Team Genius: The New Science of High-Performing Organizations (4 page)

Besides providing a devastating indictment of unconstrained government growth, Parkinson’s law had an interesting secondary effect: it forced social scientists, if they were to defeat the negative impact of the law, to determine the right number of bureaucrats for any given task.

Parkinson himself was fairly flexible. He believed that almost any team of fewer than twenty members could be made to work effectively, arguing that adding any more members would, once again, devolve the group into multiple smaller teams. He did, however, make one exception: in his experience, he said, a team of
eight
members would never reach a consensus decision—apparently because (as 4×4 or 2
³
) there was no tiebreaker. That would seem to contradict the ubiquitous eight-man military units—until one remembers the staff sergeant or master sergeant in command, whose decision is final.

THE MYSTERY OF DUNBAR’S 150 AND 1,500

Driven by the demands of Parkinson’s law, social scientists and anthropologists fanned out to study everything from factories
to primitive tribes—and made some stunning discoveries. Studying everything from Hutterite religious communities to the Yanomamo people of the Brazilian jungle, as well as ethnographic literature, the British anthropologist Robin Dunbar discovered that the same human group sizes appear over and over. He called them “clusters of intimacy,” and he identified “cliques” of five people, “sympathy groups” of twelve to fifteen people, and “bands” of up to thirty-five members.
⁶

But Dunbar’s biggest discovery, now named after him, was that there appeared to be an upper limit to team size. Precisely: 147.8, normally rounded up, as the “Dunbar number,” to 150. It is a number that appears with stunning regularity. For example, the Yanomamo people have consistently over the centuries divided their tribes whenever they approach 200 members. As for the Hutterites, they long ago learned to split their colonies when their population reaches 150. In the Domesday Book, the average size of Welsh and British villages at the time of the Norman invasion was . . . 150 residents.

Almost everywhere you look, you can find the Dunbar number. For hundreds of years, that basic unit of almost every army in Western civilization, the company, has consisted of roughly 150 members. The likely number of “friends” you will have on Facebook or regular followers on Twitter? 150 to 190. The total population of households to whom Brits sent Christmas cards in 2000 (that is, before email greetings)? 153.5. Dunbar’s explanation? The figure of 150 seems to represent the maximum number of individuals with whom we can have a genuinely social relationship, the kind of relationship that goes with knowing who they are and how they relate to us. To put it another way, it’s the total number of people you would not feel embarrassed about joining uninvited for a drink if you happened to bump into them individually in a bar.
⁷

Robin Dunbar hasn’t stopped at 150 as an archetypical team size. There are other Dunbar numbers as well:

•
      
3 to 5: This is the circle of our very closest friends. Here Dunbar agrees with other social scientists.

•
      
12 to 15: The group of friends and family whose death we would deeply mourn. Again, Dunbar agrees with other researchers. He also adds an interesting example of a group of this size, born in historical experience: juries. This group is also connected with the notion of
deep trust
—that is, it’s the number of people from whom you can accept a small amount of betrayal without severing ties.

•
      
50: This is a new number. Dunbar derives it, as usual, from historical sources; it is “the typical overnight camp size among traditional hunter-gatherers like the Australian Aboriginals or the San Bushmen of southern Africa.”
⁸

•
      
150: The Dunbar number. In the words of Dave Snowden, a Welsh knowledge management expert, the Dunbar number is “the number of identities that you can maintain in your head with some degree of acquaintances that an individual can maintain. It does not necessarily imply that you trust them, but it does mean that you can know something about them and their basic capabilities. In other words you can manage your expectations of their performance and abilities in different contexts and environments.”
⁹

Note that these natural groups in Dunbar’s theory seem to scale by a factor of three. To date, no one seems to have come up with an explanation for this underlying multiplier.

•
      
1,500: This largest Dunbar number appears to be something of an outlier, one for which Dunbar himself doesn’t seem to have an explanation. Are there one or two still-missing Dunbar numbers in between the last two, whose value and purpose are as yet undiscovered? Dunbar has suggested that there might be one at 500, which would correspond to the average person’s
number of acquaintances. Then comes this number, 1,500, which once again has a number of historic and contemporary examples and an underlying link to basic human behavior.

In terms of examples, 1,500 members is roughly the size of a large military battalion, the smallest unit capable of independent operation. It is also, as Dunbar has noted, the average size of a tribe in hunter-gatherer societies, and the number of people who speak the same language or dialect. But perhaps the most common and compelling example of the use of 1,500 as a limit on group size can be found in the corporate world.

Hewlett-Packard Co. in the 1950s and 1960s is sometimes described as the greatest corporation of all time. It was as technologically innovative as Apple was fifty years later, it offered more progressive employee programs (stock options, profit sharing, flextime, and so forth) than any company before or since, and—not surprisingly—it set records for employee happiness and morale that have never again been matched by a large corporation.

Interestingly, in 1957, as the company approached 1,500 employees, the two founders began to sense that something had changed in their relationship to other HPers—and something radical had to be done. Wrote David Packard in his autobiography,
The HP Way
, “It [was] increasingly difficult for Bill and me to know everything well and to have a personal knowledge of everything that was going on.” So they decided that it was time to break up HP from a monolithic company into a divisional one. Bill Hewlett said, “Out of this came the concept that what we should probably do is divisionalize. We had [nearly] 1,500 people at that point and we thought it was too big. By dividing up into two or three small units, we might be able to keep that personal touch.”

From that day to the present, HP’s divisions have traditionally split up when they reach 1,500 employees. That decision is credited with making Hewlett-Packard in the 1960s and 1970s one of
the most nimble large companies of all time. What Bill and Dave sensed intuitively, and executed with their typical decisiveness, has been imitated by countless other companies in the years since. The 1,500-employee division is one of the most common organizations on the global business landscape.

