go to content

This Is What Happened When We Tested Game Theory On Ordinary People

Who's more likely to betray you: a Tory or a Labour voter? Find out below.

Posted on

Recently, a professor in a US university played a sneaky trick on his students.

WHAT KIND OF PROFESSOR DOES THIS

The students were asked to choose whether they wanted to improve their grade on their final assignment by two points or six points.

But there was a catch: If more than 10% of the class selected the six-point option, then no one got any bonus points at all.

This is related to two classic game theory thought experiments: the prisoner's dilemma, and the tragedy of the commons.

Both of which are a bit too complicated to explain here. You can learn about them here and here. (The prisoner's dilemma is perfectly encapsulated by the old game show Golden Balls, and if you've got seven minutes, it's worth watching this clip, because it's an amazing demonstration of game theory in action.) In short, though, they're about how, in certain situations, rational, self-interested decision-making can lead to outcomes which are worse for everyone.

That's how it is with the professor's sneaky offer.

If everyone cooperates, then everyone gets the two points, right? But if you know that everyone is going to cooperate, then it's in your best interest to betray them and to sneakily grab the six points.

Of course, everyone else is thinking the same thing. So everyone behaving in their own best interest leads to no one getting anything.

Apparently, the professor has done it every year since 2008 but only in one year did the students actually get the bonus points. Every other time, too many people betrayed the group.

So BuzzFeed and the polling company YouGov wondered: how would the Great British Public react in the same situation?

People who are polled by YouGov earn points, which are eventually translated into money. YouGov asked 1,766 people a tweaked version of the professor's question that offered additional points to those taking part – this mattered because it meant the people we were testing had the chance to win actual rewards.

This is what YouGov's users were asked:

Select whether you want an additional 25 or 100 points added to your account. But there's a small catch: If more than 10% of respondents for this poll select 100 points, then no one gets any additional points. Your responses will be anonymous.
It isn't definitive, of course, but it's one of the largest-scale applications of game theory ever carried out.

As the professor found, usually enough people choose to go for the self-interested option to ruin it for everyone. In our survey, 41% of people chose 100 points, meaning that no one got any points at all.

More interestingly, YouGov then broke down the data. For instance, older people are the least likely to choose the self-interested 100-point option.

YouGov

Which you can sort of understand, since they've already got all the houses and pensions. They don't need to steal stuff.

And this is the most interesting one. Apparently Conservative voters are the most likely to screw other people over and go for the 100 points.

YouGov

It's worth noting, though, that UKIP voters are second only to the Lib Dems in their team-player-ish-ness. Labour, the party of collective action, are nearly as self-interested as the Tories.

It's also worth noting that however you break it down – by politics, by age, by region, by gender – not one single group would have ended up with the money.

If you'd only asked Lib Dems, then you'd have come closest, with only 30% going for the 100-point self-interested option. But that's still three times higher than the cut-off.

A health warning. This is just one poll, and the more you break it down, the less reliable it becomes.

And polls are sometimes unreliable anyway. But the results do meet the standard of statistical significance: i.e. they're probably not due to chance. So until we know more, if you meet a Tory-voting London-based 25-to-39-year-old, keep your hand on your wallet.

Tom Chivers is a science writer for BuzzFeed and is based in London.

Contact Tom Chivers at tom.chivers@buzzfeed.com.

Got a confidential tip? Submit it here.