Three uniformed organisations and one coordinating office responded to the World Trade Center on the morning of 11 September 2001, each of them established its own command, and no single commander held the incident as a whole. The published investigations state that plainly, and they also record what it cost in information that existed inside one agency and never reached the people in another agency who needed it. This piece, the third of four on the communications record of the response in New York, reads the command arrangements the way a professional reads an after action file, with the 9/11 Commission Report as the principal source.

Who was operating, and what each of them set up

The 9/11 Commission Report, published in 2004 by the National Commission on Terrorist Attacks Upon the United States, devotes its ninth chapter to the emergency response in New York, and it names the principal first responders as the Fire Department of New York, the New York Police Department, the Port Authority Police Department and the Mayor’s Office of Emergency Management. Those four organisations had different legal bases, different radio systems, different dispatch arrangements and different internal command language, and all four were operating in and around the complex within minutes of the first aircraft impact. Before anything can be concluded about coordination, it is worth being precise about what each of them actually was, because the national terminology most readers use today did not exist in 2001 and applying it backwards produces a tidier picture than the record supports.

The FDNY as the Commission describes it had about 11,000 members and was headed by a fire commissioner who, unlike the police commissioner, held no operational authority, with operations headed by the chief of department. Its response ran on an alarm escalation system numbered from a first alarm to a fifth, each level carrying a pre-established complement of units, and the Commission notes that it was common practice before 2001 for units to arrive with extra personnel and for off-duty firefighters to respond to major incidents. The Commission states that the FDNY response began within five seconds of the crash and that by 9:00 a.m. many senior FDNY leaders, including seven of the eleven most highly ranked chiefs in the department along with the commissioner and many of his deputies and assistants, had begun responding from headquarters in Brooklyn. The Commission’s narrative describes FDNY command being established in the lobby of the North Tower, a separate FDNY operation in the lobby of the South Tower, and a further FDNY command location on the street outside the complex, which means that the fire department alone was working from more than one post before any question of interagency command arises.

The NYPD brought a different structure to the same site. The Commission records that the police department supervised the city’s 911 system, staffed by roughly 1,200 operators, radio dispatchers and supervisors who were civilian employees of the department, and that a 911 call concerning a fire was transferred to FDNY dispatch. Police operations at the complex included Emergency Service Unit rescue teams entering the towers and an aviation unit airborne over the site, and police command was established at its own location outside the buildings rather than in either lobby. The Port Authority Police Department was the third uniformed organisation, responsible for the complex itself, and the Commission is blunt about its state of preparation for a multi-command response, stating that as of 11 September the Port Authority lacked any standard operating procedures governing how officers from its several commands would respond to, be staged at, and be used at a major incident at the World Trade Center, and that although the PAPD had an agency-wide radio channel, not all of its commands could access it.

Two collapse times anchor everything that follows, and I am not going to quote them from memory, because they are among the most commonly misreported figures associated with the day. The times generally published are 9:59 a.m. for the South Tower and 10:28 a.m. for the North Tower, with a 2002 review in the journal Critical Care giving 10:29 for the fall of the North Tower, and the National Institute of Standards and Technology reporting the collapse initiation times to the second in its federal building and fire safety investigation, the NCSTAR 1 series of 2005. 7 World Trade Center collapsed later in the afternoon, with published accounts giving a time of about 5:20 p.m. If you need a defensible time for a report or a class, take it from NIST or from the Commission’s own chronology rather than from any summary, including this one.

Direction and Control of Emergencies: the protocol that already existed

New York had a written allocation of command authority before the attacks, and its existence is the reason this case study is useful rather than merely sad. The 9/11 Commission Report states that in July 2001 Mayor Rudolph Giuliani updated a directive titled “Direction and Control of Emergencies in the City of New York”, whose stated purpose was to ensure the optimum use of agency resources while eliminating potential conflict among responding agencies which may have areas of overlapping expertise and responsibility. The mechanism the directive used was designation: for different types of emergency it named an appropriate agency as the incident commander, with that agency then responsible for the response. The Commission describes the July 2001 document as an update, which means an earlier version existed, and I have not been able to verify the date and title of that earlier version from a primary source, so I am leaving it out rather than printing a year I cannot stand behind.

