4.

THE LAW OF THE NETWORK

How does one build a team with seven thousand swim buddies?

—GENERAL STANLEY MCCHRYSTAL

It wasn’t a fair fight. The U.S. Army’s Task Force in Iraq in late 2003 was arguably the world’s most sophisticated fighting machine with abundant resources and advanced technology. It was facing a poorly resourced, ill-educated bunch of extremists. So why was the U.S. Army losing?

The Task Force was led by one of the army’s star commanders, General Stanley McChrystal, who took charge in October 2003. He had at his disposal unmatched firepower, armored vehicles, cutting-edge surveillance, and new technologies such as precision weapons, GPS, and night vision. Descending from blacked-out helicopters that could locate a specific rooftop in a sea of buildings with pinpoint accuracy, operators communicated via headsets with pilots controlling drones that provided constant video surveillance.

The Task Force had achieved “the holy grail of military operations: near-perfect ‘situational awareness,’” writes McChrystal in his book Team of Teams. “This was the first war in which we could see all of our operations unfolding in real time. Video feeds from unmanned aerial vehicles (UAVs or drones) gave us live footage of missions, while microphones carried by our operators provided audio. We enjoyed access to data on population, economic activity, oil exports, generation of electricity, and attitudes (through polling); we were connected to our partner organizations in real time.”¹

Before coming to Iraq, the Task Force’s teams had several decades of successfully executed, precise, surgical operations. They had received the most rigorous training in the history of special operations. For several decades, the U.S. Army had been preparing for exactly the kind of challenge they were now facing in Iraq: asymmetric warfare. By any standard, these were high-performing teams, strong on trust and bonded together by common purpose.

On paper, the rabble of frustrated Sunnis led by Jordanian extremist Abu Musab al Zarqawi had no chance. They were ill-educated, ill-trained, underresourced, poorly equipped, and making do with primitive homemade weapons assembled in safe-house basements from propane tanks and expired Soviet mortars. They had no coherent plan. They were making it up on the fly. They depended on face-to-face meetings and hand-carried letters. They were dogmatic and extreme in their conduct and views. They weren’t military geniuses or tactical masterminds.

It wasn’t just that the Task Force should have been winning this engagement. By every traditional military calculus, it should have been crushing its enemy. But it wasn’t. The Task Force was losing badly and consistently and the situation was deteriorating.

At first, McChrystal couldn’t understand how a poorly organized bunch of misfits could be defeating his crack force. It took him time to figure out the problem with his awesome military machine: The Task Force was a machine, while the enemy was operating as a flexible network. And in a turbulent environment with instant and pervasive communications, a machine—even a big, sophisticated, well-funded machine—is no match for a network.

The Task Force had ample access to big data, but that wasn’t much help in terms of prediction. The Task Force was just too slow. McChrystal found himself being asked to make decisions and give approvals on matters on which the teams themselves were better placed to make the call. The decision-making apparatus was simply slowing down the ability of the Task Force to move as expeditiously as it needed to. He asked himself: Why am I being asked to make these decisions? What am I contributing?

“In the time it took to move a plan from creation to approval,” writes McChrystal, “the battlefield for which the plan had been devised had changed. By the time it could be implemented, the plan—however ingenious in its initial design—was often irrelevant. We could not predict where the enemy would strike, and we could not respond fast enough when they did.”²

McChrystal could also see that the problem wasn’t collaboration within the teams themselves, but rather collaboration between the teams: “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.”³

The teams “had very provincial definitions of purpose: completing a mission or finishing intel analysis, rather than defeating [the enemy]. To each unit, the piece of the war that really mattered was the piece inside their box on the org chart; they were fighting their own fights in their own silos. The specialization that allowed for breathtaking efficiency became a liability in the face of the unpredictability of the real world.”⁴

McChrystal could see that his superb teams were embedded in an authority-based bureaucracy in which communications and decisions flowed slowly and vertically. “Stratification and silos were hardwired throughout the Task Force,” he writes. “Although all our units resided on the same compound, most lived with their ‘kind,’ some used different gyms, units controlled access to their planning areas, and each tribe had its own brand of standoffish superiority complex. Resources were shared reluctantly. Our forces lived a proximate but largely parallel existence.”⁵

