Team of Teams: New Rules of Engagement for a Complex World

by Stanley McChrystal

This is not to say that simply throwing more computers or Navy SEALs at a problem is the answer; the key lies not in the number of elements but in the nature of their integration—the wiring of trust and purpose. Parallel computing, joint cognition, and the oneness of a team all work toward the same goal: building a network that allows you to solve larger, more complex problems. The creation and maintenance of a team requires both the visible hand of management and the invisible hand of emergence, the former weaving the elements together and the latter guiding their work. Programs like BUD/S are designed to foster emergent intelligence that can thrive in the absence of a plan.

The crew’s attachment to procedure instead of purpose offers a clear example of the dangers of prizing efficiency over adaptability.

NASA believed that the dwindling ability of flight crews to adapt to unforeseen events stemmed from the captains’ attempts to control and plan for everything in a vehicle that had become too sophisticated for that to be possible. Champions of the iconic Mission Control room where hundreds of experts crowded into one space to facilitate real-time communication and adaptation (which we will investigate further in later chapters), they concluded that building trust and communication between crew members was more important than further honing specific technical skills.

“Ladies and gentlemen, the plane is no longer the problem.”*

The solution, which came to be known as Crew Resource Management (CRM),* was developed in consultation with social psychologists, sociologists, and other experts, and focused on group dynamics, leadership, interpersonal communications, and decision making.

In medicine as in aviation, technology had outpaced the capacity of any individual practitioner to be on top of it all; once this was recognized, there was a movement toward “cross-functional trauma teams,” with more even distribution of authority and leadership. Research showed that these changes cut average times for complete resuscitation in half, from 122 to 56 minutes.

It is no coincidence that CRM and EMS systems emerged at roughly the same time. Preston Cline, the associate director of leadership ventures at the University of Pennsylvania’s Wharton School, has spent years researching “Mission Critical Teams” (MCTs)—small teams whose failure will likely lead to loss of life, and whose time frames for action often involve critical periods of ten minutes or less.

The proliferation of such groups reflects the increasing complexity of the world—or rather, the tactical understanding that responding to such a world requires greater adaptability, and adaptability is more characteristic of small interactive teams than large top-down hierarchies. We can now do the things we used to do—get from New York to Portland, raid a building, provide trauma care—more quickly and effectively than we could fifty years ago, but doing them has become more complex and confusing, to the point that they are beyond the effective control of a single person.

“A combination of increased mobility, increased information, and increased impact means that we have reached a tipping point,” notes Cline. “Previously, we had a historical pattern of disruption followed by stabilization—‘punctuated equilibrium’—but now that pattern itself has been disrupted. Today, we find ourselves in a new equilibrium defined by constant disruption. This creates the kinds of problems that only MCTs can solve.”

We had honed the traits of trust and purpose at the team level, but our organization at large was the complete opposite—it was a classic command. Because our Task Force was used to clean lines and right angles thinking,

Fundamental structural differences separate commands from teams. The former is rooted in reductionist prediction, and very good at executing planned procedures efficiently. The latter is less efficient, but much more adaptable. The connectivity of trust and purpose imbues teams with an ability to solve problems that could never be foreseen by a single manager—their solutions often emerge as the bottom-up result of interactions, rather than from top-down orders. In recent decades, teams have proliferated across domains previously dominated by commands in response to rising tactical complexity. The adaptability of the Task Force’s teams represented a valuable start, but we would have to build that same adaptability at a much greater scale.

There is a catchy acronym in the consulting world, “MECE,” which stands for “mutually exclusive and collectively exhaustive.”

Great teams are less like “awesome machines” than awesome organisms.

As soon as we looked at our organization through the lens of the team structure—searching for weaknesses in horizontal connectivity rather than new possibilities for top-down planning—similar choke points became visible between all our individual teams. We referred to them as “blinks.”

In a now famous 1999 Institute of Medicine study, “To Err Is Human,” it was estimated that between 44,000 and 98,000 people died every year as a result of medical errors. Even if the lower estimate of the study is used, deaths due to medical errors would have been the Centers for Disease Control and Prevention’s seventh-leading cause of death in 1998, meaning that more people died as a result of medical errors than from motor vehicle accidents (43,458 deaths), breast cancer (42,297 deaths), or AIDS (16,516 deaths).* A new study published in September 2013 asserts that the number of deaths due to medical error is dramatically higher: 210,000 to 400,000. Either estimate would have put medical errors as the third-leading cause of death in the CDC’s 2011 ranking. If the estimated 100,000 deaths due to hospital-acquired infections are included, this loss is equal to twenty Boeing 747 airliners going down every week.

