Binary Labels Reinforce Systemic Discrimination


S.M. Amadae is a lecturer in politics at the University of Helsinki. She is also a 2019-20 Berggruen Institute fellow at the Center for the Advanced Study of the Behavioral Sciences at Stanford University.

While the United States is mired in pandemic and cultural conflict, in Finland the big political debate is over odd and even numbers. Finnish state identity numbers specify that if the last digit is even, that person is female. If the final digit is odd, the individual is male. Recently, the government decided that this identity system supports implicit bias and discrimination, and plans to overhaul it with gender-neutral numbers.

However, a petition against overhauling the system has more than 50,000 signatories. People signing the petition argue that gendered identity numbers are harmless. Why spend billions of dollars to reconstruct the administration of these tax and social security identifiers?

There are reasons to believe that the Finnish (indeed any) binary marking system used to classify individuals is far from innocent. Yet the underlying foundation for such a system is not widely understood.

This foundation can be seen in historical segregational systems like the one-drop rule in the U.S., which labeled any person with a Black ancestor as Black, or the Belgian legal division of Hutus and Tutsis into distinct groupings in colonial East Africa. Even the culturally implicit classification of people as white and nonwhite, or male and female, can serve as the basis of culturally accepted, although unequal, norms. These norms may not be enforced by the rule of law, but they enable the social dominance of members of one group over members of the other.

In the 1970s, the economist Thomas Schelling, who would go on to win the Nobel Prize in 2005, pioneered a model of how discriminatory patterns of segregation arise. He called it the checkboard model. It demonstrates that if two different groups of people live in an integrated neighborhood, it only takes slight bias for segregation to result. As Schelling demonstrated, if individuals seek to have at least half their neighbors like themselves, a segregated neighborhood soon takes shape. Segregation, Schelling showed, arises from individual decisions with partial prejudice. There doesn’t have to be a collusive scheme to expel unlike people from one’s streets.

Another model that illustrates how systemic oppression operates is called the Hawk-Dove game. It is also known as chicken. Say two cars face off at opposite ends of a one-lane road. When signaled, the two drivers accelerate toward each other. Winning the game means staying on course, with the other driver swerving to avoid a collision. In a draw, both drivers swerve. Or neither swerves, and there is a deadly collision.

Hawk-Dove was first used to model nuclear brinkmanship: If two nuclear-armed nations are heading for conflict, which one might blink first? However, another game was more closely studied during the Cold War: Prisoner’s dilemma. Prisoner’s dilemma describes a situation in which two criminals are arrested, the police don’t have enough evidence to convict either one on a serious charge, but do have enough to charge either with a less-serious crime, and so offer each prisoner an opportunity to betray the other in exchange for leniency.

Prisoner’s dilemma has been applied to many areas of game theory, like nuclear arms races, collective action, distributing public goods, free riding and bargaining. It also underlies the biologist Richard Dawkins’ selfish gene theory, in which an individual’s actions are geared toward self-preservation at the genetic level.

Prisoner’s dilemma assumes the equal power of actors. It does not result in systematic inequity or relations of dominance. Hawk-Dove, on the other hand, is a powerful tool to reveal the mechanism by which systemic discrimination arises when an arbitrary binary marker is introduced into a population.

Throughout the Cold War, Prisoner’s dilemma was regarded as the toughest situation from which to achieve cooperation. In this game, all actors always prefer conflict over cooperation. A situation resembling Hawk-Dove, in which an actor — say the U.S. — prefers to unliterally submit rather than to fight was regarded as inconceivable. Choosing to submit would expose a defeatist attitude — “better red than dead.” Thus, the U.S. nuclear posture developed to demonstrate to all prospective adversaries that Americans have both the intention and the capability to prevail in any kind of warfare, even protracted nuclear war.

In comparison to Prisoner’s dilemma, Hawk-Dove has not received much attention for being informative about any important class of social dilemmas or conflicts. But it has two features that make it well-suited for understanding systemic discrimination.

First, Hawk-Dove reflects a situation in which backing down from conflict may be more reasonable than engaging in hostility. Hostility is costly. This is specifically the case, as in the example of two cars racing toward each other on a single-lane road, when individuals risk significant harm by engaging in conflict.

In a Hawk-Dove encounter applied to social conflict, actors confront each other over a resource, such as food, shelter or status. If both concede, the resource is divided equally. If one concedes and the other attacks, the attacker claims the resource. If both attack, the resource is lost. The degree of possible injury remains constant within each game. Between games, conflict is variable and can range from verbal insult to bodily harm, even death.

Hawk-Dove reflects a situation in which backing down from conflict may be more reasonable than engaging in hostility. Hostility is costly.

The second important aspect of Hawk-Dove pertains to the impact of introducing a binary tag. If individuals form a homogenous population, they cannot develop relations of dominance. As in Schelling’s checkboard model, without any defining characteristic, individuals are indistinguishable. They must treat each other equally. However, once a binary tag is assigned, over a period of repeating interactions between individuals, members of one group will come to systematically dominate members of the other group.

