Sunday June 04, 2017

Dumb bots might help you get along with your colleagues

The game view in Nicholas Christakis's lab experiment.

The game view in Nicholas Christakis's lab experiment. (Hirokazu Shirado/Yale University)

Listen 9:16

Every week, the Spark gang works together to make the show. And your work is probably similar. You've got to collaborate with other people to solve problems, and think creatively.

For most of us, that means clearly seeing the issue in front of us, and filtering out anything else. We want to see the signal, and avoid the noise.

But what if that's wrong?

Nicholas Christakis is the Sol Goldman Professor of Social and Natural Science at Yale University.


Nicholas Christakis.

He's been wondering whether a little noise - and a little nonsense - could actually be a good thing.

When you go to a physical library, you typically look up a book in a card catalogue, Nicholas offers as an example.  And then when you find that book in the stacks, you might notice that the book sitting next to it on the shelf is actually better. That's noise, Nicholas explains.

And it's not just simple things like perusing a library; "noise" affects the evolution of an entire species: Without noise like gene mutation, animals and plants would never manage to adapt to their changing surroundings.

But too much noise, he adds, is also bad. "You get a mutant that can't survive."

Because we tend to be so focused on eliminating that sort of noise in our own actions, collectively, we can end up getting stuck -- like in traffic or unable to make a decision for the collective good, because we can't see beyond our own efforts to succeed: the tragedy of the commons.

So Nicholas and his colleagues devised a simple experiment, where people, working together online, placed one of three coloured lights on a network. There was only one rule - there couldn't be any identical colours next to each other. But the individual players could only see their immediate area, and not the whole network. And guess what… although they may have made things work in their proximity, they failed as a collective.

And the players, who all seem to have done their jobs, start blaming each other.  "Everyone thinks, oh it's some other jerk… and the network gets stuck."

So Nicholas wondered whether adding some simple bots to the process would help.


Bots introducing "noise" to the colour patterning experiment. (Hirokazu Shirado/Yale University)

And what they discovered was when the bots periodically added colours at random, and didn't pay attention to what colour they were next to, the humans in the experiment were forced to think more collectively and creatively about solving the issue. In other words, adding some "noise" actually spurred people to solve the problem more efficiently.

"And they end up with cascading successes, instead of cascading failures."

So how do we introduce that noise in a safe way into everyday life and work, and get groups to make better collective decisions?

Nicholas concedes that the study, published in the journal Nature, was attempting prove a point, at least for now. But he says the experiment suggests that if we could learn to tolerate a little more dissent, or even nonsense, in our workspaces, we might achieve better and more creative results.