Communications, writes McChrystal, were “even worse between the Task Force and our partner organizations: the CIA, FBI, NSA, and conventional military units with whom we had to coordinate operations. Initially, representatives from these organizations lived in separate trailers, with limited access to our compound. Built in the name of security, these physical walls prevented routine interaction and produced misinformation and mistrust. The NSA [National Security Agency], for instance, initially refused to provide us with raw signal intercepts, insisting that they had to process their intelligence and send us summaries, often a process of several days. They weren’t being intentionally difficult; their internal doctrine held that only they could effectively interpret their collections. Passing out raw data invited misinterpretation with potentially disastrous consequences.”⁶

McChrystal knew he had to do something different. But what? He knew that “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,” he writes, “the very traits that make teams great can often work to prevent their coherence into a broader whole. How does one build a team with seven thousand swim buddies?”⁷

The Task Force was too big to turn into one big team, yet it could not leave each team to its own devices. It needed rapid coordination of both information flow and collaborative action across the entire enterprise. The challenge was to achieve trust and purpose in a large group of teams and agencies without creating chaos.⁸

McChrystal knew he had to eliminate “the deeply rooted system of secrecy, clearances, and interforce rivalries.” He had to reverse the principle of limiting information on a “need to know” basis to one where “everyone knows everything” so that “every man and woman in our command understood his or her role within the complex system that represented all of our undertakings. Everyone needed to be intimately familiar with every branch of the organization, and personally invested in the outcome.”⁹

He could see that the ability to adapt to complexity and continuous unpredictable change was more important than carefully prepared plans. Rapid horizontal communications were more important than vertical consultations and approvals. Individual team excellence was not enough: Collaboration among teams was vital to achieve overall group performance. Teams had to be able to make decisions as needed, without seeking approvals higher up the command. Big data would never offer respite from the unrelenting need for continuous adaptability.

McChrystal’s approach was a very simple idea: Create a team of teams. This meant turning the Task Force from a bureaucracy into a network (see Figure 4-1). A network is a group or system of interconnected people or things. An organizational network is a set of teams that interact with and collaborate with other teams with the same connectivity, interaction, and passion as they do within their own small team. An organizational network is founded on the Law of the Small Team. But it requires more. Each team needs to look beyond its own goals and concerns and see its work as part of the larger mission of the collectivity.

In effect, McChrystal had to transform the preexisting hierarchy of authority into a hierarchy of competence. Decisions had to be based on who was best placed to make the decision, not their position in the formal hierarchy. McChrystal had to find ways not only “to reverse the information flow—to ensure that when the bottom spoke the top listened, but also “breach the vertical walls separating the divisions of our enterprise. Interdependence meant that silos were no longer an accurate reflection of the environment. Events happening all over were now relevant to everyone.”¹⁰

Figure 4-1. The difference between command of teams and a network.

Source: S. McChrystal, T. Collins, D. Silverman, and C. Fussell, Team of Teams: New Rules of Engagement for a Complex World (New York: Penguin Publishing Group, 2015), p. 24.

Embracing the seeming chaos of a network was a difficult personal decision for McChrystal to take. For centuries, military organizations had been built on this vertical and horizontal stratification. Although progress had been made in empowering individual teams to take initiative on the battlefield, the overall top-down command structure was still in place in the basic military structure.¹¹

How did he transform an authority-based pyramid to a competence-based network?

First, it meant bringing all the key actors together in a common physical space to enable horizontal information flows, promote deeper understanding, and encourage initiative and decision making. “How people behave is often a by-product of how we set up physical space. At Balad [the makeshift command center], 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,” McChrystal writes.¹²

As he further describes this open space, “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. Personnel were placed strategically throughout the space, depending on their function—those with access to real-time information critical to ongoing operations were closer to the center of the room, those with a longer-term focus were on the fringes, so they could focus on other work. Any of them, however, could walk freely across the room for quick face-to-face coordination. With the touch of a button on the microphone, everyone’s attention could be captured simultaneously.”¹³

This wasn’t just about symbolic egalitarianism. The cultivated chaos of the open space encouraged interaction between employees distant from one another on the org chart. Putting himself in the middle of it enabled McChrystal to keep a finger on the pulse of the Task Force. The open space enabled serendipitous encounters, with the recognition that no one could know in advance what connections and conversations would prove valuable.

Second, McChrystal himself embodied open communications in a daily briefing session that lasted an hour or two and in which everyone could understand how the overall operation was fitting together. It was an evolution of the daily standup (described in Figure 2-1, Common practices of Agile small teams). In McChrystal’s daily briefing session, all the actors could see what was going on, grasp the connections between different actions, and make their contribution. The daily briefing that took place in the open space was a physical embodiment of the kind of collaboration that McChrystal envisaged. Electronic connections enabled units in the partner organizations—at the CIA, FBI, NSA, and conventional military units elsewhere in Iraq, the United States, and Europe—to listen in and participate. The daily give-and-take among people at all levels in the hierarchy was a dramatic embodiment of the fluid communication and collaboration needed for success.

