Roblog

Imagine you had to meet someone in New York City. You couldn’t communicate with them in advance; you only knew that you had to meet them somewhere in the city at some point on a particular day. Where and when would you choose to go, in order to maximise the chances of meeting them?

The economist and game theorist Thomas Schelling posed this question to students. The most common answer given by them was to meet at 12pm by the information desk at Grand Central station. It’s not that 12pm is a particularly useful time to meet, or that Grand Central station is a particularly useful place – you could easily rattle off a list of a hundred other places that are equally suitable. But in trying to pick a point, you’re trying to guess what other people might pick. 12pm feels like a particularly salient time, and Grand Central station feels like a particularly salient place to meet people. It feels like what other people will pick, and so it’s what you should pick.

For an even simpler example, imagine that you’re tasked with choosing one of the four squares from this diagram. At the same time, in a different room, your friend is being given the same task, but you can’t communicate with them. If you pick the same square as your friend, you’ll both get a reward; if you each pick different squares, you’ll both get nothing.

Technically – if you were looking at this through a mathematician’s eyes – no square is superior to the others. You and your friend could pick any of them and win, or pick any of them and lose. And yet it seems incredibly obvious that you would pick the top-right, red square because it stands out. And, indeed, that’s what happens when this game is played for real. Most people pick the red square, and are correct to do so.

These focal points, points around which people can organise and coordinate even without communicating with one another, are called Schelling points after Thomas Schelling. They’re a great example of how human beings can coordinate their actions even without communication, as a result of the mental models they have about other people’s behaviour.

Schelling points are a useful tool for thinking about organisations. They help us to think about what people do, and where they go, in the absence of communication.

Imagine you’re working in a professional services business with lots of clients. You want to know what the latest developments are on a particular project. Where do you go? Clearly, you could solve this problem with communication. You’re not in the artificial environment of an academic experiment; you can just go and ask the person who’s leading the project, and they’ll give you an update. But that’s slow, and interrupts them, and means you need to wait on a reply. It also scales very poorly – what if that person is getting ten people asking them for updates every day? Coordination that requires communication is costly, and so it’s still worth exploring solutions that don’t involve communication.

If you have a system that everyone uses, and that provides an obvious and salient summary of projects, then that will become the place people working on projects will post their updates, and it will become the place people wanting to check on those projects will go to check on them. People will use it without having to communicate with one another. You will have created a Schelling point, a way for people to coordinate without communicating.

A big shared drive full of random documents tends to lack Schelling points; a more structured knowledge base is more likely to grow them. A clearly structured and easy-to-navigate website will develop popular Schelling points; a jumbled-up one won’t. Schelling points can be deliberately designed, can be consciously built into tools, and can be strengthened by the consistent use of those tools.

Ultimately, though, Schelling points are effective because of social norms. I know what’s salient to me, and I know what’s salient to you; we share an understanding of the world, and that allows us to converge on something without communication. Nobody designed Grand Central station as a Schelling point; it just is one, as a result of an unknowable emergent process within our collective psyche.

Schelling points, then, are enormously powerful. They allow for people to coordinate without communicating, even in enormously large organisations. They can be designed thoughtfully into processes to make them more effective.

But there’s a danger to them, too. Schelling points can be accidental as well as designed, and they’re also self-strengthening: the more people coordinate around a single point, the more plausible that point is as a choice, and so the more people will coordinate around it, and so on. Schelling points emerge whether we like it or not, and can prove stubbornly hard to displace. We are as much the prisoners of them as we are the designers of them; we’ll likely still be meeting people in Grand Central station for many years to come.