Most communities and organisations evaluate their gatherings by what was decided. What was on the agenda, what was agreed, what actions were allocated and to whom. These are reasonable things to track. They are also, it turns out, almost entirely the wrong things to focus on if you want to understand why some groups consistently think better together than others.

Alex Pentland, director of MIT’s Human Dynamics Laboratory, spent years trying to answer a deceptively simple question: what actually makes some groups outperform others? His research team equipped more than 2,500 individuals across a wide range of teams and industries with wearable electronic sensors that tracked not what people said but how they communicated. Tone of voice, body language, turn-taking patterns, who spoke to whom, for how long, and in what sequence. The content of the conversations was largely irrelevant to what the data revealed. The pattern of the conversations predicted almost everything.

The most striking finding was this: communication patterns were as significant a predictor of group success as intelligence, personality, and talent combined. Pentland’s team could identify which groups would outperform simply by looking at the interaction data, without ever meeting the people involved or knowing anything about what they were working on. What they were measuring, without initially knowing it, was what Pentland would later call idea flow.

The pattern that predicts performance

Idea flow is not about having better ideas. It is about how ideas move through a group. Whether they circulate freely or get stuck. Whether they reach everyone or only the few. Whether the group is genuinely thinking together or performing the appearance of collective thinking while one or two people do the actual intellectual work.

Pentland’s research identified three communication dynamics that consistently distinguished high-performing groups from those that struggled. The first was energy: the number and quality of exchanges among group members, with face-to-face conversation producing by far the most value. The second was engagement: the degree to which members communicated directly with each other rather than routing everything through a central figure. The third was exploration: the extent to which the group reached beyond its own boundaries to gather perspectives and information from the wider environment.

Of these three, engagement produced some of the most practically significant findings. In high-performing groups, members talked and listened in roughly equal measure. Contributions were short. People addressed each other directly, not just the leader or the chair. There were side conversations, back-channel exchanges, the informal two-person moments that most meeting facilitators try to suppress. These were not distractions. They were, in Pentland’s data, one of the primary mechanisms through which groups actually processed ideas together.

The groups that looked most orderly, in which one person spoke at a time and everything was directed through the centre, were frequently the least generative.

Hub and spoke versus the web

There is a structural pattern that emerges in almost every community meeting, every team away-day, every residents’ forum, every staff consultation. Someone sits at the front or the head of the table. Questions are directed toward them. Responses come from them. The conversation flows inward and outward from a single point, like the spokes of a wheel all meeting at the same hub.

This is not always the result of a dominant personality or poor facilitation. It is often simply the physical architecture of the room, the position of the chairs, the direction the tables face, the implicit signal that there is someone here whose job it is to receive the group’s input and give something back in return. The structure produces the behaviour before anyone has said a word.

Pentland’s research suggests that this hub and spoke pattern, however familiar and however comfortable it feels, is one of the most consistent suppressors of collective intelligence available to a group. When ideas flow only through the centre, they are filtered, delayed, and reduced. The person at the hub becomes a bottleneck. Their perspective, however well-intentioned, shapes what the group is able to think together far more than it should.

The alternative is a web: a pattern of conversation in which ideas move between all members, in which the centre is not a person but a shared question, and in which the energy of the group is distributed rather than concentrated. Pentland’s data showed consistently that groups with this web-like pattern of conversation outperformed hub and spoke groups on creativity, problem-solving, and decision quality, regardless of the intelligence or experience of the individuals involved.

The implication is both humbling and hopeful. The quality of what a group produces is not determined primarily by who is in the room. It is determined by how the room is structured and what kind of conversation that structure makes possible.

The lunchroom table and the coffee break

One of the most striking findings in Pentland’s research concerned not the content of meetings but what happened around them. In a study of a bank call centre, his team found that the best predictors of team productivity were energy and engagement outside formal meetings. Together, these two factors explained one third of the variation in performance between teams.

Acting on this finding, the researchers advised the centre’s manager to change the employees’ coffee break schedule so that everyone on a team took their break at the same time. The suggestion ran counter to standard efficiency logic: staggered breaks kept the phones covered. But the manager, faced with persistently underperforming teams, tried it. Average handling time fell by more than 20 percent among the lowest-performing teams and by 8 percent across the centre overall.

Nothing about the formal structure of the work had changed. No new training had been delivered. No new process had been installed. The only thing that changed was that people who worked together were given unstructured time to talk to each other, informally, away from their workstations. The idea flow improved. The performance followed.

The same principle applies to a neighbourhood meeting, a community planning session, or a team away-day. The most important conversations are not always the ones that appear on the agenda. They are the ones that happen before the formal session begins, in the ten minutes after it ends, in the corridor while people are putting on their coats. These are not peripheral to the group’s work. They are often where the group’s actual thinking takes place.

