Have a look in the corridor. Gather valuables. Get dressed. Find children. Find pets. Move to the balcony.
That is not a list of mistakes. It is the list of things people actually did after a fire alarm went off, with cameras running, across a set of Canadian evacuation drills conducted from 1993 to 1995. Afterward the occupants wrote down what they had done, & those six things came up more than anything else. The researchers who ran the drills, Guylene Proulx of the National Research Council of Canada & Rita Fahy of the National Fire Protection Association, published the list in 1997.
Every item on it is reasonable. Every item is what a careful, responsible adult would do. And every item happens before the person has decided to leave.
Read the list again with your own household standing in it. The corridor is your hallway. The valuables are the passports & the laptop. The children are your children. The pets are the dog who hides under the bed when he is frightened. Nothing there is panic, & that calm is precisely the problem.
What People Actually Did in the North Tower
The clearest look anyone has at this comes from the most closely studied emergency in modern memory. After the north tower of the World Trade Center was struck in 2001, federal investigators spent years finding the people who got out & asking them, in detail, what they did between the moment they knew something was wrong & the moment they started down the stairs. NIST published what they found in 2005.
They did not run. Mostly, they talked. Seven in ten of the 440 north tower survivors reached in the telephone survey spoke with somebody else before they began to evacuate. Nearly half of that same group of 440 stopped to collect personal belongings first. Others went looking for the people they worked with.
None of that is strange behavior, which is the whole point. It is what any of us would do at a desk on an ordinary morning when something odd happens. What it costs is time. For the occupants from the ground floor up to floor 76, the typical wait before starting to evacuate was three minutes. Some were moving inside of a minute. Others had still not started down more than an hour after they first knew something was wrong.
Then the investigators did something more useful than counting. They modeled the delay to find out what was actually driving it, & the answer contradicts what most people assume. It was not that anyone was short of information & went hunting for more. Looking for information barely moved the clock at all. What predicted the delay, more than anything else, was simply doing things. Talking. Gathering. Telephoning. Searching. The minutes did not go into thinking. They went into doing.
Two smaller findings from the same analysis run backwards from intuition, & both point the same way. More evidence of danger in the surroundings, the things people could see & hear for themselves, made them slower to start rather than faster. And the further a person was from an exit, the sooner they got moving. All of that is the finding of the federal investigation into the towers, not an interpretation of it.
Why the Calm Is the Dangerous Part
The assumption underneath most household emergency thinking is that people come apart. The research says close to the opposite, & it has said so for a long time. Dennis Mileti & John Sorensen, in the standing federal assessment of how the public responds to emergency warnings, written for Oak Ridge National Laboratory in 1990, put it plainly: people are a hearty lot, & they are not easily convinced that the unthinkable is happening to them. Their central finding is that the first response to a warning is usually not action. It is checking. Getting more information, talking it over with somebody, waiting to hear it again. That assessment dates to 1990 & it has not been overtaken; NIST cited it as foundational again in 2018.
Proulx, the same researcher who filmed those drills, later gave the behavior a name. She called it milling: looking at what other people are doing, discussing the situation, working out together what the right move is. She was careful to say this is not a defect. It is how a group of human beings establishes that something is true. Writing in 2007, she named the reason milling begins at all, which is that an alarm proves nothing by itself. People need something more, a smell, a sight of smoke, before they will accept that the alarm means what it says. Only a minority of the occupants in those Canadian drills said that when the alarm sounded, they thought it might be a real fire.
In one of those drills the alarm went off during a snowfall, & people dressed for the weather before they went outside. Nobody there was being foolish. They were being sensible about a situation they had not yet decided was an emergency. That is the entire mechanism, in one small domestic act.
So this is not a problem of information, & it is certainly not a problem of nerve. It is a problem of belief, wearing the clothes of ordinary competence.
Every one of those acts is what a reasonable person does while life is still normal. Doing them is how a person says, without knowing it: not yet.
