Tuesday, December 11, 2007

Tuesday Column: The prisoner's dilemma, in Bali

In almost every police interrogation movie there's this classic scene: The good cop walks into the cell housing one of the suspects and says: “We know the pair of you did it, your mate in the other cell has just confessed. Now why don't you fill us in on the details?”

Unable to confirm whether or not his or her partner really has spilled the beans the prisoner will be tempted to blab.

More than half a century ago researches at the RAND Corporation in the United States came up with a more complex version of the problem and named it the Prisoners Dilemma...

In it the police don't have enough evidence to nail either of the pair on the big charge that carries a ten-year jail term - only a lesser charge carrying six months.

They keep the prisoners apart and offer each an identical deal.

“I you betray your partner we will give you immunity and you won't spend an extra day more in jail. Your partner will get 10 years.”

“The only exception is if you each betray each other. Then you each get 5 years.

“So what's it going to be? Remember, if your partner blabs and you don't, you are facing a decade in jail.”

It is in the prisoners interests for neither to blab. Then each would be each out in months.

But if they each try to look after their own interests each will blab and be up for 5 years.

Economists have long been intrigued by the dilemma because it suggests that there's an exception to the more general truth that if everyone acts in their own interests society as a whole will be better off.

Ross Garnaut has been wrestling with it for months.

He's the economic guru appointed by Kevin Rudd and Australia's eight state governments in April to produce Australia's version of Britain's Stern Report on climate change.

He'll deliver his first progress report, to the Queensland Cabinet, today.

But a fortnight ago in the HC Coombs Lecture Theatre at the ANU he gave Canberra residents an early insight into his thinking.

He said that climate change was a “genuine international prisoners' dilemma”.

It is in most country's individual interest not to cut emissions. It would cost it money and by itself would do little to halt climate change.

But if every country acts in what it sees as its own individual interest, the end result - for all of them - will be catastrophic.

He said by contrast trade negotiations, often thought of as a prisoners dilemma, were not the real thing.

The truth is that opening a country up to imports almost always benefits it, even if many of the citizens of the country don't believe it.

It makes sense for nations such as Australia to cut tariffs even if others are not. Garnaut himself helped demonstrate that in the mid-1980's when he advised Bob Hawke to begin dismantling our tariff barriers without waiting for the rest of the world.

But climate change is different. It really wouldn't make narrow economic sense for Australia to cut its own carbon emissions if other nations didn't, a point made repeatedly by John Howard until he switched sides and set up his prime ministerial task force on emissions trading this time last year.

Climate change negotiations area all about getting everyone to put aside what appears to be their self interest in order to actually look after their interests – and Garnaut doesn't think it will be easy.

(I like to think of it not as a dilemma faced by prisoners, but as a dilemma about milk. My father told me this story when I was a boy. A village decides that on a special day its fountain will spout forth milk instead of water. All each family has to do is pour a cup of milk into the fountain overnight. Each family figures that it will save scarce money if it pours in a cup of water instead. No-one will notice and it won't make much difference. When the fountain is turned on in the morning, there is no milk, only water.)

But things aren't completely bleak. The surprising thing about the prisoner's dilemma is that when it takes place in the real world (actually in the world of the economics laboratory, where real people are invited to play a real version of it for real money against real people who they can't even see and will never see again) a very large proportion of people don't act as the economists say they will. They co-operate against what appears to be their narrow self-interest and find their trust repaid.

The other finding, very relevant to the Bali negotiations, is that where games involving the prisoners dilemma are repeated the co-operation rate jumps. It becomes in players' long-term self interest to develop a history of co-operating even if doesn't help them in the short-term.

And there's something else that will help. While it will be costly to a nation such as Australia to cut carbon emissions and while we will get little benefit unless the big emitters such as the United States and China do so as well) the best guess suggests that for Australia those costs are small.

The latest, produced by economic modellers from Monash University and the CSIRO looks at a range of options from cutting Australia's net emissions by 40 per cent to 100 per cent by the middle of the century.

It finds that whatever option is adopted by the middle of the century Australia's real incomes will be roughly three times as big as they are today.

Our grand children will be much better off than we are now whatever we do, just as we are much better off than our grand parents were.

But there is a caveat. Their analysis looks only at the costs of cutting carbon emissions - not at the potential benefits in terms of tragedies avoided if the entire world cuts them.

Those potential benefits are huge. They make Australia's decision look like not much of a dilemma at all.

3 comments:

swio said...

The other finding, very relevant to the Bali negotiations, is that where games involving the prisoners dilemma are repeated the co-operation rate jumps."

I'd forgotten about that little tid bit. This could be interpreted to suggest that by unilaterally acting first on committing to carbon reductions Australia could be jump starting a virturous circle of increasing co-operation on reducing carbon emissions. Perhaps early positive contributions have a disproportionately powerful effect on the final outcome?

Also relevant to Bali and carbon emissions are the results of game theory experiments where participants are allowed to punish those who "cheat" at some cost to themselves. I can't remember the reference but it was found that people were willing to pay a very high cost to punish cheaters, much higher than what would be rational (I think the rational amount was to never punish unless the cost was zero). This also suggests that if the world is able to come to an agreement on Bali but some participants fail to sign up (ie they cheat) then those that did sign up may be willing to absorb a surprisingly high cost to themselves (perhaps in the form of trade sanctions?) to "punish" the carbon "cheaters"

Peter said...

Here's a reference to work on a related topic - people's willingness to pay to punish winners:

http://pmhatingwinners.notlong.com

Anonymous said...

WHAT ECONOMISTS BELIEVE vs WHAT THEY SOMETIMES ASSUME

A nice article Peter, but I take issue with the way you represent economists in the following statement:

"The surprising thing about the prisoner's dilemma is that when it takes place in the real world ... a very large proportion of people don't act as the economists say they will."

Perhaps some economists think, or at least say, that people will only act to maximise their narrowly defined self-interest, but I think you'd find them in the minority. Most if not all my colleagues are aware of the existence, for instance, of interdependent utility functions and the scope for (seemingly) altruistic behaviour. What the experiments you referred to show is that people do not maximise narrow self-interest. Yet the assumption that people do maximise narrow self-interest in simply that - an assumption. It is also, as you would be aware, a very powerful and useful one, but it is one that in my experience is relaxed when necessary. Take a look at PC reports on gambling, foot and mouth disease and social capital, for instance. Climate change, it would seem to me, is another of those times.

Tom N.

Post a Comment

COMMENTS ARE CLOSED