A simple model for why moderate beliefs rarely prevail.

I attend the weekly chaos and complexity seminar at the University of Wisconsin when I am in Madison WI during the summer. I want to pass on the paper by Marvel et al. in Physical Review Letters that was discussed this past week, as well a few clips from a review by Zyga discussing its purely mathematical "model of ideological revolution.":

We live in a world of extremes, where being fervently for or against an issue often becomes the dominant social ideology – until an opposing belief that is equally extreme emerges to challenge the first one, eventually becoming the new social paradigm. And so the cycle repeats, with one ideological extreme replacing another, and neither delivering a sustainable solution. Political revolutions, economic bubbles, booms and busts in consumer confidence, and short-lived reforms such as Prohibition in the US all follow this kind of cycle. Why, researchers want to know, does a majority of the population not settle on an intermediate position that blends the best of the old and new?

The model of ideological revolution begins with a community consisting of four types of individuals: those that currently hold an extreme opinion A, those that hold the opposing extreme opinion B, those that hold neither A nor B (the moderates), and those that hold A indefinitely and never change their minds (the A zealots).

To run the model, two individuals are randomly selected to interact with each other, with one randomly chosen to be the speaker and the other the listener. If the speaker is an A or B and the listener is a B or A, respectively, the speaker changes the listener's beliefs to AB. If the listener is an AB, then the listener becomes an A if the speaker is an A, and becomes a B if the speaker is a B. Moderate speakers cannot change a listener's beliefs; only extremists rally others toward their cause.

Running this basic model, the researchers found that the proportion of zealots strongly affects the outcome. When zealots are below a critical value, the system remains similar to how it started. But above a critical value, the zealots quickly convert the entire population to A...Marvel noted that "a raft of alternatives to our basic model (built from different assumed interactions) all show the same threshold behavior: when the committed believers reach a certain fraction of the community, they are capable of converting everyone to their perspective," Marvel said. "This suggests that a similar threshold may appear in real systems even when those real systems have dynamics somewhat different from our basic model. As the American anthropologist Margaret Mead is claimed to have said, 'Never doubt that a small group of thoughtful, committed citizens can change the world. Indeed, it is the only thing that ever has.'"

The researchers tested seven different strategies for increasing the moderate subpopulation in the model...only one could effectively expand the moderate subpopulation – and the strategy was based not on social interaction but on other environmental stimuli, which might take the form of a media campaign in real life. By integrating this new parameter into the model, the number of moderates increased without threat of extinction.

"The one successful strategy, nonsocial deradicalization, involves a particularly strong sort of encouragement of moderation; for example, its terms with the new parameter are independent of the size of the moderate population," Marvel said. "Hence, our findings suggest that this strong form of encouragement may be necessary for spreading a balanced perspective in a sustainable way."

The researchers note that this strategy should be regarded with caution, given that they have not attempted to show that the model's dynamics accurately represent the real world, with its multiple small-scale ideologies, fragmentation of opinions, and other intricacies. Nevertheless, they hope that this general framework for testing possible strategies that encourage moderation may lead to the discovery of more sophisticated methods.