Bigotry is often tolerated among those that hold free speech and free association in high regard.

Yet stereotypes and ethnocentrism can make most everyone worse off.

In a previous article titled The Non-Aggression Strategy, I discussed how the TIT-FOR-TAT is the best strategy in an iterative prisoner’s dilemma in a wide variety of environments.  I labelled the TIT-FOR-TAT strategy the “Non-Aggression Strategy” for it’s similarity to the libertarian Non-Aggression Principle. When a majority of players in an iterative prisoner’s dilemma play TIT-FOR-TAT, then all players are better off because of the increased amount of cooperation.

When players of an iterative prisoner’s dilemma are categorized by arbitrary labels like blues and greens, and are allowed to determine their strategy based on their opponent’s label, it is a stable strategy for most players to be ethnocentric.  That is to play TIT-FOR-TAT or cooperate with players with the same label but always defect with players of different labels than theirs.

Now if a single player tries to break with the convention and cooperate or play TIT-FOR-TAT with the out-group they will be punished by trying to cooperate with defectors.

Since TIT-FOR-TAT is basically the non-aggression strategy, this means that the non-aggression strategy will fail in this environment. This is because,  players can play TIT-FOR-TAT with members of their own group, but playing TIT-FOR-TAT with members of other groups that will always defect, will reduce their score more than defecting against them.  In fact if all other opponents always defect against outsiders, there is no strategy that will do better than defecting with the members of other groups.  Ethnocentrism is incompatible with libertarianism, because it creates an environment that prevents non-aggression from being a successful strategy.

Also ethnocentrism makes almost everyone worse off. When players are hostile to those outside their group, all encounters between members of different groups reduce the scores of both players in the encounter.  This makes the total score of all players less due to the lack of mutual cooperation, between groups. This is similar to tragedy of the commons where everyone has an incentive to do things that make most people worse off. I call this the tragedy of stereotypes.

This means xenophobia and bigotry are a stable phenomena, and because it reduces the total accumulated scores of all players scores it harms almost everyone.  However if one group is larger than the other the group, they will not be harmed equally.  This is because the minority will have more negative interactions than the majority.

All that is needed is for ethnocentrism to be stable is for players to be identified by some arbitrary label so that all players can easily identify them, and for players to start determining strategy based on those labels. In this way arbitrary labels can become self-enforcing stereotypes.

Is important to understand that TIT-FOR-TAT will beat ethnocentrism if the ethnocentrism has not become the dominate strategy. And a population dominated by TIT-FOR TAT can’t be invaded by ethnocentrism.  It is only after most players are already using the ethnocentrism strategy that  an ethnocentric population won’t be invaded by TIT-FOR TAT.

Yet if TIT-FOR TAT can’t be invaded by ethnocentrism why does it seem ethnocentrism has been so dominant in humans? It could be that in the environment in hunter and gatherer times, when humans competed for limited resources, enable the emergence of ethnocentrism, referred to as parochialism in my post, Altruism and Racism Cause War, and War Selects for Altruism and Racism.

It is also possible that a pure ethnocentric strategy, that is always cooperate with in-group and always defect with out-group, is more robust in a noisy confusing environment.  That is, TIT-FOR TAT is susceptible to getting caught up in draining feuds if a player mistakenly defects, or mistakenly thinks a player has defected. There are modifications to TIT-FOR TAT that make it more robust to noise. Maybe ethnocentrism is more robust to noise that even these modified TIT-FOR TAT strategies.  This is not clear yet and needs further investigation.

Now since the labels do not have to have to be based on any objective differences to form these self-enforcing stereotypes, we should be careful in using labels to define ourselves and others. We should try to avoid falling into the US versus THEM strategy that is so natural for us, because it can cause self-enforcing stereotypes to emerge.  When they do emerge we can all lose opportunities due to the lack of cooperation that to these self-enforcing stereotypes engender.

In online forums in which flair are used to denote ideology, I often see people supporting weak arguments of those that have the same flair and downvoting strong arguments of those that have the flair of opposing the forum’s ideology. It is easy to fall into the US versus THEM strategy.  Sometimes labels might hinder the discussion more than help.

What does this all mean?  Well even though racism and other stereotypes do not violate the Non-Aggression Principle, as long as people do not physically trespass against others or commit fraud, these stereotypes might engender a culture that is more tolerant to initiations of the violence against the out group.  This could eventually lead to self-enforcing stereotypes and an environment in which the Non-Aggression Strategy fails.  Libertarians should not tolerate racism, sexism and other forms of bigotry, because it could engender a culture that would create an environment in which the Non-Aggression is not a viable strategy and everyone is worse off.

It may seem very pessimistic that the most successful strategy (the Non-Aggression Strategy), is unable to defeat self enforcing stereotypes once they are established, even though they make everyone worse off.

The good news is that it is possible, the original conditions that selected for racism no longer exist.

See Altruism and Racism Cause War, and War Selects for Altruism and Racism.  Altruism towards in-group and antagonism towards the out-group was selected for when groups were in zero sum competition for resources. Wars are now less frequent with less fatalities due to the non-zero sum environment that came with the advent of agriculture and then further magnified by markets, so there is now less selection for this behaviour.

Free markets change the payoffs such that there is much more incentive to cooperate. The gains from the division of labour through markets make cooperation much more beneficial.  Gary S. Becker won a Nobel Prize in part for showing how markets discourage discrimination based on race rather than merit. So a society with a free market might be more able to avoid self-enforcing stereotypes.  The challenge is that the market can’t eliminate all prejudice. Even if prejudice does not result in violence, less cooperation due to prejudice, means less productivity for all.

So, in addition to market incentives we may need cultural suppression of prejudice.  As Mises said:

Liberalism, however, must be intolerant of every kind of intolerance. If one considers the peaceful cooperation of all men as the goal of social evolution, one cannot permit the peace to be disturbed by priests and fanatics. Liberalism proclaims tolerance for every religious faith and every metaphysical belief, not out of indifference for these “higher” things, but from the conviction that the assurance of peace within society must take precedence over everything and everyone.

-Ludwig von Mises


The Evolution of Cooperation, Chapter 8, The Social Structure of Cooperation, in the section on Labels, Stereotypes, and Status Hierarchies, by Robert Axelrod

How to cope with noise in the iterated prisoner’s dilemma by J Wu, R Axelrod

The Evolution of Ethnocentrism by Ross A. Hammond and Robert Axelrod

Liberalism by Ludwig von Mises