The daily briefings were a constant affirmation that there was no established script that was being followed willy-nilly. Leaders were engaging others at the level of choice. There was no telling in advance what would be decided. The leaders did not have an outcome that was being imposed. They were open to possibility and surprise. Everything that was said was of potential consequence. The leaders had to avoid acting as though they were pressing for a predetermined conclusion. They had to be willing to allow for alternative outcomes, whatever the cost to their own egos or the pride of their own units. They had to accept that the best answer might come from anywhere in the network.

Third, McChrystal pushed decision making and ownership down to the lowest possible level for every action. In effect, McChrystal was creating an operation that functioned as a horizontal network based on trust, common purpose, shared consciousness, and empowered execution.

He largely removed himself from the decision-making process. Rather than having to approve all important operational decisions, he created an environment where those closest to the action could make immediate decisions. This meant only occasionally demanding the final say. McChrystal had to honor the judgment of his people, whenever he could. There were times of course when he had to “make the call.” McChrystal occasionally pulled rank—because of external constraints or significant risk of failure. But he didn’t do it often, or lightly. He had to recognize that “less is more.”¹⁴

This in turn meant creating high levels of trust so that teams accepted the responsibility to make decisions without seeking approvals. Instead of holding people accountable, he had to help people be accountable to each other. McChrystal created public forums for network members to discuss and evaluate each other’s actions; these forums also showcased members as examples to encourage higher performance.

Fourth, there was an exchange of staff between teams. McChrystal would “take an individual from one team—say, an Army Special Forces operator—and assign him to a different part of [the] force for six months—a team of SEALs, for example, or a group of analysts.” As McChrystal writes, “Predictably, initial resistance was intense. ‘Our teams train in entirely different ways,’ I was informed. I was told that I needed to understand that the tight bonds inside assault teams came from working with trusted comrades over years—to insert an outsider is an unwise and unfair risk to operators already performing the most difficult of missions. Simply put, it was anathema to the entire history and tradition of special operations selection, training, and war fighting.”¹⁵

But McChrystal persisted. “Although it was a ‘forced’ initiative, once the mandate was in place, elite units were naturally incentivized to send their best operators and leaders. These individuals would be representing their organization, so unit pride would drive them to select the best examples from an already highly selective sample set. Many of these top-of-the-pack personalities were also the types that had a natural ability to connect with others—especially in an environment where leadership and one’s capability as an operator were a critical measuring stick among peers.”¹⁶

Fifth and perhaps the most difficult, McChrystal had to unlearn what it means to be a leader. A great deal of what he thought he knew about how the world worked and his role as a commander had to be discarded. He writes:

I began to view effective leadership in the new environment as more akin to gardening than chess. The move-by-move control that seemed natural to military operations proved less effective than nurturing the organization—its structure, processes, and culture—to enable the subordinate components to function with “smart autonomy.” It wasn’t total autonomy, because the efforts of every part of the team were tightly linked to a common concept for the fight, but it allowed those forces to be enabled with a constant flow of “shared consciousness” from across the force, and it freed them to execute actions in pursuit of the overall strategy as best they saw fit. Within our Task Force, as in a garden, the outcome was less dependent on the initial planting than on consistent maintenance. Watering, weeding, and protecting plants from rabbits and disease are essential for success. The gardener cannot actually “grow” tomatoes, squash, or beans—she can only foster an environment in which the plants do so.

Although I recognized its necessity, the mental transition from heroic leader to humble gardener was not a comfortable one. From that first day at West Point, I’d been trained to develop personal expectations and behaviors that reflected professional competence, decisiveness, and self-confidence. If adequately informed, I expected myself to have the right answers and deliver them to my force with assurance. Failure to do that would reflect weakness and invite doubts about my relevance. I felt intense pressure to fulfill the role of chess master for which I had spent a lifetime preparing. But the choice had been made for me. I had to adapt to the new reality and reshape myself as conditions were forcing us to reshape our force. And so I stopped playing chess, and I became a gardener.¹⁷

Implementing the Law of the Network led to a remarkable acceleration in the flow of information and the speed of decisions. It was not as though the Task Force had upped its game through the discovery of some secret treasure trove of data or a technological breakthrough in surveillance. What made the difference was that the Task Force was no longer a rigid machine. It had become “an adaptable, complex organism, constantly twisting, turning, and learning to overwhelm its protean adversary.” It had mastered the Law of the Network.¹⁸