COMMAND OF TEAMS

Small teams are effective in large part because they are small—people know each other intimately and have clocked hundreds of hours with each other. In large organizations most people will inevitably be strangers to one another. In fact, the very traits that make teams great can often work to prevent their coherence into a broader whole.

How many “cooks” is too many? It depends. In a small kitchen or office, four might be the ideal number. For a company with operations the size of Walmart, the break point is much higher. For some activities, like having an engaging conversation, diminishing marginal returns sets in after a few people. For other tasks, like producing a mechanical item via assembly line, you can add just as much value with the hundredth employee as with the first. For teams, this range is considerably narrower. Athletic teams, for instance, usually consist of fifteen to thirty people. Army Ranger platoons are composed of forty-two soldiers. SEAL squads contain between sixteen and twenty people. Beyond such numbers, teams begin to lose the “oneness” that makes them adaptable. As the proverbial kitchen fills up, communication and trust break down, egos come into conflict, and the chemistry that fueled innovation and agility becomes destructive. In many cases, this loss of adaptability dooms the enterprise.

Imagine the closest roommate relationship you’ve ever had and multiply that by one hundred. The bonds within squads are fundamentally different from those between squads or other units. In the words of one of our SEALs, “The squad is the point at which everyone else sucks. That other squadron sucks, the other SEAL teams suck, and our Army counterparts definitely suck.” Of course, every other squad thought the same thing.

British anthropologist Robin Dunbar theorized that the number of people an individual can actually trust usually falls between 100 and 230 (a more specific variant was popularized by Malcolm Gladwell as the “Rule of 150” in his book Outliers). This limitation leads to a kind of tribal competitiveness: victory as defined by the squad—the primary unit of allegiance—may not align with victory as defined by the Task Force. The goal becomes to accomplish missions better than the team that bunks on the other side of the base, rather than to win the war. In other words, the magic of teams is a double-edged sword once organizations get big: some of the same traits that make an adaptable team great can make it incompatible with the structure it serves.

“As a team gets bigger, the number of links that need to be managed among members goes up at an accelerating, almost exponential rate.” In his handbook Leading Teams, Hackman reminds us of “Brook’s Law”: the adage that adding staff to speed up a behind-schedule project “has no better chance of working . . . than would a scheme to produce a baby quickly by assigning nine women to be pregnant for one month each . . . adding manpower to a late software project makes it later.”

As we sat in our makeshift command center in Balad, reading reports of AQI bombings, we realized that our goal was not the creation of one massive team. We needed to create a team of teams. It may sound like a kitschy semantic distinction, but it actually marked a critical structural difference that turned the aspiration of scaling the magic of the team into a realizable goal.

But on a team of teams, every individual does not have to have a relationship with every other individual; instead, the relationships between the constituent teams need to resemble those between individuals on a given team: we needed the SEALs to trust Army Special Forces, and for them to trust the CIA, and for them all to be bound by a sense of common purpose: winning the war, rather than outperforming the other unit. And that could be effectively accomplished through representation.

By this point I knew that defeating Zarqawi and his organization could not be accomplished by a traditional command—even a command composed of teams as capable as our own. We would have to match AQI’s adaptability while preserving our traditional strengths, and this would necessitate an unprecedented transformation—it would require a true team of teams. Accomplishing this would involve a complete reversal of the conventional approach to information sharing, delineation of roles, decision-making authority, and leadership.

Although our Task Force’s constituent teams exemplified adaptability, a commandlike superstructure constrained the organization at large. This “command of teams” approach was more flexible than a conventional command, but was still not adaptable enough to deal with the complexities of the twenty-first century and battle AQI. Although teams have proliferated across many sectors, they have almost always done so in the confines of broader commands. More and more organizations will need to overcome this hurdle and become more adaptable. Unfortunately, many of the traits that made our teams so good also made it incredibly difficult to scale those traits across our organization. We were also up against some fundamental constraints. Building a single team the size of our Task Force would be impossible. The solution we devised was a “team of teams”—an organization within which the relationships between constituent teams resembled those between individuals on a single team: teams that had traditionally resided in separate silos would now have to become fused to one another via trust and purpose.

Though we rarely use that phrase in real life, it is an accurate depiction of military and broader organizational sentiments about the value of information: given the overwhelming volume of, and myriad sensitivities around, information, the default is not to share.

Such absurdities reflect the truth that most organizations are more concerned with how best to control information than how best to share it.

