Emotionally Dealing With Cheaters

It’s time for some emotional maturity when it comes to agents engaged in systems.

Hawks and Doves

A few decades ago, a couple of scientists applied game theory to evolutionary development. The result was evolutionary game theory and it gave us some neat new ways to understand evolution.

The most famous example of this work is known as “hawks and doves”. The premise is that given a certain set of circumstances, individuals within the same species competing for finite resources may have more than one strategy for obtaining those resources. Divided simplistically, an individual may behave in a “dove”-like fashion ie standing down from conflict (after an initial bluff of force) and sharing resources with other doves it discovers, both of whom are non-violent or a “hawk”-like fashion ie following through on threats of violence and not sharing resources with others. The dove strategy has low costs but also low rewards while the other has high cost but also higher potential reward.

The theorists produced a graph that showed which strategy might be optimal under which conditions. There is an awful lot more to evolutionary game theory however I mention this as an illustrative example.

Depending on the conditions, you have a certain percentage of hawks and doves. The more hawks you have, the more costly it is to be one as you might starve or be killed. The more doves you have, the higher the reward for being a hawk (violent and not sharing) since almost everyone you meet is a dove and will back down. In the high dove scenario, it makes sense for more individuals to become hawks since the rewards are high and the risks very low. Once you have several hawks though, the chances of meeting another hawk are higher and it once again becomes too costly to be one compared to the expected reward. There is a certain balance that appears between the types of strategy. The exact ratios depend on the situation, but what I want to point out is that you always have some hawks.

Social Strategies/Cheaters

Evolutionary game theory goes on to analyse circumstances where individuals within a species co-operate to a certain extent, rather than compete.

Humans often co-operate, and this behaviour is said to be ‘social’. Humans are one of the most socially sophisticated animals on the planet.

In a highly social society there are systems of rules in place and a few different strategies for success. Social systems are typically based on varying degrees of trust, you need to trust that other people will follow the rules and humans have evolved extremely complex skills and heuristics to assess trustworthiness in others.

One strategy in trust based systems is to fake trustworthiness, not abide by the rules, screw people over and reap the benefits. This strategy risks being caught and completely shunned, which may even lead to death. It is a high-risk, high-reward strategy that is the social equivalent to a hawk.

It seems to be that any social system that has rules based on trust is also open to the possibility of cheating being a viable strategy. In any social game you always have some cheaters.

Stopping Cheaters

Our social evolution has given humans pretty powerful tools to spot a cheater, helping groups to keep cheaters to a minimum. Many of our formal systems also have safeguards to attempt to weed out cheaters.

However, I frequently come across the assumption that it would be ideal to completely eliminate cheaters. This is wrong.

Eliminating cheaters is not possible. All games of any complexity have rules and therefore can be cheated. The more dove-like rule-followers a game has, the greater the rewards are for cheating and so the likelihood of having cheaters in the game increases. I believe that past a certain point you have a power-law situation with cheaters, where the energy expended to detect and remove cheaters grows exponentially the lower the number of cheaters becomes. Eventually, the measures taken to eliminate the cheaters become more injurious for everyone than the harm the cheaters are causing.

A recent example from my life: at a Pride parade of tens of thousands of people a political group of a dozen participants was denied entry, but unofficially they ‘broke in’ at the end of the column and marched anyway. Despite the fact that the organisers had made reasonable efforts to prevent the group from marching that year the outrage was huge, so the following year the organisers implemented a security system requiring all 10,000 people to acquire official wristbands, the parade needed extra staff, security barriers and to change the assembly point and shorten the parade route. The measures far exceeded the harm.

Eliminating cheaters is not necessarily desirable. Cheaters have to develop great skill to cheat, skills that are often prized in general, from thorough attention to detail to ingenuity, innovation and improvisation. In David Chapman’s essay ‘Geeks, Mops and Sociopaths’ the sociopaths – the cheaters – have an important function. They use their skills to market the New Thing made by the Geeks, making money and enriching culture, even if they reap an unfair share of the rewards. Sociopaths in general are quite likely to be cheaters but also quite likely to be very useful, like surgeons.

Emotional maturity

The extreme measures taken to eliminate cheaters seem to be often caused by the emotional pain of being a rule-follower and seeing or knowing that there are cheaters. Cheaters seem to reap great rewards, while the costs of the strategy are less visible. The emotional reaction can blind people to other causes for cheating, such as injustice (which I think may have played a part in the Pride parade example).

This emotional overreaction goes for double when money is involved – the obvious example is rabidity over benefit/welfare cheaters. I think the extra effort expended on trying to prevent cheaters is one of the key reasons that universal benefit works out cheaper.

People also seem to overestimate how many cheaters there are. I would expect any system to be capable of supporting around 10% hawks/cheaters. However from internet reading I’ve done about crimes, false claims about crimes (eg false insurance claims, false rape claims) are around 2%. This seems absurdly low.

The emotional desire to eliminate cheaters is prioritised over the rational knowledge that cheaters are inevitable, exist in low numbers and are even desirable. We need to sort that out because so often the measures to prevent cheaters are worse than the cheaters themselves.