The Convergence Trap

I was sitting in a room with a bunch of people debating a topic. The atmosphere was fantastic. Ideas were popping up flawlessly, everyone quickly aligned on a direction. You could feel in your skin that the meeting was productive. Maybe the most productive one of the week.

Until one of my colleagues unmuted himself and said we were solving the wrong problem – the one that didn’t really matter.

He was right.

This made me wonder what his superpower was about. We had seemingly diverse backgrounds, opinions, and seniority. Even though, we felt short in direct comparison. I could acknowledge it was about a skill gap, but the difference shouldn’t be exponential in the outcome produced; it should be linear at most. Yet problems seemed to avoid him, no matter how complex the topic he was covering, and he always delivered more than was expected.

If you’ve known me for a while, you probably noticed that I’m an advocate of collaborative thinking. This is superior for solving problems, learning from each other, and building trust. Once mastered, it provides significantly better results than solitary thinking in the discovery phase. But it comes with a price.

While thinking alone has no immediate effect on others, group thinking is all about dynamics, relations, and hierarchy, which are prone to being influenced in unpredictable ways. When we think alone, our mistakes remain contained within our own reasoning. When we think together, our thinking becomes a system where systems theory applies. Ideas cause other ideas, people influence other people, and even the order of thoughts makes a difference. A wrong assumption made by one person can become the starting point for the next person’s reasoning, which then becomes evidence for someone else. Like a domino effect. The blast radius isn’t, in that case, one person’s thinking. It is the thinking of everyone who comes after them.

One of the reasons this happens is that the timing of ideas really matters. The first convincing explanation has an advantage over the latter because everyone else is in a position to respond. Someone strengthens the thought by providing evidence. Another person proposes an improvement. Someone else connects it to a past event. The conversation itself becomes richer, but is limited to the chosen direction. This is closely related to research on anchoring and biased hypothesis testing. People tend to evaluate evidence in ways that support an existing position rather than considering opposite interpretations. The first hypothesis can become an anchor before the group has consciously decided that it should be one. Once it is shared, we start evaluating everything that follows in relation to what has already been said.

The social structure of the conversation might influence the thinking even before an idea is evaluated. It’s related to production blocking, where people have to wait while someone else is speaking, maintain their own thoughts in unreliable working memory, and keep track of the conversation at the same time. These interruptions can reduce the amount and flexibility of ideas generated by the group. When we imagine a meeting consisting of seven people bringing seven independent perspectives into it, it’s actually an imagination that doesn’t happen in reality. Some ideas never get expressed because the conversation has already moved on. Or someone was afraid of the validity of their point. Or they felt they were not in charge because of their social position. Therefore, the group is not only evaluating a set of ideas. It’s influencing which ideas become available for evaluation in the first place.

This especially happens in ambiguous environments, where people relate to their experience and patterns that worked before. Let’s imagine an engineering team investigating a production incident. Someone suggests it’s a database overload. It sounds like a reasonable hypothesis, and it comes from a person who’s right most of the time. So the team starts looking at queries, connections, and CPU. Someone connects it to a similar incident from last year. Another person proposes increasing capacity. Someone else suggests an architectural change that would solve it strategically. After two hours, the team has produced a sophisticated plan for dealing with the problem. Finally, it turns out it’s not a database problem at all.

The first hypothesis set the tone for the whole conversation. It sounded plausible, which turned off the critical thinking. This is why premature convergence is more subtle than simply choosing the wrong solution. We often think of convergence as something that happens at the end of the thinking process, after we have explored the possibilities and evaluated them carefully. In reality, it can happen much earlier, sometimes before the group has even agreed on what question it’s trying to answer.

What’s more, the most dangerous type of it can happen between smart people doing exactly what organisations expect good teams to do: listen carefully, build on each other’s ideas, look for evidence, create alignment, and eventually move towards a decision. That’s why the problem is so difficult to notice. A meeting that has much friction in the form of disagreement feels unproductive, while a meeting in which everyone quickly develops the same idea feels like strong collaboration. But the feeling of progress can sometimes come from the fact that the group has stopped exploring, rather than from the fact that it has found the right direction.

Looking back at that meeting, I think this was what I had been seeing in my colleague without having the right language to describe it. He wasn’t better at solving the problem, nor was he much more informed than us. He was better at questioning the problem we had started solving. While the rest of us were making our explanation increasingly sophisticated, he was still checking whether the explanation itself should be discussed. He was applying something close to first-principles thinking to really understand what the problem was before trying to solve it. That distinction matters because expertise often makes convergence easier. The more experience we have, the more patterns we recognise, the faster we can classify a situation and move towards a familiar solution.

Organisations are great at creating mechanisms for convergence. We have meetings, roadmaps, frameworks, decision documents and alignment sessions – all designed to turn many individual perspectives into coordinated, mutual action. We need those mechanisms. The problem is that we have far fewer for protecting divergence before convergence starts. We often ask people to challenge each other, but we don’t always create the environment in which genuinely independent hypotheses can survive long enough to be compared. Asking someone to challenge the dominant idea after twelve people have already built on it is very different from giving them the opportunity to develop a different explanation before the group has settled on one.

Research on the consider-the-opposite strategy points to a possible solution to this problem: deliberately generating an alternative explanation can help counter the tendency to protect the explanation we already have. In practice, that might be as simple as asking what we would be discussing if the first idea had never been proposed, what we are no longer considering, or what evidence would make our current explanation wrong.

I still believe that collaborative thinking is one of the most powerful ways of solving complex problems. But perhaps we need to become more conscious of the moment when collaboration starts turning into convergence. The goal is to keep the benefits of collaboration without losing the space for independent thinking. It is to make sure that we don’t confuse the feeling of alignment with the quality of our thinking.

Maybe this was my colleague’s superpower. He wasn’t necessarily better at solving problems. He was better at noticing when we were solving one that didn’t really matter.

1. Hypothesis testing and biased assimilation (Lord, C.G., Lepper, M.R. & Preston, E., 1984): showed that people tend to evaluate evidence in ways that support their existing beliefs, while deliberately considering the opposite interpretation can reduce this bias. The study is particularly relevant here because it shows how the initial position can shape how subsequent evidence is interpreted.
2. Production blocking (Nijstad, B.A., Stroebe, W. & Lodewijkx, H.F.M., 2003): describes how having to wait for others to speak can interfere with idea generation. In group discussions, people have to remember their own ideas while listening to others, and some ideas are lost before they can be expressed. The effect can reduce both the number and flexibility of ideas generated.
3. Consider-the-opposite strategy (Hirt, E.R. & Markman, K.D., 1995): found that deliberately generating alternative explanations can reduce biased judgments. Instead of evaluating only the explanation that is already available, considering plausible alternatives encourages people to search beyond the initial interpretation.