Consider, for example, a population of a hundred individuals who repeatedly meet each other in randomly paired encounters over many interactions. Without any defining marker, for each individual to maximize personal gain, they must play a randomized pattern. They sometimes submit (playing dove), and sometimes they are aggressive (playing hawk). Just like in the children’s game rock, paper, scissors, no individual can win in the long run. All must randomly change strategies. Otherwise, their opponent can predict their move, and thus win.

However, in a sample population of a hundred people, if some individuals are marked as blue and the rest as red, over successive interactions, either blue or red will come to dominate. Blue players will always play hawk when they encounter red players, who always play dove, or vice versa.

Once a convention becomes established, it is impossible for any single individual, or small group of individuals, to alter this pattern. Individuals could resist this population-wide norm, such as a red player who grows tired of submitting and resolves to always challenge blue players by henceforth playing hawk. However, they will unfailingly be harmed by facing endless new attackers.

I developed this Hawk-Dove binary (HDB) model as a logical analysis. It can be simulated with a computer programmed to represent a population of individuals who seek to maximize personal gain under the conditions of being tagged as members of one or another group. These simulations reinforce the logical outcome of the HDB model. They demonstrate the mechanism by which the binary marker necessarily results in the systematic domination of members of one group over members of the other group.

Experimental evidence further corroborates this result. Students who played this game against each other virtually always ended up with either red-tagged or blue-tagged players dominating within less than 16 rounds of a 32-round game.

In 1968, Jane Elliott observed how, if individuals are divided into two groups — by eye color, for example — one group rapidly grows accustomed to dominating the other. Elliot suggested to her students that brown-eyed people are smarter and more capable than blue-eyed people. The markers are wholly arbitrary. Nevertheless, the members of the advantaged group came to justify their privilege by developing beliefs that they were superior. Winners developed a narrative to explain to themselves why they prevailed over members of the other group, and they saw the others as inherently inferior. Losers developed low self-esteem and internalized their lower status as their own fault.

In a homogeneous population, actors are not individually recognizable and cannot treat one another in any coordinated manner. Hierarchical order cannot be established. However, introducing a binary tag makes it possible to treat others with discrimination. The markers enable a systematic pattern of hierarchy to emerge, without any additional assumptions that some actors are more capable or more aggressive than others. Hierarchy reduces chaos. We are able to predict with virtual certainty, once we see which group prevails over the other, what role every individual will play in every interaction.

Given that, prior to the marker being applied to individuals, it is strictly a matter of chance what any given player will do in any interaction, this reduction of entropy — this resistance to devolving into chaos — is consequential. It allows the population as a whole to make efficient use of available resources. This is because, in the homogenous population, with everyone interacting randomly, conflict is normal. Every time two actors both play hawk, precious resources are lost.

In populations with a binary tag, on the other hand, because one individual always submits, no resources are lost in conflict. Thus we cannot dismiss the evolutionary role the HDB mechanism may have played in advancing civilizations by allowing hierarchy, order and efficient — although inequitable — resource use.

There is still much to be learned from HDB. Given the accidental nature of the binary tag, there is no apparent feature that would lead to members of one group dominating over others, rather than the inverse. It does not obviously explain why minorities or majorities will dominate.

How then can we explain systemic patterns which seem to hold across different cultures? For example, why is it the case that males typically dominate females in most societies around the world? Why is lighter skin often advantaged over darker skin?

Without any intended bias, self-interest combines with the obvious means to treat people differently based on their identities, resulting in a global pattern of subjugation.

With respect to male domination over females, Cailin O’Conner, who is one of the few to have studied Hawk-Dove in detail, argues that a socially constructed bundling occurs. Human identity as male or female builds on biologically fixed traits, such as physical strength or pregnancy. Identity markers incorporate stereotypical behavioral roles that cannot be causally correlated with these physical traits. Males may be more effective hunters in some settings, and females give birth. However, causation due to these physical differences does not account for the fact that, in many societies, females are much more likely responsible for household chores and receive less pay for equal work.

O’Connor argues that if actors with a binary tag have some bargaining advantage in one setting, such as sharing food after a big game hunt, then this bargaining power can be transferred to other routine encounters with out-group members. Thus, some physically based advantage in one area of life can be translated across all areas correlated to the identity marker. This results in extreme role divisions. One example of such an absolute case is the doctrinal rules in traditional Asian Buddhism. Nuns must defer to monks in all matters, and may never criticize them. Here, the most senior female categorically has less status than the lowest ranking male novice monk, who may be a child.

Now we can understand the stakes of enforcing binary markers on population members in Finland, with its gendered identity documents. HDB shows us how systemic discrimination will emerge in any population whose members are characterized by dualistic tags such as male and female, or white and nonwhite.

As in Schelling’s checkboard model, it doesn’t have to be true that individuals are biased — although they may be. Rather, the model shows that without any intended bias, self-interest combines with the obvious means to treat people differently based on their identities, resulting in a global pattern of subjugation.

This hierarchy is efficient, but it is unfair. Countering it requires raising consciousness, challenging the recognition of binary identity markers as relevant to individuals’ treatment and accepting the legitimacy of oppressed groups to protest in solidarity.