What Practice Actually Buys You
John Leach, a psychologist who has spent his career on why people freeze, came at this from the other end. He worked through the official inquiries into a run of shipwrecks & aircraft emergencies, & he interviewed survivors himself. What he published in 2004 is the most useful thing in this literature for anyone with a family, because it puts a clock on the one thing nobody thinks to time: deciding.
If a person already has the right response prepared & stored, it can fire in about a tenth of a second. If they have a handful of prepared responses & only have to pick one, that takes a second or two. If they have nothing prepared, & the brain has to build the response from scratch, that construction takes at least eight to ten seconds under good conditions, & Leach adds the necessary caution that conditions like danger may slow it down further.
Eight to ten seconds reads like nothing on a page. It is not nothing in a dark hallway, once for every decision, with two other people standing there waiting to see what you do before they do anything at all.
The remedy sits inside the same finding, & it is not the one people expect. Practice does not make anybody braver. It moves the task out of the category the brain has to compose & into the category it only has to select.
Running an Advance on Your Own House
In executive protection there is a name for solving this ahead of time. It is called the advance. Somebody walks the route before the principal does, counts the exits, decides in advance what would change the plan & who does what if it changes. Of twenty years in federal service, roughly three of mine were spent leading protective operations at the cabinet level & roughly seventeen in federal investigation & support work, & the advance is the one practice from that world that carries cleanly into a private one. Not because a family needs a detail. Because the whole purpose of an advance is to spend the decisions early, while there is time to make them, instead of in a hallway at two in the morning.
One caution before the list. Scaled down to a household, this is our professional judgment, not a research finding. Nobody has run a controlled test of a rehearsed family against an unrehearsed one, & there is no sense pretending otherwise. What the research establishes is the mechanism underneath it. One of the four elements rests on that research directly, & I will say which.
- A named trigger. Decide now, out loud, in plain words, what counts as the event. The smoke alarm sounding twice. The ground shaking. The word "fire" said by anybody in the house. This is the element with research behind it, & Proulx is the reason: an alarm does not prove itself. If a household has not agreed in advance what counts, it will end up settling the question during the emergency, in the hallway, out loud.
- One action per person. Not a plan. One action, per person, carried out without waiting to be told. This is what keeps a household off the list at the top of this article, because a person doing their own assigned job is not gathering valuables.
- A place to meet. One named spot outside that everybody walks to. It exists so that nobody has to go back through the house looking for anybody, which is the act most likely to carry a person deeper into a building instead of out of it.
- One person outside to call. Somebody who does not live there, whom everyone checks in with. The head count resolves in one place, & nobody has a reason to stay inside making calls to work out who is already out.
Then practice it, which is the part with the mechanism under it. This does not mean a drill in the corporate sense. It means a two minute conversation at the dinner table, a few times a year, until every person can say their own action without having to reach for it. If the children can recite it, the household has it. If only one adult can, the household does not.
None of this makes an emergency safe, & it would be dishonest to sell it that way. The pause does not disappear. Milling is normal, & Proulx would tell you it is useful, & no household plan removes the moment when a person has to accept that this is really happening. What an advance does is shorten that moment, & spend in advance the decisions it would otherwise eat. At Scopos Strategies the advance is standing practice, run before a principal moves rather than assembled after something goes wrong, & the version that fits a family is smaller than the professional one but not different in kind. The thinking gets done early, so that when the moment arrives there is very little left to decide.
If the smoke alarm went off in your home tonight at two in the morning, what is the first thing each person would do, & has anyone in the house ever said it out loud?
Sources
- Proulx, Guylene, and Rita F. Fahy: "The Time Delay to Start Evacuation: Review of Five Case Studies," Fire Safety Science: Proceedings of the Fifth International Symposium, International Association for Fire Safety Science, 1997, pages 783 to 794. https://publications.iafss.org/publications/fss/5/783. Canadian evacuation drills conducted from 1993 to 1995, instrumented with more than thirty video cameras and post-drill questionnaires. The source of the pre-movement activity list that opens this article, of the drill held during a snowfall in which occupants dressed before leaving, and of the finding that an average of 20 to 25 percent of the occupants surveyed thought the alarm might signal a real fire emergency. The paper states that range without printing a questionnaire sample size, which is why this article describes it as a minority rather than as a precise figure. Measured average times to start moving ran from 36 seconds in an office drill to 8 minutes 35 seconds where the alarm was hard to hear.