The Office of Emergency Management sat alongside that directive rather than above it. The Commission records that Mayor Giuliani created the office in 1996 with three basic functions, the first being a Watch Command that monitored the city’s key communications channels including FDNY dispatch and NYPD radio frequencies, the second being the improvement of the city’s response to major incidents by planning and conducting exercises and drills involving multiple city agencies and particularly the NYPD and FDNY, and the third being management of the city’s overall response, under which liaisons from relevant agencies and the mayor and his senior staff would report to the Emergency Operations Center once it was activated. The Commission adds that an OEM field responder would be sent to the scene to ensure that the response was coordinated, which is a real and specific role, and one that shows up in the record of that morning in at least one place where it worked.

What the Commission concluded about how all of this performed is the finding that matters, and it needs to be stated as the Commission’s conclusion rather than as anybody’s opinion. The Commission found that the response operations lacked the integrated communications and unified command that the directive contemplated, and that these problems existed both within individual agencies and between them. That conclusion concerns practice rather than paperwork, and the distinction is the whole point of this article. A directive that allocates lead agency responsibility by incident type answers the question of who is nominally in charge of what, and it does not by itself produce a group of chiefs who have stood at the same table before, who know each other’s names and radio designators, and who have a shared habit of building one plan, because that habit is built in exercises with the other agencies physically present or it does not exist on the day.

There is a further problem with allocation by incident type that any emergency manager reading this will recognise from their own plans. An incident of this scale is simultaneously a fire, a mass casualty event, a structural collapse, a criminal act, a transportation emergency and a continuity of government problem, so a table that assigns lead agency by category has to be interpreted under time pressure by people who did not write it. The Commission’s treatment of command and control in New York is the principal source for the finding that the interpretation did not converge on a single commander, and where later characterisations of that finding are contested, which they are, I deal with that in the seventh section rather than smoothing it over here.

What the aviation units reported, and where it went

The police department had aircraft over the complex, and the observations those crews made are the clearest illustration in the whole record of information that existed and did not travel to the people whose decisions depended on it. The 9/11 Commission Report states that NYPD Aviation determined that rooftop rescues from the Twin Towers were not possible that day, and it records the consequence for the call takers, namely that because the 911 operators were not informed of that determination, they could not knowledgeably answer callers who asked whether to go up or down. That is the Commission’s own finding and it concerns the flow of information inside the city’s emergency system, not the conduct of any individual operator, several thousand of whose calls are described elsewhere in the same chapter.

The later aviation observations are the ones usually cited, and they need to be handled with their sourcing attached. The Gotham Gazette’s summary of the Commission’s findings on emergency response states that NYPD Aviation radioed immediately that the South Tower had collapsed, that at 10:08 a.m. an aviation helicopter pilot advised that he did not believe the North Tower would last much longer, and that there was no ready way to relay this information to the fire chiefs in the North Tower. The Commission’s own narrative contains additional aviation transmissions about the condition of the buildings at earlier times, and I am not going to print those times from memory, because the exact wording and timing of each transmission is the kind of detail that circulates in altered form. A reader who needs the transmissions themselves should work from the Commission’s chapter and its endnotes, and from the NIST emergency response volume in the NCSTAR 1 series, which reconstructs response communications from recordings and interviews.

The mechanism here is worth stating precisely, because the popular version of this story is that one agency withheld information from another, and that is not what the investigations describe. The aviation observations moved perfectly well inside the police department, reaching police units and the police chain of command, and both NYPD rescue teams in the North Tower knew that the South Tower had collapsed and left the building. What did not exist was a path from the police radio system to the fire department’s command posts that anybody could use quickly, and that absence is a structural property of two separate commands on two separate systems rather than a decision made by a person. Part two of this series dealt with what the investigations establish about the radios inside the towers, and I am not going to restate any of it here beyond noting that the in-building problem and the interagency problem were different problems with different causes.