What Hewlett and Packard sensed on the ground has also been described by Dunbar in human terms:

You have an inner, inner core of intimate friends and relations, of about five, and then there’s the next layer out, it’s about 15. If you like to think of those as best friends, perhaps, they’re the people you might do most of your social Saturday evening barbeques with, and that of course, includes the five inside. And then this next layer out is 50 (you might think of those as good friends), and the 150, your friends. And then we know there are at least two more layers beyond that: one at 500 which you might think of as acquaintances, so again this is including everybody within the 150 as well; finally, one at 1,500 who are basically the number of faces you can put names to.
¹⁰

WHAT DAVID PACKARD AND BILL HEWLETT KNEW

Another way of looking at Dunbar numbers is that five is the number of your most intimate friends and partners (“clique”) and is a number that, not coincidentally, corresponds with the limits of our short-term memory. Fifteen is the number of people with whom we can have deep trust in the face of almost any turn of events (“sympathy group”). Fifty is a familial grouping, a small tribe with whom we can securely travel in dangerous country (“band”). One hundred fifty is the optimal size for a group of people living together in a community (“friendship group”); it corresponds to the number
of people whose individual characteristics and behaviors the human brain can effectively remember. Five hundred is the number of people with whom we can remain nodding acquaintances (“tribe”). And fifteen hundred is the limit of our long-term memory, the total number of people to whom we can mentally attach a face if we hear his or her name (“community”).

Why this last number, 1,500, works is self-evident. If you live in a community of 150 people, you feel pretty comfortable walking down the street, because you know everyone in town. A stranger—that is, a potential threat—is instantly obvious. With 500 members in the community, that comfort is strained a little, because you are likely to run into people whom you know but haven’t spoken to in a long time. But still, strangers stand out. But with 1,500 neighbors, that comfort ends. Now you begin encountering people who you are not quite sure if you’ve ever met before . . . and the world suddenly becomes a much riskier place.

For Bill Hewlett and Dave Packard, the first threshold was passed in the late 1940s, when they had to end their tradition of personally handing out Christmas bonus checks and greeting everyone by name. By the early 1950s, everyone was still in the same building complex on Page Mill Road in Palo Alto, but the two founders now found themselves nodding at employees whose faces they knew but couldn’t name. The tipping point came in the late 1950s, when Bill and Dave began encountering employees they had never seen before and whose only proof of employment was their name badge. And that’s when they made their move.

By the end of the 1980s, HP had more than forty divisions, all of them consisting of about 1,500 employees—and was regularly creating new ones. The company even developed a standard architectural design for its divisional facilities, accommodating for a maximum capacity of . . . 1,500 employees.

Other corporations—in emulation of one of the world’s most successful companies, or via trial and error—soon found themselves
replicating this model. The PC division of Sony in Japan, for example, has 1,500 employees. Interestingly, but probably not coincidentally, 1,500 is usually considered the maximum employment of a medium-size company. And a surprising number of companies—such as Facebook, Google, and Twitter—hit the 1,500-employee mark in the ramp-up to their initial public offering . . . suggesting that there may be an even deeper link between going public, with the inevitable transformation in employment that takes place inside a company between employees of a private versus a public firm, and the boundaries of a natural community. Many employees of private firms remark, often with despair, that their “company is no longer the same place” after the IPO, that it is no longer a big “family but now an anonymous organization that is filled not with true believers but with salary men and women intent on their benefits and résumés.”

THE HARD MATH OF IMPOSSIBLE CONNECTIVITY

Before we close out this section about the manifold forces at work in making teams of certain sizes both inevitable and desirable, there is one last structural force, beyond genetics, that appears to be driving the creation of teams. It is
the mathematics of networks
.

The reason the Internet is so powerful—and Metcalfe’s law, as discussed in chapter 1, so useful—is that each new user arriving on the Web doesn’t just add his or her solitary new node, but the billions of new connections that are then created. To understand the magnitude of this effect, let’s look at the smallest number of connections in a team and progress forward, showing the total number of connections:

2 members = 1 connection

3 members = 3 connections

4 members = 6 connections

5 members = 10 connections

6 members = 15 connections

16 members = 120 connections

32 members = 496 connections

Notice that after the low-digit numbers, the equation settles down to N(N − 1)/z, where
N
is the number of team members. The sheer complexity of the network grows much, much faster than the number of team members does (at the Dunbar number of 1,500, the number of interconnections reaches 1,124,250).

And that creates an obvious problem. Human beings can handle, much less maintain, only comparatively small numbers of connections. That’s why relationships degrade so quickly as the number of team members grows. Most of us are pretty good at remaining in contact with five or six other people on a constant basis. But doing so is a whole lot tougher with a dozen or more. At fifty? Not even those rare individuals with a photographic memory for faces and names can still stay in touch with the team in the same way as they did with a half dozen others. Even with the tools of social networks, texting, wide area networking, and global wireless telecom, we
still
don’t have the time or the bandwidth to continuously maintain hundreds or thousands of close personal connections.

That’s why bigger teams almost never correlate to a greater chance of success. The noted Harvard psychologist J. Richard Hackman once said, “Big teams usually wind up just wasting everybody’s time.”
¹¹
He explained that

[the] fallacy is that bigger teams are better than small ones because they have more resources to draw upon. A colleague and I once did some research showing that as a team gets bigger, the number of links that need to be managed among members goes up at an accelerating, almost exponential rate.
It’s managing the links between members that gets teams into trouble. My rule of thumb is no double digits. In my courses, I never allow teams of more than six students. Big teams usually wind up just wasting everybody’s time. That’s why having a huge senior leadership team—say, one that includes all the CEO’s direct reports—may be worse than having no team at all.
¹²