Solve Communication Complexity With Networks Of Small Teams

Joost
Written by
- 4 min read

Work is solving other people’s problems. Most progressive companies on our Bucket List think they do that best when structured as networks of teams, rather than hierarchical pyramids. Teams in radically decentralized networks are often self-managed and highly autonomous. And these teams are often very small. They rarely consist of more than 15 people. But why are self-managed teams in these networks typically so small? There are very good reasons.

They have to do with Metcalfe's Law, and the structure of the human mind—which has evolved such that we suck at handling large numbers of relationships.

Metcalfe's Law

Let’s start with this law, originally presented around 1980 by Robert Metcalfe. He was an Internet/Ethernet pioneer, and his law captures many of the effects of such networks.

It says that the effect of a (telecommunications) network is proportional to the square of the number of connected devices. Which also explains why networks like the Internet have become so powerful. Simply, a network's value increases exponentially with size. That is, the more people using it, the more their participation enhances it.

This exponential growth is generally considered a good thing. Especially in digital networks.

From digital to social

But Metcalfe's law has another side, especially when it comes to social interactions. Imagine how Metcalfe's law translates easily to the social dynamics at play between team members.

article photo

So, for a team increasing from three to four team members, although there is only one more person, the number of efficient communication lines suddenly doubles from 3 to 6.

Similarly, imagine being in a team of five team members. Here there are 10 such communication lines at play. But double the number of team members to ten, and you suddenly have a complex network of 45 communication lines.

As more people try to communicate with each other, the number of interactions increases immensely. Here we hit a problem—our brain simply can't handle this.

Limitations of the mind

The structure of the mind is that it can't really handle very large numbers of relationships. Scientists like Robin Dunbar have told us for decades that our social world is very small-scale.

They say there are approximately 5 people we can have tight relationships with, and another 15 people we can have slightly less intense relationships with. Think of sports teams, for example. They are rarely being bigger than 15 people.

The same principles apply in business and management. There are natural limits to the number of people we can communicate, coordinate and perform with. There are limits to the number of colleagues you happily drink coffee with, and even fewer that you invite over for dinner.

Transaction costs

Communicating, coordinating and contracting take place in all organizations. And these interactions should take place as efficiently as possible, because they are hidden costs. Academics call them transaction costs.

Many decades ago, organizations invented hierarchies (including the role of ‘middle manager’) to solve this problem. Managers were installed to lower transaction costs inside organizations. As they did so, they took these tasks away from front-line employees, and reduced costs.

Two problems

There are two things that make the ‘inventing’ of managers (and their spans of control) problematic.

First, the wellbeing of the front-line employees: they did not benefit much from this change. Most people don’t like to be stripped of tasks like communication and coordination. It didn't really motivate or engage front line staff.

Second, the possibilities that digital technologies brought to internal communications meant transaction costs of one-to-one and one-to-many interactions dropped basically to zero. This questions the manager's role from a 'transaction cost' perspective.

Conclusion

One result is evident in the progressive organizations. They have departed far from traditional managerial hierarchies. Instead, they structure themselves as decentralized networks of teams.

These networks have no (or few) middle managers. They feature highly autonomous teams where members take care of communication, coordination and contracting themselves.

But as Metcalfe's Law shows, when there are no managers, teams must be small. And small enough that members do not get overloaded with communication and information which our brains can't handle!

Subscribe to our newsletter

Share


leave a reply

Replies (8)

Tomas Horak

Tomas Horak

I see a basic issue with comparing computer networks to human or social network. And like all models, there is always a simplicistic trap somewhere. Here it‘s the assumption, that human to human communication can be expressed graphically as one straight line and mathematically as a value of 1.
The system of one single person is too complex to be expressed with mathematical values. Here Metcalfe‘s ideas seem to be valid. The more human systems are connected to each other, the more exponentially the complexity is rising. But what is the use of expressing human or social systems with mathematical values? And what about the ethics of doing so? What if the economy itself would transform into a system, where absolute values are coexisting with relative ones?

| | 17 | Flag
Anita

Anita

It's similar to the scrum book which says that an agile team should be up to 9 people, for similar reasons - it's shown that small teams work better than big ones, and you trust each other and build relationships in small teams better than in big ones.

As the guide says, the team should should be 'small enough to remain nimble and large enough to complete significant work '.

| | 5 | Flag
Chuck Blakeman

Chuck Blakeman

Great picture of one of the issues with having managers. The bigger issue is the lack of distributed decision making. When managers make decisions for others, they dehumanize those others. If there is one manager and two team members, there are only six lines, but decision making still only goes one direction. Managers should be eliminated primarily because they dehumanize the people around them.

| | 4 | Flag
Chris

Chris

This is an excellent article. I think Chuck Blakeman (above) has it right. Team size is important as demonstrated by the charts in the article, but the real catch here is that teams must also be given the authority to make decisions and set the best course of action given the business needs and goals. I believe that a team can elect a leader, rather than a manager because it is useful to have someone dedicated to management type activities and to communicate with the rest of the organization. That person does not need to be an official "manager" though.

| | 1 | Flag
Complex

Complex

Surely 4 people create 9 different relationships. That big diagram is fundamentally wrong:

Relationships = (Members-1)²

Simple small network maths?

| | 0 | Flag
Joost

Joost

Surely 4 people create 9 different relationships. That big diagram is fundamentally wrong:

Relationships = (Members-1)²

Simple small network maths?

Complex

Hi Complex,

Melcalfe's law states:

Connections = nodes(nodes-1)/2

If you take that law, then 4 nodes is 6 connections...

| | 0 | Flag
David

David

Hi Complex,

Melcalfe's law states:

Connections = nodes(nodes-1)/2

If you take that law, then 4 nodes is 6 connections...

Joost

Correct, as a connection between two people is a single, bi-directional connection, not one in each direction.

Thanks for the post - my manager shared it to explain her new team structures.

| | 0 | Flag
AA

AA

Remember Conway's law! "Any organisation that designs a system (defined broadly) will produce a design whose structure is a copy of the organisation's communication structure"

| | 0 | Flag
Flags are private, only visible to forum moderators. Be specific: "It's spam/off-topic/inappropriate because..."
submit

Leave a reply

This field is required. Only use letters, digits, spaces and minus characters.
This field is required.
This field is required.
We only use your email for spam detection purposes.