The problem is that the logic of “need to know” depends on the assumption that somebody—some manager or algorithm or bureaucracy—actually knows who does and does not need to know which material.

The problem, at one level, was obvious: we were failing to create useful bonds between one team and the next. The work done by our operators and analysts was inextricably linked, and yet we had placed the two groups in separate organizational silos—we had given them blinders—in the name of efficiency. Our players could only see the ball once it entered their immediate territory, by which time it would likely be too late to react. With no knowledge of the constantly shifting perspective of their teammates, they would have no idea what to do with the ball once they got it.

Frederick Ordway summarized the nature of their achievement as follows: “The Apollo project . . . is generally considered as one of the greatest technological endeavors in the history of mankind. But in order to achieve this, a managerial effort, no less prodigious than the technological one, was required.”

The interface failures, however, exposed an inherent problem: independent small groups were very effective at exploratory work, but trouble erupted when the projects of the disparate teams had to be integrated into a vehicle going into orbit.

Given these failures, in 1962 NASA leadership had doubts about the feasibility of Kennedy’s goal. “Most of us in the Space Task Group thought [Kennedy] was daft,” recalled a NASA executive. “I mean, we didn’t think we could do it. We didn’t refuse to accept the challenge, but God, we didn’t know how to do [Earth] orbit determination, much less project orbits to the Moon.”

In 1963, NASA brought in George Mueller to build the managerial foundation of the Apollo program, and he brought a sea of organizational change. His vision for NASA was that of a single interconnected mind—an emergent intelligence like the “joint cognition” that defines extraordinary teams. As NASA director Wernher von Braun framed it, Mueller brought the perspective of an electrical engineer who aspired to create a managerial “nervous system,” whereas von Braun, a mechanical engineer, saw organizations as reductionist contraptions.

What Mueller instituted was known as “systems engineering” or “systems management,” an approach built on the foundation of “systems thinking.” This approach, contrary to reductionism, believes that one cannot understand a part of a system without having at least a rudimentary understanding of the whole. It was the organizational manifestation of this insight that imbued NASA with the adaptive, emergent intelligence it needed to put a man on the moon.

In the two years after Mueller was brought on, Apollo transformed from a group of loosely organized research teams into a tightly run development organization. Even the engineers most ardently opposed to systems management found that many technical problems could be solved only by sharing information. As von Braun put it, “The real mechanism that makes [NASA] ‘tick,’ is . . . a continuous cross-feed between the right and left side of the house.” In half a decade, a space program that had once been a national embarrassment became the best in the world.

In fact, even for NASA, as historian Howard McCurdy has noted, “maintaining . . . organizational culture as practiced by the first generation of employees turned out to be most difficult to do.” After Apollo, its well-integrated system of units slid into a competitive set of independent entities; its open communications calcified with bureaucracy. One employee characterized NASA in 1988 as “the Post Office and the IRS gone to space.”

As McCurdy notes, FBC was the antithesis of systems engineering. Systems engineering was “formal, elaborate and expensive.” It was inefficient. But it worked.

This is the difference between “education” and “training.” Medical school is education, first aid is training. Education requires fundamental understanding, which can be used to grasp and respond to a nearly infinite variety of threats; training involves singular actions, which are useful only against anticipated challenges. Education is resilient, training is robust.

We did not want all the teams to become generalists—SEALs are better at what they do than intel analysts would be and vice versa. Diverse specialized abilities are essential. We wanted to fuse generalized awareness with specialized expertise. Our entire force needed to share a fundamental, holistic understanding of the operating environment and of our own organization, and we also needed to preserve each team’s distinct skill sets. We dubbed this goal—this state of emergent, adaptive organizational intelligence—shared consciousness, and it became the cornerstone of our transformation.

RECAP Like NASA before it, our Task Force found itself confronted with a complex problem that demanded a systems approach to its solution; because of the interdependence of the operating environment, both organizations would need members to understand the entire, interconnected system, not just individual MECE boxes on the org chart. Harnessing the capability of the entire geographically dispersed organization meant information sharing had to achieve levels of transparency entirely new to both organizations. In traditional organizations, this constitutes culture change that does not come easily. It demanded a disciplined effort to create shared consciousness.

In the private sector also, physical space has for a century been used to facilitate and enforce efficiency and specialization. Along with factory assembly lines, the architectural frames of white-collar work have evolved to maximize efficiency. In the nineteenth century, “countinghouses” where partners and clerks worked side by side at identical rolltop desks began to disappear, replaced by subdivided offices. As the volume of clerical and administrative work grew, white-collar professions began importing the reductionist ideal of specialization from the factory floor. Management historian Alfred Chandler observed that the role of the merchant, which once embraced “exporter, wholesaler, importer, retailer, shipowner, banker and insurer,” split—like Adam Smith’s pin production—into multiple specialized businesses in the late 1800s.

