Group decision making

It’s said that during World War II, when journalists asked American soldiers why they went to war, the common answer was “for Mom and apple pie”. Whether true or not, the idea is that “Mom and apple pie” are universal American values that nobody can argue against. Common refrains like this are useful because they let you off the hook of answering the actual question. They’re like a conversational get-out-of-jail-free card.

I think the corporate version of Mom and apple pie is the appeal to group decision making. Specifically, when asked to make a decision, people find it easier to say “let’s get more input on this question”, or “why don’t we make sure that so and so weighs in”. After all, “two heads are better than one”.

I’ve always had an aversion to hearing this. I suspect that in most cases, appeals to add more people to a decision are just a convenient dodge, but I’ve never been able to articulate a good argument against it. I think it’s because in the moment, it’s hard to argue against adding a specific person to a group without sounding like an asshole. “I don’t think Phil has much value to add here” may be true, but it’s not a winning argument. So instead, here’s a general theory on why adding more people to a group decision is probably a bad idea.

As you add more people to a group, two things happen. First, the total number of people in the group increases. Second, the total amount of “thoughts” in the group also increases. As more people join a group, the amount of group thoughts increases correspondingly.

Here we see the seductive appeal of adding more people to the group. People look at this graph and expect that because both functions progress linearly, there’s at least no harm in adding more people, because as the group grows, so too does the aggregate amount of “thinking” in it. In cultures that prize risk mitigation and penalize members for mistakes, this is a function to be maximized. “Nobody ever got fired for thinking things over one more time”.

The problem is that while group size and thought progress linearly, group complexity and value progress exponentially and inversely to each other. Here’s what that looks like in a graph:

As you add more people to the group, the complexity of the group increases in proportion to the square of the group size. This is because each group member needs to coordinate with each other to schedule times to meet, hear each other out, and reconcile their opinions. The amount of duplication also increases as more and more members rehash largely similar issues over and over again. Meanwhile, the probability that each incremental thought added to the group is novel and valuable decreases in inverse proportion to the aggregate amount of thought previously incorporated into the group. There are only so many bases to cover.

The moral of the story is that we sacrifice more than we gain in adding more than a handful of people to a group tasked with making a decision. So what to do instead? I propose the following:

  • Keep groups small by restrict decision making authority to a handful of people. 3 seems optimal. If this feels too difficult, break the problem down into smaller ones.
  • Gather input asynchronously and on a voluntary basis – e.g., have people comment or write in on a public document instead of setting up synchronous meetings.
  • Judge a decision based on process, not outcome, to eliminate the incentive to pass the buck on taking action.
  • Make peace with not knowing everything. Colin Powell said that he takes decisions when he has between 40-70% of the requisite information. Any less and you’re shooting from the hip, but any more and you’ve waited too long. It’s unlikely that many of us will face decisions with higher stakes than Colin Powell did. If 70% of the available information was good enough for him, it’s probably good enough for you too.