Tic-Tac-Toe and the Squirm Test: Building Trust and Shared Understanding

Elliott's Monster Face

Trust. Shared understanding. Shared language. I constantly mention these as critical elements of collaboration. But how do we develop these things, and how do we know if we’ve got them?

We can start by playing Tic-Tac-Toe, then by applying the Squirm Test.

Tic-Tac-Toe

Many years ago, one of my mentors, Jeff Conklin, taught me a simple exercise that gave me a visceral understanding of why trust, shared understanding, and shared language were so important, as well as some clues for how to develop these things.

First:

  1. Find a partner.
  2. Play Tic-Tac-Toe.

Try it now! What happened?

Hopefully, nobody won, but don’t stress too much if you lost. It happens!

Second: Play Tic-Tac-Toe again, except this time, don’t use pen or paper.

Try it now! What happened?

In order to play, you needed to come up with shared language to describe positions on the board. You probably managed, but it was almost certainly harder.

Third:

  1. Play Tic-Tac-Toe without pen or paper again, except, this time, play on a four-by-four board — four rows, four columns.
  2. Play until it’s too hard to play anymore, until someone has won, or until there’s a dispute about what the board looks like. When you’re done, both you and your partner should draw what you think the board looks like without looking at each other’s work. Now compare.

Try it now! What happened?

Research suggests that we can hold between five to nine thoughts in our head at a time before our short-term memory begins to degrade. This is why American phone numbers have seven digits. It’s also why three-by-three Tic-Tac-Toe (nine total squares) without pen and paper felt hard, but doable, whereas increasing the board by just one row and column (sixteen total squares) made the game feel impossible.

How many ideas do you think you’re holding in your head after just five minutes of moderately complex conversation? How often are you using some kind of shared display — a whiteboard, a napkin, the back of an envelope — to make sure that everyone is tracking the same conversation?

While you were playing, how much did you trust that the two of you were seeing the same board at all times? Were you right?

In cooperation theory, the most successful groups trust each other by default. You almost certainly assumed that your partner knew the rules of Tic-Tac-Toe and was playing it fairly to the best of his or her ability. If you already had a strong relationship with your partner, you probably trusted him or her even more.

But trust is fragile, and it’s not always relational. If it’s not constantly being reinforced, it weakens. A lack of shared understanding is one of the easiest ways to undermine trust.

With the Delta Dialogues, we were dealing with a uniquely wicked and divisive issue — water in California. As a facilitator, you always want to get the group out of scarcity thinking. But water is a zero-sum game, and no amount of kumbaya is going to change that. Moreover, we were dealing with a half-century legacy of mistrust and a group of participants who were constantly in litigation with each other.

We did a lot of unique, relational work that played an important role in the success of Phase 1 — rotating site visits, asking people to share their favorite places in the Delta, implementing a buddy system, leaving plenty of space for breaking bread. But we were not relying on these things alone to build trust.

Our focus was on building shared understanding through a mapping process that allowed the group to see their ideas and track their conversations in real-time. Prior to our process, the group had been attempting to play multi-dimensional Tic-Tac-Toe with thousands of rows and columns… and no pen and paper. We brought the pen and paper, along with the ability to wield it skillfully.

Like many of my colleagues, I believe strongly in building and modeling a culture where people are engaging in powerful, constructive, sometimes difficult conversation. Unfortunately, it’s not enough to get people into a circle and to have them hold hands and talk about their feelings. The more wicked the problem, the more inadequate our traditional conversational tools become, no matter how skillfully they are wielded.

This recognition is what separates the Garfield Foundation’s Collaborative Networks initiative from similar well-intentioned, but misguided initiatives in the nonprofit and philanthropic worlds, and it’s why I’m working with them right now. We happen to be employing system mapping (and the talents of Joe Hsueh) right now as our “pen and paper” for developing that shared understanding, but it’s how and why we’re mapping — not the specific tool itself — that separates our efforts from other processes.

The Squirm Test

How do you know how much shared understanding you have in the first place? And if you choose to employ some version of “pen and paper” to help develop that shared understanding, how do you know whether or not you’re intervention is effective?

Many years ago, I crafted a thought experiment for doing exactly that called the Squirm Test.

  1. Take all of the people in your group, and have them sit on their hands and in a circle.
  2. Have one person get up and spend a few minutes describing what the group is doing and thinking, and why.
  3. Repeat until everyone has spoken.

You can measure the amount of shared understanding in the group by observing the amount of squirming that happened during the presentations. More squirming means less shared understanding.

You can implement the Squirm Test in all sorts of ways, and it even appears in different high-performance communities in real-life. For example, the Wikipedia principle of Neutral Point of View is essentially the Squirm Test in action. If you read an article on Wikipedia, and it makes you squirm, edit it until the squirming goes away. If enough people do that, then that article will accurately reflect the shared understanding of that group of people and will thus achieve Neutral Point of View.