The record also contains a check-in that worked and a check-in that did not, and the pair is more instructive than either one alone. The Commission states that the first NYPD Emergency Service Unit team entered the West Street level lobby of the North Tower and prepared to begin climbing at about 9:15 a.m., that they attempted to check in with the FDNY chiefs present, and that they were, in the Commission’s word, rebuffed. In the South Tower, by the Commission’s account, an Office of Emergency Management field responder who was present ensured that the NYPD rescue teams checked in with the lobby chief, and in that case both agreed that the rescue team would ascend in support of FDNY personnel. One interagency liaison standing in a lobby produced a coordinated arrangement in one building and the same arrangement did not occur in the other, which tells you how thin the mechanism for interagency contact was and how much depended on whether a particular liaison function happened to be present at a particular door.

The coordinating office at 7 World Trade Center

The office whose entire purpose was to coordinate city agencies was housed inside the complex that was attacked. The Commission states that the OEM’s headquarters was located at 7 WTC, that some had questioned locating it both so close to a previous terrorist target and on the twenty-third floor of a building that would be difficult to access if elevators became inoperable, and that there was no backup site. Those three facts belong together, because the objection recorded in the report is not hindsight added in 2004 but a contemporaneous concern about siting, and the absence of an alternate is what turned a siting question into an operational one.

On the sequence and timing, the sources I can check are not fully consistent, and I am going to give you both rather than pick. The 9/11 Commission Report states that OEM was evacuated at 9:30 a.m., and that figure is the one cited from the report in later summaries and filings. The 9/11 Memorial’s published timeline attributes the evacuation decision to a report by a United States Secret Service agent about the possibility of additional hijacked aircraft. A 2002 review of the response published in the journal Critical Care instead describes the OEM building as having been heavily damaged at 10:29 by the fall of the North Tower and evacuated then, with command and control re-established elsewhere. The Commission is the stronger source on the timing and I would use its 9:30 a.m. figure, while noting that at least one peer-reviewed account published a year earlier describes the sequence differently, and that a reader citing this in a plan should quote the Commission directly.

What I have not been able to verify is where the city re-established a coordination facility later that day. Published accounts name a pier on the Hudson and they do not agree on which one, so I am leaving the location out rather than printing a number I cannot source, and the point does not depend on it. The point is that no alternate emergency operations centre was in operation at the moment the primary one was abandoned, which meant that the function of interagency coordination was interrupted during the period when the agencies most needed it, and that the field responder role became the whole of the coordination mechanism at the site rather than one part of it. In a statement submitted to the Commission after the attacks, the city’s Office of Emergency Management described the arrangements that had been built since, including a lead agency system with a unified operations post for incidents where multiple agencies act on a variety of emergency fronts, and a full backup Emergency Operations Center with facilities for the mayor and his key staff.

Every agency reading this has the same problem in a smaller package. Your emergency operations centre is very often in the same building as your dispatch centre, your administrative offices and your primary radio equipment, which is efficient and is exactly what a budget committee will approve, and it means that one event can take all four. The question is not answered by having a designated alternate location on a page in the plan, because a designated location becomes an operational alternate only when somebody has confirmed that the phones and radios there work, that the people who would staff it know how to get in after hours, and that whatever data and mapping the room depends on is reachable from that network. A location that has been named in an annex and never occupied during an exercise has not yet been shown to work, and the way to find out is to occupy it while nothing is wrong.

What a unified command would and would not have changed

This is the part of the argument where people usually claim too much, so I want to be exact about the limits before I state the case. A unified command would not have improved portable radio coverage inside the towers, because the physics of a steel high rise does not care how many chiefs are standing at the same table, and the coverage and channel loading problems established by the investigations were properties of the radio systems and the buildings. A unified command would not have changed the structural behaviour of either tower, the timing of either collapse, or the conditions above the impact zones, all of which NIST examined in the NCSTAR 1 series and none of which is sensitive to command arrangements. Anyone who tells you that a single incident commander would have prevented the outcome of that morning is making a claim the investigations do not support.