- Averill, Jason D., Dennis S. Mileti, Richard D. Peacock, Erica D. Kuligowski, Norman Groner, Guylene Proulx, Paul A. Reneke, and Harold E. Nelson: "Federal Building and Fire Safety Investigation of the World Trade Center Disaster: Occupant Behavior, Egress, and Emergency Communications," NIST NCSTAR 1-7, National Institute of Standards and Technology, September 2005. https://nvlpubs.nist.gov/nistpubs/Legacy/NCSTAR/ncstar1-7.pdf. The federal investigation into the 2001 attacks, built on more than a thousand new survivor interviews. Source of the median time to initiate evacuation, 3 minutes for occupants from the ground floor up to floor 76 and 5 minutes for those nearer the impact region on floors 77 to 91, with individual reported delays running from 1 minute to 80 minutes. Source of the activities reported by the 440 WTC 1 survivors in the telephone survey, of whom 70 percent talked to others, 46 percent gathered personal items and 23 percent searched for others, a set of figures totalling more than 100 percent because one respondent could report several actions. Source also of the causal model finding that taking pre-evacuation actions was the greatest predictor of delay, that seeking additional information had minimal impact, that environmental cues and passively received information both increased delay, and that the further an occupant was from an exit the quicker they began to evacuate.
- Mileti, Dennis S., and John H. Sorensen: "Communication of Emergency Public Warnings: A Social Science Perspective and State-of-the-Art Assessment," ORNL-6609, Oak Ridge National Laboratory, prepared for the U.S. Department of Energy, August 1990. https://www.osti.gov/servlets/purl/6137387. The standing federal assessment of the empirical literature on public warning response. Establishes that warning confirmation, not panic, is the documented first response, and states in its own words that "people are a hearty lot and are not easily convinced that the unthinkable (a disaster) can happen." It is a synthesis of the empirical literature rather than a single study, and its authors say so.
- Kuligowski, Erica D.: "A Review of Public Response to Short Message Alerts under Imminent Threat," NIST Technical Note 1982, National Institute of Standards and Technology, 2018. https://nvlpubs.nist.gov/nistpubs/TechnicalNotes/NIST.TN.1982.pdf. A current federal review that cites the 1990 Oak Ridge assessment as foundational. The currency check confirming the warning-confirmation finding is still the operative understanding.
- Proulx, Guylene: "Response to Fire Alarms," Fire Protection Engineering, Society of Fire Protection Engineers, First Quarter 2007. https://www.sfpe.org/publications/fpemagazine/fpearchives/2007q1/2007q11. Defines milling as looking at what others do and discussing the best course of action, and calls it an essential step in the decision-making process rather than a defect. Establishes that a fire alarm by itself is rarely sufficient to trigger evacuation without additional cues, such as the smell or sight of smoke, to certify that the alarm is credible.
- Leach, John: "Why People 'Freeze' in an Emergency: Temporal and Cognitive Constraints on Survival Responses," Aviation, Space, and Environmental Medicine, 75(6), June 2004, pages 539 to 542. https://pubmed.ncbi.nlm.nih.gov/15198281/. A qualitative synthesis of official inquiry reports and witness testimony from five maritime and six aircraft disasters, together with the author's interviews of 19 survivors. It is cited here for mechanism, not for prevalence. The timing structure this article rests on: a prepared and embedded response in as little as 100 milliseconds, a choice among prepared responses in 1 to 2 seconds, and construction of a new response in at least 8 to 10 seconds under optimal circumstances and longer under threat. Leach draws those cognitive intervals from the working memory literature rather than from an experiment of his own, and says so in the paper.
This article was produced with AI assistance. The perspective and security expertise are the author's own.