Fallibility of intuition
The model opens up possibility that its not just average anger that matters but also distribution of radical elements.
The writer is a researcher and an adviser to the Dutch government on public policy. He can be reached on twitter @mrsultan713
Social scientists often lament the hubris of lay people and policymakers who blindly trust their intuitions and gut feelings, commonly dubbed as the ‘conventional wisdom’. Economists and increasingly political scientists have constructed disciplining devices such as models, to keep our logic grounded in the face of complexity. The prominent British mathematician George E.P Box had said, “all models are wrong, but some are useful”. This is a deep insight. In my last article, I outlined how models make us see hidden social patterns. Here I demonstrate how models act as ‘thinking aids’ and help us reach counter-intuitive conclusions. To make my point, I will use a simplified version of the Granovetter threshold model and apply it in the context of participation in revolutionary movements to show one of the many utilities of models.
Instead of using highfalutin jargon and presenting models as a black box of esoteric mathematical complexity, I will explain one simple model.
I begin with agents or players. Here players are prospective revolutionaries who want to overthrow a government. Now, assume that the people join the movement only if there are other people also willing to join the movement. Each prospective revolutionary has a ‘threshold’ to join the movement. Now, consider two societies, say Afghanistan and Pakistan, where people with different thresholds (to join the movement) live.
For simplification, assume that there are only four people in each country. In Afghanistan, each person has the same threshold, i.e., each person will join the movement if one other person joins the movement. In Pakistan, on the other hand, the first person joins the revolt without any other person joining the movement (his threshold is zero), the second person needs one, the third two and the fourth person needs three other people to join the revolutionary movement. In which society will more people join this movement?
Notice on average, people in Afghanistan are ‘angrier’, i.e., require fewer people from its populace to join the revolt relative to Pakistan. Our intuition would let us believe that as the people in Afghanistan are more prone to join the revolt (as their threshold is lower) so a greater number of people will turn out on the streets of Afghanistan. In fact, this is not necessarily true when we carefully think with our model.
Notice that no one in Afghanistan will join the revolt as one person is required for anyone (and everyone) to join the movement. On the other hand, Pakistan will witness a cascade effect. The first person will require no other person to join the movement. Of course, there are many factors that affect people’s decision-making. We can, of course, accommodate very complicated relationships in more advanced models. However, the result of the model holds in more ‘realistic’ settings.
The main point to note is if peer pressure to join the revolt is present we can reach counter-intuitive outcomes which are not easy to explain without the aid of models. The results of the model opens up the real possibility that it is not just the average anger in society that matters but also the distribution of radical elements in society. It shows how only a few extreme elements can precipitate a domino effect and eventually engulf a large section of society. It explains why most ‘pundits’ could not predict the Arab Spring, the Orange Revolution or the Iranian Revolution.
The main point I would like you to take from this article are the limits of explaining complex phenomena based on ‘gut-feeling reasoning’. Models, though not perfect, act as thinking aids and help us reach conclusions which we might not think of otherwise. Next time when you are making a blanket claim on the cause and effect of a particular phenomenon, do not forget to humble yourself by remembering the fallibility of intuition.
Published in The Express Tribune, February 22nd, 2014.
Instead of using highfalutin jargon and presenting models as a black box of esoteric mathematical complexity, I will explain one simple model.
I begin with agents or players. Here players are prospective revolutionaries who want to overthrow a government. Now, assume that the people join the movement only if there are other people also willing to join the movement. Each prospective revolutionary has a ‘threshold’ to join the movement. Now, consider two societies, say Afghanistan and Pakistan, where people with different thresholds (to join the movement) live.
For simplification, assume that there are only four people in each country. In Afghanistan, each person has the same threshold, i.e., each person will join the movement if one other person joins the movement. In Pakistan, on the other hand, the first person joins the revolt without any other person joining the movement (his threshold is zero), the second person needs one, the third two and the fourth person needs three other people to join the revolutionary movement. In which society will more people join this movement?
Notice on average, people in Afghanistan are ‘angrier’, i.e., require fewer people from its populace to join the revolt relative to Pakistan. Our intuition would let us believe that as the people in Afghanistan are more prone to join the revolt (as their threshold is lower) so a greater number of people will turn out on the streets of Afghanistan. In fact, this is not necessarily true when we carefully think with our model.
Notice that no one in Afghanistan will join the revolt as one person is required for anyone (and everyone) to join the movement. On the other hand, Pakistan will witness a cascade effect. The first person will require no other person to join the movement. Of course, there are many factors that affect people’s decision-making. We can, of course, accommodate very complicated relationships in more advanced models. However, the result of the model holds in more ‘realistic’ settings.
The main point to note is if peer pressure to join the revolt is present we can reach counter-intuitive outcomes which are not easy to explain without the aid of models. The results of the model opens up the real possibility that it is not just the average anger in society that matters but also the distribution of radical elements in society. It shows how only a few extreme elements can precipitate a domino effect and eventually engulf a large section of society. It explains why most ‘pundits’ could not predict the Arab Spring, the Orange Revolution or the Iranian Revolution.
The main point I would like you to take from this article are the limits of explaining complex phenomena based on ‘gut-feeling reasoning’. Models, though not perfect, act as thinking aids and help us reach conclusions which we might not think of otherwise. Next time when you are making a blanket claim on the cause and effect of a particular phenomenon, do not forget to humble yourself by remembering the fallibility of intuition.
Published in The Express Tribune, February 22nd, 2014.