What a unified command changes is narrower and entirely real. It creates a physical and procedural place where information held by one agency reaches a decision maker in another agency in the ordinary course of business, without anybody having to decide to make a special call. When the fire chiefs, the police chiefs and the facility’s own police command are working from one post with one operational period plan, an observation from an aircraft reaches the table because the person receiving it on the police net is standing next to the person who needs it, and the same is true of a report from a stairwell or a status from a building engineer. That is the entire mechanism, and it is worth being clear that it is a mechanism for moving information rather than a source of authority over physics.

The places where that mechanism would have applied on the morning are identifiable from the Commission’s own findings without speculating about what anyone would have done with better information. Call takers were answering questions about whether to go up or down while the aviation determination that rooftop rescue was unavailable sat inside the police department, and the fire chiefs in the North Tower had no ready path to receive aviation assessments of the building’s condition, which the Commission records along with the fact that police units in that building knew the South Tower had collapsed. Each of those is a point where a shared command post changes what was physically in front of a decision maker, and I am not prepared to say what any decision would then have been, because that is precisely the speculation this series avoids.

There is also a cost side that honest analysis has to carry. A unified command post takes time to assemble, it concentrates senior people in one location, and it introduces its own information bottleneck if the post is run without a discipline for what gets briefed and when. Agencies that adopt unified command as a slogan rather than as a practice produce a large room full of principals, no written operational period objectives, and worse information flow than they had with separate posts, which is a failure I have watched happen at exercises. The version that works is small, it has named seats, it produces one plan for each operational period, and it has a communications unit that can tell you which channel each agency’s field units are actually on.

Why separate command posts feel correct from inside each one

The most useful thing about this case for a working agency is that the failure mode is the ordinary one rather than an exotic one. Separate command posts are not a symptom of dysfunction and they do not feel like an error while they are happening, because each command is functioning correctly by the standards of the agency that set it up, in a location that agency chose, on a radio system it owns, staffed by people who know each other’s capabilities and who are executing a structure they have practised. That is true simultaneously of a fire command in a lobby, a police command on a street corner and a facility police command inside a building its officers know better than anyone else on scene, so nothing registers as an error from inside any of them and there is no observer position from which the whole arrangement is visible to the people running it.

What is missing is visible only from outside all three, and no one on the incident is standing outside all three. Each commander knows what their own units are doing, believes the other agencies are handling their own responsibilities, and has no routine way to notice that a fact known at one post has not arrived at another, since the absence of a message is indistinguishable from the absence of anything to report. A gap in interagency information flow produces no alarm, no error tone and no entry on a status board, which is why it survives a debrief where everyone reports that their own command functioned well and everyone is telling the truth.

The same property explains why this is rarely caught in routine operations. Most multi-agency incidents in most jurisdictions are small enough that the missing information would not have changed anything, so the arrangement is validated hundreds of times by incidents that did not test it. A structure fire with police doing traffic control works fine with two separate commands, and so does a highway crash, a water rescue and a small hazardous materials spill, because the information each agency holds is information the other agency does not need. The arrangement only fails when one agency holds a fact that determines another agency’s tactical decision, and those incidents are rare enough that a department can go a decade without encountering one.

There is a further complication in jurisdictions with strong agency cultures, which is most of them, and it does not require anybody to behave badly. An agency that has carried the operational load for a hazard for thirty years reasonably regards its own command structure as the competent one for that hazard, and asking it to place its commander at a shared post with another agency’s commander reads as a challenge to that competence unless the arrangement has been worked out in advance and rehearsed. That is a governance problem rather than a personality problem, and it is solved with written agreements that name roles and with exercises that give people the experience of the arrangement working, rather than with an instruction issued on the day.

