While thinking about what it even means for a society to be ‘fair’, I came to define sub-terms that I think would be useful to use more widely. The inspiration comes from positive and negative liberty. As with the trend of ‘positive’ rights implying some sort of action or protection, I see ‘positive fairness’ as being an active protection against discrimination - for example via ‘positive action’. On the other side, ‘negative’ rights imply some sort of inaction or passivity, in this case I see ‘negative fairness’ as any mechanism that’s designed to discriminate ‘fairly’, e.g. by distributing rewards unequally according to either effort or needs. In other words: As with other social rights, negative fairness is the ability to discriminate, and positive fairness is the protection from discrimination. ¹

It seems to me that the trend in society is that both negative and positive fairness are desirable things, similar to how society generally wants both negative and positive rights of other varieties. We end up having to decide, almost on a case-by-case basis, which things should be on the ‘negative’ side of the spectrum, and which things should be on the ‘positive’ side of the spectrum. To use a simple example, we generally agree that the ‘freedom from being murdered’ is more societally useful than ‘freedom to murder’, which is why murder is illegal in practically every society. For more topical examples, we generally agree that preferential treatment on the basis of e.g. gender, heritage or religion should be protected against (e.g. anti-discriminatory laws), but we also generally agree that preferential treatment on the basis of differences in effort, ability, or choice, are justified.

So far so good, right? But the problem arises when we consider transitive causality. Let me demonstrate an illustrative example: In the absence of other information, would you consider it fair that somebody applying to a college or faculty with a lower admission rate (due to e.g. its higher popularity) has a lower acceptance rate than somebody applying to a faculty with a higher admission rate? I think the general answer is ‘yes, that’s fair’ - the argument goes that the choice of which faculty to apply to is one the person makes, therefore it’s not unfair for somebody who knowingly picks a more difficult college to get into to end up not being accepted. The question is this: How does this picture change if I add the information that the choice of college is highly associated, perhaps even causatively linked, with the applicant’s gender? Scenarios like this is exactly where the (in)famous Simpson’s paradox ends up illustrating the point perfectly: in the classic example, UC Berkeley’s individual departments are on average unbiased (or even slightly biased towards women), yet women still have a lower admission rate than men overall, due to the transitive causative link between the gender and the choice of department.

The question we want to be asking is this: To what extent are we trying to balance positive and negative fairness? Is it ethical to hire a worse applicant over a more qualified one simply on the basis of their gender? ² I think that to some extent, we consider this type of preferential treatment ‘fair’, when their qualification itself is causatively linked with gender. In this instance, we are sacrificing negative fairness (qualification –> reward) in order to promote positive fairness (gender -/-> reward). It becomes a question of policy, almost on a case-by-case basis; and moreover, it becomes a question of scope. Going back to the UC Berkeley example: Should we rate the fairness of UC Berkeley’s hiring process on a per-department level, or should we rate it overall? Should we evaluate the fairness of an economy on a per-sector basis, or should we evaluate it overall? Should a ‘fair’ hiring process completely obscure the applicant’s gender, nationality etc., even if this means the resulting admission rates will not at all be balanced with respect to the demographics of the society? I don’t think there are any obviously right answers to this, because shifting the focus one way or the other eliminates one type of fairness but necessarily introduces another type of unfairness - so it becomes a question of which type of unfairness we perceive as worse.