Toward the end of Phase 1 of the Delta Dialogues, we designed a whole day around the Squirm Test. We had participants capture on flipcharts what they thought the interests and concerns were of the other stakeholder groups. Then we had participants indicate whether these points represented themselves accurately and whether they found shared understanding on certain issues surprising. There was very little squirming and quite a bit of surprise about that fact. It was a turning point for the process, because the participants were able to see in a visceral way how much shared understanding had been built through all of their hard work together.

Last week, I was describing the Squirm Test to Rick Reed, the leader of Garfield Foundation’s Collaborative Networks initiative. He pointed out a discrepancy that I had not thought of before. “People might not be squirming because there’s no shared understanding. They might be squirming because, after seeing the collective understanding, they realize that they’re wrong!”

This is exactly what happened with the Garfield Foundation’s first Collaborative Network initiative, RE-AMP. When you are able to see the whole system, including your place in it, you may discover that your frame is wrong. The kind of squirming that Rick describes is the good kind. Understanding how to design for it is a topic for another day!

Maximizing Collective Intelligence Means Giving Up Control

Ant City

Today marks the 45th anniversary of the Mother of All Demos, where technologies such as the mouse and hypertext were unveiled for the first time. I wanted to mark this occasion by writing about collective intelligence, which was the driving motivation of the mouse’s inventor (and my mentor), Doug Engelbart, who passed away this past July.

Doug was an avid churchgoer, but he didn’t go because he believed in God. He went because he loved the music.

He had no problem discussing his beliefs with anyone. He once told me a story about a conversation he had struck up with a man at church, who kept mentioning “God’s will.” Doug asked him, “Would you say — when it comes to intelligence — that God is to man as man is to ants?”

“At least,” the man responded.

“Do you think that ants are capable of understanding man’s will?”

“No.”

“Then what makes you think that you’re capable of understanding God’s will?”

While Doug is best known for what he invented — the mouse, hypertext, outlining, windowing interfaces, and so on — the underlying motivation for his work was to figure out how to augment collective intelligence. I’m pleased that this idea has become a central theme in today’s conversations about collaboration, community, collective impact, and tackling wicked problems.

However, I’m also troubled that many seem not to grasp the point that Doug made in his theological discussion. If a group is behaving collectively smarter than any individual, then it — by definition — is behaving in a way that is beyond any individual’s capability. If that’s the case, then traditional notions of command-and-control do not apply. The paradigm of really smart people thinking really hard, coming up with the “right” solution, then exerting control over other individuals in order to implement that solution is faulty.

Maximizing collective intelligence means giving up individual control. It also often means giving up on trying to understand why things work.

Ants are a great example of this. Anthills are a result of collective behavior, not the machination of some hyperintelligent ant.

In the early 1980s, a political scientist named Robert Axelrod organized a tournament, where he invited people to submit computer programs to play the Iterated Prisoner’s Dilemma, a twist on the classic game theory experiment, where the game is repeated over and over again by the same two prisoners.

In the original game, the prisoners will never see each other again, and so there is no cost to screwing over the other person. This changes in the Iterated Prisoner’s Dilemma, which means there’s now an incentive to cooperate. Axelrod was using the game as a way to try to understand the nature of cooperation more deeply.

As it turned out, one algorithm completely destroyed the competition at Axelrod’s tournament: Tit for Tat. Tit for Tat followed three basic rules:

  • Trust by default
  • The Golden Rule of reciprocity: Do unto others what they do unto you.
  • Forgive easily

Axelrod was intrigued by the simplicity of Tit for Tat and by how easily it had trounced its competition. He decided to organize a followup tournament, figuring that someone would figure out a way to improve on Tit for Tat. Even though everyone was gunning for the previous tournament’s winner, Tit for Tat again won handily. It was a clear example of how a set of simple rules could result in collectively intelligent behavior, highly resistant to the best individual efforts to understand and outsmart it.

There are lots of other great examples of this. Prediction markets consistently outperform punditry when it comes to forecasting everything from elections to finance. Nate Silver’s perfect forecasting of the 2012 presidential elections (not a prediction market, but similar in spirit) was the most recent example of this. Similarly, there have been several attempts to build a service that outperforms Wikipedia by “correcting” its flaws. All have invoked the approaches people took to try to beat Tit for Tat. All have failed.

The desires to understand and to control are fundamentally human. It’s not easy to rein those instincts in. Unfortunately, if we’re to figure out ways to maximize our collective intelligence, we must find that balance between doing what we do best and letting go. It’s very hard, but it’s necessary.

Remembering Doug today, I’m struck — as I often am — by how the solution to this dilemma may be found in his stories. While he was agnostic, he was still spiritual. Spirituality and faith are about believing in things we can’t know. Spirituality is a big part of what it means to be human. Maybe we need to embrace spirituality a little bit more in how we do our work.

Miss you, Doug.

Artwork by Amy Wu.