Where the published accounts disagree, and how to read them

Three bodies of published work carry most of the weight on this subject and they were produced for different purposes, which is the first thing to understand about reading them together. The 9/11 Commission Report of 2004 was the product of a federal commission examining the attacks and the response to them, and its ninth chapter is the principal source for command and control in New York. The NIST federal building and fire safety investigation of the World Trade Center disaster, published as the NCSTAR 1 series in 2005, includes volumes on occupant behaviour and egress and on emergency response communications, and it is the source built on systematic reconstruction of recordings, interviews and physical evidence. The McKinsey and Company review of the FDNY response, prepared for the department and released in 2002, was an internal management review commissioned by the department itself, which gives it a different standing again, because a review a department commissions about itself is evidence of what that department was prepared to accept as findings.

These sources do not agree on everything, and where they disagree I am not going to resolve it. The Commission’s characterisations of command and control in New York were publicly contested at the time of its 2004 hearings in New York by former city officials and by people who had held senior positions in the departments, and the substance of that disagreement is on the record in the hearing transcripts and submitted statements published by the Commission. If you are going to cite the Commission’s conclusion that the response lacked unified command, the professional way to do it is to cite it as the Commission’s conclusion, note that it was disputed by city officials at the time, and let the reader go to the transcripts. The dispute concerns what the arrangements on the day amounted to and how much of the outcome is attributable to them, and that is a genuinely contested question rather than a settled one.

The claims that circulate most widely are the ones least anchored to the published record, and two of them deserve labelling. The first is the claim that one agency deliberately withheld warnings from another, which the investigations do not establish and which misdescribes the mechanism the Commission actually documents, namely the absence of a usable path between two separate command structures. The second is the claim that a single change in command arrangements would have produced a different outcome, which no investigation concluded and which cannot be tested. Both circulate because they are simpler than the record, and both are worth correcting in the same breath as they are repeated, particularly in a training room where they will be absorbed as fact.

My own use of these sources in this article follows one rule that I recommend for any historical case study used in training. Where a claim rests on a single source, the source gets named in the sentence that carries the claim, so that a reader can see the weight the claim is bearing. The Commission is the source for the directive, for the description of OEM’s functions and location, for the absence of a backup site, for the check-in at the North Tower lobby and the check-in at the South Tower lobby, and for the conclusion about integrated communications and unified command. Where I have used a secondary summary, such as the Gotham Gazette’s account of the Commission’s emergency response findings for the aviation transmission at 10:08 a.m., I have said so in the same sentence, and a plan or a lesson plan that cites these events should go to the primary documents rather than to me.

The exercise that finds this, and the one that hides it

An exercise detects a command ownership problem only if it is designed to detect one, and most exercises are designed in a way that guarantees it stays hidden. The common pattern is a tabletop with one agency hosting, a scenario written by that agency’s planner, and injects that test that agency’s decisions, with the other agencies present as guests who are asked what they would do. Everyone leaves satisfied, the after action report records good interagency cooperation, and the question of who would have been in command was never put, because the host agency was obviously in command of its own exercise in its own building.

The design that surfaces the problem is not expensive and it is uncomfortable, which is the main reason it is rare. Write a scenario in which the decision that matters depends on a fact that only one agency’s units can observe, put the agencies in separate rooms with only the communications paths they actually have, and give the facilitator authority to refuse any information transfer that would not have occurred in reality. A police observation from a drone or an aircraft, a utility company’s knowledge of what is in a pipe, a school district’s knowledge of where students were moved, or a hospital’s knowledge of its own diversion status will all serve. The evaluation objective is written as a question about information arrival, meaning whether the fact reached the decision maker in the other agency, by what path, and how long it took, and that is the only measure the exercise needs to produce to be worth running.

The second element is the one people skip. Before the exercise, get the written answer to who commands a multi-agency incident in your jurisdiction, and read it out loud in the room. In much of the country the answer is a combination of state statute, a county emergency operations plan, a mutual aid compact and an unwritten local understanding, and the four are not always consistent with each other. Verify the current arrangement with your state emergency management agency and your county attorney rather than relying on what the previous chief told you, because the answer changes when compacts are revised and nobody circulates the revision to operations personnel.