“Until the 1980s,” writes Carlota Perez in her book Technological Revolutions and Techno-Economic Paradigms, “the prevalent organization was the one that served as the optimal framework for deploying the mass-production revolution: the centralized, hierarchical pyramid with functional compartments. The structure was applied by almost every corporation. . . . With the advent of computers and the Internet, large pyramids now appear rigid and clumsy. In its place the decentralized flexible network structure, with a strategic core and rapid communication, has shown the capacity for accommodating much larger and more complex global organizations as well as smaller ones.”¹⁹ Perez predicted that the network model would eventually encompass a very wide range of organizations, both public and private.

The network model of organization is key to making the whole organization Agile. As McChrystal found in a military setting, merely plugging Agile teams into a pyramidal hierarchy doesn’t lead to the fluid horizontal communications necessary in a fast-changing context. Instead, communications tend to flow vertically, up and down the chain of command. The customer or end-user is often not in the picture at all. Many opportunities for improved performance are missed. In this setting, small gains by individual teams are possible, but the whole organization continues to operate as a clumsy, slow-moving bureaucracy.

To the conventional management thinker, large, efficient networks are an oxymoron. They are not managerially possible. This is the same thinking described in Chapter 2 that hinders the acceptance of the possibility of self-organizing teams that are both innovative and disciplined. Conventional managers know that a large network will be chaotic and won’t be able to get things done. The human experience over thousands of years with Greek, Roman, and Chinese armies has shown that to be so. Big organizations require authority-based hierarchy. It’s not even worth thinking about anything different.

Large and effective networks are also thought to be impossible because of the constraint posed by what sociologists call “the Dunbar number.” This is the number that was first proposed in the 1990s by British anthropologist Robin Dunbar as the cognitive limit to the number of people with whom any individual can maintain stable social relationships—relationships in which an individual knows who each person is and how each person relates to every other person. Opinions differ as to the exact size of the limit, somewhere between 100 and 250. The commonly used value is 150.

The thinking is that in groups larger than 150, hierarchical rules and norms are required just to maintain a stable, cohesive group, let alone establish one that could implement a large, complex task. This constraint wasn’t relieved by the arrival of pervasive computer communications. That’s because the limit is a cognitive limit: The human brain, the sociologists said, simply can’t contain the number of interrelationships involved in any larger grouping. Until the brain gets bigger, we are limited to groups no bigger than 150.

Yet over the last fifteen years, we have seen examples of large and efficient organizations operating as networks in the world of Agile, like those of McChrystal’s force in Iraq. These examples include Riot Games, Spotify, the Network Management unit in Ericsson, and the Developer Division at Microsoft, all of which have more than 1,000 people collaborating on large, complex tasks in networks of Agile teams, rather than in pyramidal hierarchies.

Moreover, if we look beyond business organizations, we can see examples of even larger networks operating effectively. For instance, Alcoholics Anonymous has been operating as a network with more than 2 million members worldwide for more than seventy years, connecting more than 100,000 small groups.²⁰ There’s no leadership or top-level manager telling the small groups what to do. The network style of governance has helped avoid many of the pitfalls that political and religious institutions have encountered.

Even more dramatic is the example of the “cellular church,” a brainchild of Rick Warren, a pastor in Orange County, who created the Saddleback Church with a network of thousands of small prayer groups. It was based, writes Malcolm Gladwell, on “lots of little church cells—exclusive, tightly knit groups of six or seven who meet in one another’s homes during the week to worship and pray. . . . The small group was an extraordinary vehicle of commitment. It was personal and flexible. It cost nothing. It was convenient, and every worshipper was able to find a small group that precisely matched his or her interests.”²¹

What’s interesting about Warren’s Saddleback Church is that it’s not in the self-help business of solving the individual’s own problems. Like the world of Agile, its orientation is outward. It focuses attention on delivering value to others.

While we still have much to learn about the functioning of large networks, we can make some strong hypotheses as to what it takes to make them work:

1. The network has a compelling goal. Traditional managers are correct in thinking that a network will never work in a bureaucracy. That’s because work in a bureaucracy is dominated by rules, procedures, and instructions from a boss. It’s also unlikely to be effective if the goal of the firm is to make money for the shareholders and the C-suite. To function effectively, a network must be driven by a compelling shared goal. In the case of Agile management, the goal flows from the Law of the Customer: the naturally inspiring goal of adding value to customers. Once that obsession is shared throughout the organization, it matters less who is delivering the value. What’s important is the fact of delivery. Like on a basketball team, it’s less important who scores the goal: What matters is whether the goal is scored.

