How To Raise The Collective Intelligence Of Teams


In the most recent Meditation, I wrote about True Asynch, adopting the thinking of researchers Christoph Riedl and Anita Williams Woolley about ‘burstiness’: where daily communication is constrained into as narrow a timeframe as possible:

Our research suggests that such bursts of rapid-fire communications, with longer periods of silence in between, are hallmarks of successful teams. Those silent periods are when team members often form and develop their ideas — deep work that may generate the next steps in a project or the solution to a challenge faced by the group.  Bursts, in turn, help to focus energy, develop ideas, and achieve closure on specific questions, thus enabling team members to move on to the next challenge.

We’d certainly want our teams to be successful, so adopting burstiness — while a big departure from the default behavior at most companies — seems like a good idea.

But I wondered about the connection between the practice of truly asynchronous communication and team success. How does true asynch lead to greater success for teams?

Perhaps unsurprisingly, Anita Williams Woolley was the lead in another research effort that looked into team effectiveness, and she and her colleagues found that a team’s collective intelligence was influenced quite strongly by the makeup of the group.

Counterintuitively, adding people of higher intelligence to a team does not necessarily lead to higher collective intelligence, measured by teams undertaking diverse cognitive tasks. Collective intelligence is not some arithmetic average of the intelligence of those in the group: it is an emergent property arising from the interactions of the members.

Instead, the key factor in team effectiveness is the social sensitivity of its members: how well the members perceive each other’s emotions. This turned out to be strongly correlated with gender: women are generally more socially sensitive. So the fastest way to improve team effectiveness is to alter the ratio of social sensitivity (which may translate to ‘add more women’) in the group. This was an unexpected finding of the research.

But I was initially curious about burstiness of communication in effective groups, and this research didn’t test for that explicitly, but did find a negative correlation between group effectiveness and dominance of conversations by one or few members of the group. Effectiveness was highest when ‘conversational turns were more evenly distributed’, said Woolley. This lines up with my experience with teams: team meetings are most effective when everyone participates and there is little imbalance in speaking time. When I lead meetings, I work hard to achieve that.

The biggest takeaway from that research is that group collective intelligence accounts for about 40% of the variation in performance on the range of tasks in the test.

Given the earlier research about the impacts of bursty communications on group success, what specifically is different in teams that practice true asynch communications? One element of the magic comes from raising collective intelligence through social sensitivity and balanced communication in meetings, but what is it about the time that truly asynchronous teams spend apart?

Ethan Bernstein, Jesse Shore, and David Lazar researched that question deeply by setting up an experiment to see which of three kinds of cooperative work lead to the best results: groups in constant communication, groups with no interactions (working independently), and groups with bursty (intermittent) interaction. The test set was the traveling salesman problem.

The third kind of group performed best, as the researchers explained [emphasis mine]:

Groups in the intermittent social-influence treatment found the optimum solution frequently (like groups without influence) but had a high mean performance (like groups with constant influence); they learned from each other, while maintaining a high level of exploration. Solutions improved most on rounds with social influence after a period of separation.

So the second half of the magic of collective intelligence is the alternation between periods of group interaction or we-time and individual exploration during me-time. As the authors clarified:

When people interact and influence each other while solving complex problems, the average problem-solving performance of the group increases, but the best solution of the group actually decreases in quality. We find that when such influence is intermittent it improves the average while maintaining a high maximum performance.

So, achieving maximum results depends on separating the team members to work independently, then coming back together to share and to learn from each other, and then returning to solo exploring of the problem space.

Again, this is the cadence of effective use of me-time and we-time, circling from one to the other, and the alternation is key. While having access to shared information about the team members’ individual progress is helpful, it is not by itself a guarantee of success.

Bernstein and company make the bottom line very clear:

The results imply that technologies and organizations should be redesigned to intermittently isolate people from each other’s work for best collective performance in solving complex problems.

Adding it all up, the best way to raise the collective intelligence of a team involves the following:

  • Selecting members for their social sensitivity, and not specifically for intelligence.
  • When in (as few and brief as possible) meetings, try to balance communication across the group’s members, so that opportunities to learn are increased.
  • Intermittently isolate the team members, so that group influence does not lead to a decrease in group performance.

As in many other cases, these findings have had little influence in the business world, partly because the findings are relatively new, but mostly because our ways of doing our work are so deeply ingrained, and therefore, seldom questioned.

Here we have definitive proof that some of the norms of business are leading to poor performance. But we also know what to do about it.

Facebook iconTwitter IconLinkedIn icon