The interoperability guide on this site makes the argument that the hard part of interoperability is governance rather than hardware, and the New York record of 11 September 2001 is the case study it was written from, so I will let the record carry the point rather than repeating the argument. Part four of this series takes up what came out of all of this in policy, including the incident management structure that the federal government issued afterwards and the requirements attached to it, and it is worth saying here that the policy answer was built precisely on the finding that command ownership had to be established before the incident rather than negotiated during it.

What to do at your agency

• Ask your emergency manager or the person who maintains the emergency operations plan for the specific page that states who holds command of a multi-agency incident in your jurisdiction, read the page, and circulate it to your shift commanders with the date it was last revised written at the top.
• Have your training officer schedule one tabletop in the next ninety days in which a police, fire, EMS and public works representative are physically present, with the single written objective of determining who would be in command and how the other agencies would receive information from that command.
• Put one item on the agenda of the mutual aid or chiefs association meeting that already meets in your area, asking each agency to state where its command post would be established for an incident at one named large facility in your response area, and record the answers on one page.
• Confirm with your communications manager which channel or talkgroup a neighbouring agency’s incident commander would use to reach yours during an incident, then have two supervisors test that path this month from vehicles rather than from the shop, and write down the result with the date.
• Locate your designated alternate emergency operations centre, go to it with the person who holds the key, confirm the radios and telephones in that room work today, and note in the plan what is missing rather than assuming the room is ready.
• Write one paragraph into your existing emergency operations plan stating what happens to coordination if your primary emergency operations centre is inside the incident area, naming the alternate location, the person authorised to make that call, and how agencies are notified of the move.
• Ask your law enforcement counterpart what their aviation, drone or specialised units would observe at a large incident and how that observation would reach your incident commander, and if the answer is a phone call to a number nobody has, fix the number and put it on your incident command worksheet.

Takeaways

• The 9/11 Commission Report of 2004 identifies the principal first responders in New York as the FDNY, the NYPD, the Port Authority Police Department and the Mayor’s Office of Emergency Management, and each of those organisations established command on its own terms rather than under a single commander for the incident.
• New York had a written protocol, the directive titled Direction and Control of Emergencies in the City of New York, which the Commission states Mayor Giuliani updated in July 2001 and which designated an appropriate agency as incident commander for different types of emergency, so the failure the Commission describes was one of practice rather than of missing paperwork.
• The Commission concluded that the response operations lacked the integrated communications and unified command contemplated by that directive, and that the problems existed both within individual agencies and between them, a conclusion that was publicly disputed by former city officials at the Commission’s 2004 hearings in New York.
• Information from NYPD aviation moved correctly inside the police department and had no ready path to the fire chiefs, with the Commission recording that 911 operators were not informed of the aviation determination that rooftop rescues were not possible, and the Gotham Gazette’s summary of the Commission’s findings recording an aviation pilot advising at 10:08 a.m. that he did not believe the North Tower would last much longer.
• The Commission records that OEM headquarters was at 7 World Trade Center on the twenty-third floor, that the siting had been questioned before the attacks, that there was no backup site, and that the office was evacuated at 9:30 a.m., while a 2002 review in the journal Critical Care describes the evacuation as following damage from the North Tower collapse at 10:29.
• A unified command would not have improved radio coverage inside the towers and would not have changed the collapses, and what it could have changed is whether a fact held by one agency reached a decision maker in another agency in the ordinary course of business.
• Separate command posts are the normal failure mode rather than the exceptional one, because each command is functioning correctly by its own agency’s standards and the absence of an interagency message produces no alarm that anyone on the incident can see.
• An exercise finds this problem only when it is designed to, meaning the agencies are in separate rooms using the paths they actually have, the decisive fact is observable by only one of them, and the evaluation objective is whether that fact reached the other agency’s decision maker and how long it took.