Designing for idea flow

Most gatherings are designed for order. Rows of chairs facing a screen. A timed agenda with no slack. Breaks kept short to maintain momentum. People seated with their own team or their own department. The room optimised for the transmission of information from the front to the rest.

Pentland’s research suggests this is almost precisely the wrong design if the goal is collective thinking rather than information delivery. The physical environment of a gathering is not neutral. It either enables idea flow or suppresses it, and most conventional meeting design suppresses it systematically.

Several practical shifts follow directly from the data:

Longer breaks are not inefficient. They are generative. The informal conversation that happens over coffee is not wasted time between the real work. For many groups it is where the real work actually happens, where the idea raised in the session gets properly examined, where the person who did not speak in the room says the thing that changes the direction, where the connections form that the formal agenda could not have planned. Building longer, unstructured breaks into a gathering is not a concession to comfort. It is an investment in the quality of the group’s thinking.

Satellite spaces matter. A single room with everyone facing the same direction produces a single, centralised conversation. Multiple smaller spaces, a side room, a corner with chairs arranged in a circle, a table in the corridor where people can break away and return, create the conditions for the parallel conversations that Pentland’s data identified as essential to high-performing groups. The community consultation that keeps everyone in the main hall for three hours will consistently produce less genuine thinking than one that deliberately creates spaces for smaller exchanges to happen alongside the plenary.

Moving people around disrupts the echo chamber. When a group sits in the same configuration for an entire session, conversation calcifies around existing relationships. The person next to you is the person you already know, and the ideas that circulate are largely the ideas that were already in circulation before the meeting began. Structured mixing, a different seating arrangement after the break, a deliberately formed small group that cuts across the usual clusters, introduces the new connections and the non-redundant perspectives that Pentland’s exploration dynamic depends on. It feels slightly awkward. That slight awkwardness is the feeling of idea flow beginning.

Informal time before the session starts is not dead time. It is the warm-up for the group’s collective intelligence. The ten minutes before a community meeting in which people are standing around with a cup of tea and talking to each other is not merely social. It is the moment in which the web of conversation begins to form. Starting a meeting abruptly, on the exact minute, with no prior informal contact, is the equivalent of asking people to sprint before they have stretched.

None of this requires a large budget or a specialist facilitator. It requires a different set of beliefs about what a gathering is for, and a willingness to design for the conversations that matter rather than the agenda that is easiest to control.

The citizenship of the side conversation

This matters for how we understand citizenship in a group setting. We have been taught, most of us, that the responsible participant is the one who waits their turn, addresses the chair, and stays on topic. The side conversation is a form of rudeness. The informal exchange is a distraction. The people who are talking to each other when they should be listening to the speaker are the ones who are not taking this seriously.

Pentland’s research suggests the opposite. The participants who generate informal exchanges, who reach across the table to a person they have not spoken to yet, who continue the conversation in the corridor, who ask the question that was not on the agenda, are not undermining the group’s work. They are doing some of its most important work. They are creating the conditions in which idea flow becomes possible, in which the group’s collective intelligence can exceed the sum of its individual parts.

Citizenship, in a gathering, is not just about what you contribute when you have the floor. It is about whether you are actively building the web of conversation, whether you are creating the connections between people and ideas that no formal agenda can manufacture. The meeting that looks most orderly is not always the one doing the most thinking. And the participant who seems most engaged is not always the one talking most loudly to the chair.

Questions for reflection

Think about the last community meeting or team gathering you attended. Was the conversation a hub and spoke, with most exchanges routed through one central figure, or a web, with ideas moving freely between all members? What effect did that pattern have on what the group was able to produce?

What small change in the physical arrangement of your next gathering, the shape of the seating, the length of the break, the structure of the opening, might shift the pattern of conversation from concentrated to distributed?

Where does the real thinking happen in your community or your team: in the formal meeting or in the conversations around it? What would happen if you designed your next gathering around that knowledge rather than despite it?

Is there someone in your regular meetings who rarely speaks in the formal session but who often has something important to say in the conversation afterwards? What would it take to create the conditions in which that contribution could happen in the room?

When you are in a gathering, are you actively building the web of conversation, reaching toward people you have not yet spoken to, creating connections that the agenda did not plan for? Or are you directing your energy toward the centre and waiting to be called?

Sources

Pentland, A. (2012) The new science of building great teams. Harvard Business Review, 90(4), pp. 60–69.

Pentland, A. (2014) Social physics: How good ideas spread, the lessons from a new science. New York: Penguin Press.