There might be no other concept that divides humanity as much as fairness. No, I’m not talking about disagreement on how to attain fairness, how to evaluate fairness or what fairness even is in the first place. I’m not talking about fairness related meta topics. I’m talking about the concept of fairness itself. I think that our desire for fairness might be the biggest flaw of humans. A flaw that is so fundamental, it might be a mathematical consequence of basic stochastics. It has to do with a certain effect I only realized this week and that happens in specific circumstances. Circumstances that were irrelevant for most of human history but that have become very common now. Let’s look at this effect and why it might be responsible for many of the most obnoxious debates we experience today.
The first assumption that needs to hold for this effect to work is that the thing people might perceive to be unjust is actually normally distributed. It doesn’t really matter what that thing is. It could be income, property, free time, blood pressure, or how often you get to ride the rollercoaster. I hope you can agree with me here that there are many situations where the measure people perceive as unjust is normally distributed. It’s not all the situations. If there is systemic bias towards certain groups of people, only a few factors dominate the product and you get a gaussian mixture with multiple peaks. That’s what one might call discrimination (like, e.g., racism), but I’m only talking at situations where enough variables are in play that the posterior distribution is normal.
The second assumption we have to make is that we are in a civilisation with a working democracy. This is the part which makes the effect relevant today where it hasn’t been for most of human history. First of all, for most of human history humans didn’t live in civilizations and second, even when they did, they mostly weren’t democracies. What I mean by democracy here is not really any concrete political system. What I mean is that the ratio of opinions about a certain variable within the population directly corresponds to the feasibility of policy decisions that change that variable. Or in other words: a smaller group of people is less likely to outvote a larger group of people on average. Of course there is lots of political theater all the time and tricks people play on each other, political coups, scandals and foul play. But we are looking at big, long term metrics here. Metrics that will not just change from one day to another from one policy change. And as long as both political camps have the same opportunity to use those tricks, their effect should average out.
Alright, so we are in a democratic civilization and argue about a normally distributed metric which some people perceive as deeply unjust. What that translates to is that people perceive the standard deviation as too high. When, e.g., comparing people from the lower quartile of the distribution with people from the upper quartile, this can be nicely emotionalized. But let’s look at the degrees of freedom we have. In most cases, we cannot shift the mean because that is tied to limited resources like funds, monthly revenue, time in a day, available land mass or the number of tickets we bought for the rollercoaster. Shifting the mean would require magically having more or less of it. We also cannot change the area under the graph, because that corresponds to the number of people (yes, I’m conflating a statistical measure with a probability density function here, but if you are a statistician and care about these things, don’t just complain but tell me how this is relevant to what I’m trying to describe here). So the only way in which policy could really improve the perceived unfairness is by decreasing the standard deviation.
And this is where you need to make the crucial observation: decreasing the standard deviation of a normal distribution without shifting the mean will benefit exactly half the population at the cost of the other half. The idea that more fairness benefits everyone is a fairy tale. While some still propagate the romantic robin-hood-esqu view of taking from the few rich and giving it to the many poor, that’s just not how it works in practice. If you are above average on that metric, you’re going to lose something. It’s the only way this works. This might seem surprising as, for example, there are ideas like taking money only form the ultra rich and using it to fund social benefits for the poor, but the money will trickle through many complex processes before we can finally look at the bank accounts of all people and will almost certainly notice that it is normally distributed and the mean hasn’t changed (safe for inflation). So if the policy worked perfectly, half the population will have less money than they had before.
Now, you might not see the problem. After all, this is exactly what we wanted. This matches a definition of fairness most people will agree with. Some have more, some have less, but there is less difference between what people have. But this is where my third assumption comes in: people are selfish. People like to use fairness as a convenient argument for policy decisions that benefit them but once they realize a policy doesn’t benefit them they try to find arguments against it. Naturally, not all people are like this; I know you have some great counterexamples, the best of which is yourself, but I’m pretty sure most people are selfish. The arguments people give for their opinions rarely have anything to do with their actual motivations because if they are selfishly motivated it is a bad look to have people notice that. Nearly everyone does it and it often happens unconsciously.
I hope you see where this is going by now and can appreciate the bitter-sweet beauty of this effect. Nature’s tendency to form normal distributions naturally divides humans into two camps: those who win from narrowing the standard deviation and those who lose. It doesn’t matter that I hate binary thinking and it doesn’t matter that it’s often a fallacy to do so, in this case the two camps just fall out of the maths. This might be the reason why politics these days tends to produce two estranged camps, and why important votes are often so incredibly close. Of course reality is more complicated than that, but I’d be surprised if this effect doesn’t at least play a role.