Moving away from the gender-based example, we can also apply the same concept to e.g. insurance and health care. The classic question being: Is it justified for somebody to have access to a higher standard of health care if they can afford it? But I think this question is seemingly less controversial: Is it justified for somebody to have access to more social benefits if they have a higher need for it? I think that in the latter case at least, the almost universal answer is going to be ‘yes’. But what if the degree of health care they require is due to self-inflicted factors? Do we consider it fair that somebody who lives a haphazard and unhealthy lifestyle has a higher risk of death than somebody who does not, or should society have an obligation to spend more effort protecting people from their own ignorance/stupidity/apathy? If we answer this question in a way that contradicts the previous question, we’re again left with a unresolvable conflict. Due to the causative link between an unhealthy lifestyle and a higher need for medical treatment, there’s no way to gain one type of fairness without sacrificing another type of fairness. Should insurance companies be allowed to charge more for individuals with a higher history of past accidents, especially accidents deemed to be within their control? If no, is it ‘fair’ that somebody who gets into fewer accidents has to pay more simply to cover the cost of somebody else’s daredevil antics and high-octane passions? Is it fair that ISPs have to charge the majority of their customers an excess price just to cover the costs of a few individuals abusing the flat-rate system to download orders of magnitude more than the average household? (And is it fair that somebody who earns more also has to pay more taxes? ³)

It seems to me as the most natural approach to try and resolve these types of conflicts is to ask the question of how much influence the individual we’re discriminating against had over their outcome. We generally agree that we should be responsible for factors within our control, while not being held responsible for factors outside our control. The reason people consider discrimination on the basis of gender or ethnicity to be undesirable is precisely because these are factors outside of our control - you don’t choose where or how you’re born. But what about discrimination on the basis of e.g. intelligence? For some reason, we (ironically) believe that people with a very low intelligence (e.g. clinical retardation) deserve extra social benefits, yet at the same time, lavishing people with a very high intelligence (i.e. an accomplished genius) with exceptional rewards and awards. It’s almost as if the wrong place to be is in the very middle. Somebody with an average intelligence has the worst of both worlds, in that they lack the natural ability to excel at problem solving, while at the same time being smart enough to have the expectation of solving those problems. In other words: for some reason, we assume that somebody of average intelligence is held responsible for their inability, while somebody of extremely low intelligence is not. And if we agree that it’s fair to reject an applicant on the basis of their low intelligence, is it fair to use somebody’s academic performance as an indicator for their intelligence? If ‘yes’, is it fair to use somebody’s religious beliefs as an indicator for intelligence? If ‘no’, what’s the fundamental difference there? Aren’t both transitively linked indicators for the actual causative metric we care about? As an even more fun example, what about discrimination on the basis of physical appearance? Is it fair that somebody born ugly has lower chances of forming a relationship? Is it fair that somebody with a naturally ‘goofier’ look makes a more successful comedian?

The problem is that I believe that it’s possible to form a transitive causative link between anything and factors outside your control. Maybe the only reason you ended up becoming a murderous psychopath was because of your genetics. Maybe the only reason you ended up drowning yourself in alcohol was because of your upbringing. Maybe the only reason you ended up applying to a more difficult department to get into was because of socioeconomic circumstances pushing you in that direction. How many degrees of ‘separation’ between the source (‘outside’ your control) and the target (‘within’ your control) of a bias do we demand in order for that transitive influence to be considered fair to ignore? How much ‘free will’ do we assign to people, and is it okay to discriminate against a group that ‘willingly’ self-marginalizes, e.g. incels or other separatist extremists? I think it’s pretty clear that trying to answer these questions with either ‘universally yes’ or ‘universally no’ gives rise to an equally insane world, with the societal consensus being somewhere in the middle ground.

The bottom line is that the ‘ideal society’ many of us seem to be fighting for is simply put, fundamentally impossible - and it becomes a question of what sacrifices we’re willing to make, in what contexts. Trying to focus too much on one relationship (e.g. gender -> pay) obscures the transitive causative links and biases coming from elsewhere, and there’s no “right” way to eliminate these types of biases. Put differently, there always will be biases and injustice - the question is, which type of bias do we prefer and why? And that’s the part where we can no longer apply logic, instead resigning ourselves to be led merely by gut feeling and chaotic social dynamics - against which there ultimately isn’t even an argument. It seems to come down to simply what people happen to feel at the time. Ironically making the most marginalized group in society those whose natural-born opinions and acquired beliefs happen to differ from the majority - and for whom, in a democracy, there can never be anti-discriminatory laws, because passing laws requires a majority opinion.