyone has a plan. It is: which of our plans has anyone ever actually run?

For most households the honest answer is none of them. Not the fire plan, not the plan for who collects the children if both parents are unreachable, not the plan for what happens if the person who holds everything in their head is suddenly unavailable.

That last one is the one we find most often, and it is the one nobody thinks to rehearse.

Family protocols and crisis coordination is one of the eight domains Sentinel Resilience assesses. It is not the domain most people expect to be assessed on (it has no equipment to buy and nothing to install) and it is consistently among the weakest.

The assessment does not ask whether a household has a plan. It asks whether the plan has been tested, whether the people named in it know they are named, and whether it still works when the obvious person is not there. Those are different questions, and the answers are usually less comfortable.