2. The network comprises small groups. These large networks are not built on a collection of individuals, like the old-style evangelists bringing huge crowds together in a big tent. Networks built on individuals will run into the legitimate constraint of the Dunbar number. “Successful movements,” says John Hagel, the director of Deloitte’s Center for the Edge, “are all organized around small, local action groups, who work together to achieve impact in very different contexts. These action groups are united by a loosely coupled network that enables them to seek help from others and to observe and learn from the diverse actions of each group what actions can achieve the greatest impact.”²² This requirement is of course consonant with the Law of the Small Team and McChrystal’s “team of teams.”

3. The groups have an action orientation. Small groups cultivate a spirit of collaboration, but it is a particular kind of collaboration. These groups are not about years of discussion, study, meditation, or reflection. They are about putting ideas into practice, not about abstract knowledge or even ideas for the sake of ideas themselves. These groups are about changing the world in a direction that the participants think is better in an important way.

4. The network is the sum of the small groups. The network doesn’t contain the small groups; rather, it is the sum of the small groups. The small groups are not groups within the organization: Conceptually, they are the organization.

5. The network’s legal framework stays in the background. In an organization functioning as a network, the legal aspects of an organization still exist. Boards of directors must be appointed. Bank accounts and contracts must have signatories. Officers are legally accountable for ensuring that money is properly spent, that compulsory government reports are signed and filed, and that any due taxes are paid. Legal procedures for recruiting, paying, and terminating any employees must be in place. But in a network, all this legal paraphernalia stays in the background: The substantive decisions are taken by the competence-based network. If that is not the case and the organization becomes a hierarchy of authority rather than a hierarchy of competence, then the network ceases to be a network: It has regressed back to a bureaucracy.

The organization as a network is one thing if the organization begins that way, like Riot Games, Spotify, Alcoholics Anonymous, or the “cellular church.” But can a top-down bureaucracy transform itself into a network? If so, how? Three main approaches have been tried.

One possible approach is top-down big-bang. Here the goal is to defeat bureaucracy in one great battle—by ripping bureaucracy out and replacing it with a network. As with Mao’s Cultural Revolution, the collateral damage here can be significant. People need time to adjust to any fundamental change in their organization’s management model, and they need to have a stake not only implementing whatever follows bureaucracy, but in the design of it as well. And if the organization moves too fast, there’s a risk of operational chaos. Moreover, a dramatic cultural change that is imposed on the organization from above may be in tension with the ethos of a self-managing organization and may create a backlash. It will generally make more sense to start as the organization means to go on. If an organization wants a highly empowered and involved workforce, then it should start by inviting everyone to help cocreate the new management model.²³ Nevertheless, the software firm Salesforce did have success with “a big bang” rollout—by carefully sidestepping the mistakes that other firms have made (see Box 4-3).

A second way is the gradualist bottom-up approach. The management sets out to replace bureaucracy incrementally, over a long period, working at it step-by-step. Says management guru Gary Hamel:

A lot of organizations have tried this [approach]. They have taken out a management layer or two, and simplified some particularly noisome processes. But they usually find that bureaucracy just grows back. The problem here is not a backlash, but half-hearted commitment. There’s a hope that one can push back bureaucracy without challenging the fundamental assumptions on which it’s built—that power trickles down; that decision rights are a function of title; that control must be imposed.

Over the decades, many organizations have run successful experiments with post-bureaucratic practices, but without a top-to-bottom commitment to bust bureaucracy, these little efforts are either soon aborted, or marginalized. No organization will ever defeat bureaucracy through half-measures—bureaucracy is too resilient. If the CEO is afraid to say “bureaucracy must die,” then it won’t.²⁴

Even when high-performing Agile teams are plugged into a bureaucracy, there is inherent tension, as the goals and modalities are different. The situation isn’t stable. Either the Agile teams will take over the organization, or more likely, the bureaucracy will crush the Agile teams.

The third approach is a combination of top-down and bottom-up, reflecting Carlson’s Law. It’s an organic process that is simultaneously revolutionary and evolutionary. In the examples of Barclays and Ericsson (discussed in Chapter 1), it was a combination of bottom-up and top-down. The Agile movement got going as a grassroots movement and then found support, encouragement, and funding from the top. When this happens, the Agile movement can have genuine enthusiasm from both the top and the bottom. Another example of this happy combination is Microsoft’s Developer Division, to which we turn in the next chapter.