New technologies enabled the construction of larger, taller buildings to house the increasingly complicated strata of the workplace. The “office building” took shape under the hand of architects such as Louis Sullivan, who envisioned structures composed of independent, standardized cells, which he likened to the hexagonal building blocks of beehives: discrete, MECE units, not to be merged.* Dictaphones and pneumatic tubes enabled discrete, directed communications at a distance without the messy inefficiencies of the countinghouse. Executives moved to separate rooms, then to plush suites, and finally to different floors to separate them from the “pools” of stenographers toiling away at desks arranged in grids, silent but for the clacking of typewriter keys. The “corner office” where a manager could separate himself from the rest of the workforce became a status symbol. (During the 2008 financial crisis it emerged that many executives commute to their offices in private elevators, further minimizing any potential interaction with employees.) It was in the early 1900s that the term “ladder” became common parlance for the corporate hierarchy.

How we organize physical space says a lot about how we think people behave; but how people behave is often a by-product of how we set up physical space. At Balad we needed a space that facilitated not the orderly, machinelike flow of paperwork, but the erratic, networked flow of ideas—an architecture designed not for separation, but for the merging of worlds. We weren’t the only ones to be trying this—there was a growing movement in the private sector to organize offices for better cooperation, too.

Firms that value innovation and creativity have spent a lot of time searching for ways to inject interactivity into work environments. In 1941 Bell Labs famously broke with tradition, hiring Skidmore, Owings & Merrill to design a campus whose spaces promoted interaction: to move from an office to a lab, for example, employees had to walk through the cafeteria where they would bump into people. The hope was that such casual interactions with peers, managers, and even custodial staff might prompt unexpected insights. In the 1970s even staid IBM experimented with early “nonterritorial” offices where engineers could come in, grab their materials, and sit anywhere in the open plan space arranged to facilitate “serendipitous encounters” with team members. As with NASA, these changes were not initially welcomed, but grew in popularity. “I was skeptical before, but I’d hate to go back to a closed office now,” said one engineer. Another was more succinct: “Don’t fence me in again.” In Silicon Valley, Google, Facebook, and other titans, as well as countless start-ups, use open plans that put different teams and different rungs of management in the same space.

When former mayor Michael Bloomberg moved into New York’s City Hall, he turned down the building’s fancy mayoral suites, and instead had the Board of Estimate Hearing Room—one of the most lavish ceremonial spa...

Anyone in the room—regardless of their position in the org charts’ silos and tiers—could glance up at the screens and know instantly about major factors affecting our mission at that moment.

The cubicle itself is a good example of management space gone wrong. Originally created by the visionary inventor Robert Propst to free workers from isolation, the cubicle has become a symbol of the impersonal culture it aimed to reform. The “Action Office II” was supposed to be customizable and reconfigurable for privacy, but also for cooperation, promoting interaction. It was designed to be arranged in organic clusters, reflecting a new conception of the office as an interconnected whole. Put into production by Herman Miller in 1967, Propst’s invention was immediately perceived as transformative. The New York Post ran an article about it titled “Revolution Hits the Office,” which argued that the days of “the completely enclosed ‘boxes’ in which the bosses isolate themselves behind monster mahogany status symbols and the inhuman row upon rigid row of steel desks with their clumsy drawers at which you sit all day” would soon draw to a close. “The success of the concept seems assured,” the article concluded.

We needed transparency that provided every team with an unobstructed, constantly up-to-date view of the rest of the organization.

RECAP Shared consciousness in an organization is either hindered or helped by physical spaces and established processes. Often, efforts to facilitate Taylor-inspired efficiencies have produced barriers to information sharing and the kind of systemic understanding we needed to pervade our Task Force. Creating transparency and information sharing at the scale we needed required not only a redesign of our physical plant, but also a rethinking of almost every procedure in our organizational culture. The daily O&I briefing lay at the core of our transformation: this pumped information about the entire scope of our operations out to all members of the Task Force and partner agencies, and also offered everyone the chance to contribute.

We needed true, not theoretical, collaboration, transparency, and trust. Putting everyone in the same room was a start.* But if we wanted instinctive, second-nature, teamlike trust, we would have to go much deeper. The stronger the ties between our teams—as with the prisoners—the higher the likelihood we would achieve the level of cooperation we needed.