Thursday, March 10, 2011

Why dictators should study Game Theory!

One of the (over)hyped features of the revolutions in the Arab world has been the role played by social media. Several articles have outlined how social media have made spreading information much faster, which has helped mobilize people to protest against the regimes. This is particularly important since these countries do not have an independent "regular" media.

However, overcoming the "coordination" problem is just one of the 2 problems that an individual faces in deciding to protest against the regime. For an individual to decide rationally that he can protest he needs to overcome 2 problems of information asymmetry

1. The "coordination" problem: Where and when the protests will be held
2. The "common knowledge for confidence" problem: For me, as a rational individual, to decide to protest, I need to know there are several other protesters. The reasoning is simple, if I am going to be one among only a handful of protesters then the dictatorship would crush the protest ruthlessly. If on the other hand I am among half a million protesters the dictatorship would be unable to do so.

Note that every individual faces the same issue. Hence what is required for a successful protest is that I know there are other protesters, and that they know the same. They should know, that I know, that they also want to protest and I should know that they know the same. Hence there should a "common knowledge" about the desire to protest for me to be able to have the confidence to protests myself. 

The role of social media is actually as important in overcoming the latter, as the former problem. When I post about the protest on Twitter or Facebook, and someone "Likes" or Retweets my message, we automatically have common knowledge.

The answer, if you are Mubarak, seems intuitive, shut down social media! However, with a little game theory you can show that this would actually be counterproductive! (the Game Theory blog "Cheap Talk" showed this well in their post However they have not talked about the former problem, and how shutting down social media would affect that )

The reasoning is as follows. The regime would not shut down social media unless their assessment was that the protests could be large-scale. But this is precisely the information that i as an individual protester seek!- whether the protests are going to be large-scale or not (and that everybody else is aware of the same!), because that is when it is rational for me to protest too! In other words, the regime has successfully consolidated all the private signals of protest, into one big public signal. Hence, paradoxically, the act of shutting down social media makes its role even more efficient in terms of overcoming the "common knowledge for confidence" problem- everyone instantly knows that everyone else is keen to protest and that they all know the same!

But, hang on, would it still not achieve the purpose of preventing the protesters from overcoming the coordination problem (the first problem outlined above)? Granted, they all now have the confidence to protest, but in the absence of social media, how would they know when and where to go? In which case, isn't shutting down social media still effective?

The answer is No, and it is due to a concept called "focal points" (which was first popularized by Malcolm Gladwell).

Consider this thought experiment: If I told you that you were supposed to meet someone in Paris the next day. I do not tell you the time or the place. The only thing that you know is that the other person has the same information. What would you do? Most people would decide to go to the Eiffel Tower at 12 noon. This is because when we have to solve coordination problems in the absence of information there are some "default" solutions that we inexorably converge towards, as we expect others to arrive at the same default solution.

Hence, when social media is shutdown, and I know there is no way to communicate with my legion of fellow protesters (I know it is a legion now by the regime's act of shutting down social media), I will protest in the famous focal point-  namely, Tahrir Square at 12 PM!

Note, that again the act of the dictatorship has been counterproductive- it ensures that every single protester converges upon the focal point, where it might have been that they would have been more dispersed otherwise!

S the act of shutting down social media helps the protesters solve both the coordination problem and the common knowledge for confidence problem, more efficiently, and optimally, than they otherwise would have!

But what should the government have done? Not doing anything would have still led the protesters to solve the problem using social media.

There were 2 options to consider. One is that given that the protesters would necessarily converge upon the focal point, to focus only on preventing people from getting there e.g., in blocking key arterial roads to Tahrir Square.

The other option is to use misinformation. Just as bad money drives good money out of circulation ( Gresham's law), bad information would drive good information out of circulation, if the Government had used social media to spread rumors about the size and nature of protests, and confused people with multiple and contradictory views of where protests would occur, that might have helped. this is however difficult to pull off since people typically automatically have a quality filter in social media networks (since you decide whom to follow or friend).

The collapse of these repressive regimes was inevitable, but by their ham handed actions driven by lack of knowledge of Game Theory, they hastened their own demise. Cheers to that!


Vishwas Narendra said...

Option 3 for the dictator: Cut off the internet even during "normal" times!

Ulysses said...

Agree! Drastic, but effective :)