In Defense of Meritocracy (And how AI might help?)
Meritocracy has become a little bit of a dirty word recently. In contemporary politics it is used to justify wealth and income inequality, and also as a justification against social justice initiatives and ideas. While the original term did indeed start off within a socio-political context (it was coined in the mid 20th century despite what one might think), today the word has a broader meaning outside of politics and we all know it. If you google the definition for meritocracy you will get the broader definition which is a society or system that rewards individuals based on their own merit (skills, effort, outcomes) rather than family, class, and wealth background.
To hone in better though, I think it's more helpful to think of the term meritocratic rather than meritocracy. When something is meritocratic that means that system or institution is trying to practice meritocracy, relatively speaking. It's worth pausing for a moment and asking then, what even is the opposite of "meritocratic"? What do we classify the lack of meritocracy as? Nepotism and in the case of where public institutions and and governments are involved: corruption.
And like everything else in the social science world, these exist on a continuum rather than as binary states.
Our World isn't Particularly Meritocratic
I didn't need to tell you that, but we do not live in a particularly meritocratic world. There are some domain exceptions to this which we'll get to, but by and large our society isn't meritocratic.
In a perfectly meritocratic society, you would statistically speaking, see close to perfect economic mobility. Economic mobility and meritocracy are not analogous, but they are highly correlated. For example economic mobility can capture movement that is not based on meritocracy if the source of that is roughly equally distributed. Consider how if hypothetically everyone gave preferential treatment to their neighbor, then because almost everyone does have a neighbor then everyone will benefit from some nepotistic non-merit-based source.
However, most of our society does not have that happenstance of nepotism being equally distributed. What actually happens is that social capital is very disproportionately skewed. Even if it wasn't, I don't think we can really justify nepotism.
Meritocracy is Good, Actually
No really, of course we want the best to make it to positions they deserve. It's not just about fairness and equity, it's also what's best for outcomes. Of course it's better for a team to have the better players make it to a team. It's better to have the better researchers doing scientific researches. And it goes without saying that it's better for better managers to be managing.
Obviously this is something we want. We want to hire the best and most qualified people for positions. Yet for some reason, so many institutions especially in the corporate and private sector are favoring connections over merits. How can this be?
Steel-manning Nepotism
I can't believe I have to be doing this, but here we are. Let's try to steel-man the argument for nepotism.
To start off, proponents will not call it nepotism of course. While meritocracy is kind of a controversial concept among those left of center politically, nepotism is universally bad and wrong and shunned by everyone. Instead the concept they will frame it as is trust and networking.
It's not about favoring friendships and relationships with existing people you already know for the sake of it, it's because those people are pre-vetted. Go on LinkedIn and you will see endless posts and comments about the value of networking. Usually as a means to find a job, but sometimes we also get a glimpse of the justification from the other side. Let's look at one:
Let's unpack this.
Trust is very important because hiring is risky.
Calls for fair and inclusive hiring practices come from moralistic approach, rather than a utilitarian means for better performance.
Personal recommendations create trust which lowers risk. You are more likely to know what you're getting.
Soft skills that are harder to measure have been pre-approved.
People just like working alongside their friends and that's valuable in and of itself!
Chemistry exists for people you've already worked with, and that helps with performance.
That's actually quite a lot of important things. Since we are steel-manning here, we should try to meet the goals of the other side. If all we should be caring about are better outcomes, then that'll be our starting position too. No moralism.
The rest of the argument can be combined into two primary points: trust to lower risk and chemistry for heightened performance overall.
There is something to consider about trust and risk. We don't actually have the tools to always measure merit reliably. Let's say you have 100 people apply for a position. There is only so much that a resume and cover letter can tell you about each individual. Of course evaluation is highly uncertain and as a result risky when you don't have much to go off of. That's where connections come in. Connections act as a pre-approval, and make that evaluation of someone less uncertain. This is good and valuable, but this method is by virtue of its nature highly limited. There's only so many people out of those 100 people that you would personally know or have connections with. The vast majority of those people are going to be excluded due to not having the right connections.
But what if we could solve that problem? What if we could have more ways to evaluate a person's merits, whatever they may be? That would be great. Despite earlier claiming that most of society is not meritocratic I did allude that there a few exceptions. How do they do it?
Sports and Academia
Let's look at professional sports, and academia, arguably the two last bastions of meritocracy. So many superstar sports players are on record either in interviews or biographies saying that part of the reason they got into their sport and loved it was because of how they felt their success was in their own hands. That they knew that working hard and practicing would pay off.
While academia doesn't have prolific superstars, it instead has robust meritocratic systems in place. Exams and grades put hard numbers on students' understanding and performance on given courses. To equalize for differences in tests across universities and schools, there are universal tests such as the Gaokao in China where all students will take the same test and graded. For graduates studies there is the GRE (Graduate Record Examination). By putting hard numbers, room for personal feelings and interpretations go aside.
These systems emphasize fairness, but by being meritocratic they are also serving the interests of the institutions to get the best students and researchers in. Meanwhile professional sports teams are private companies with all the same financial incentives that all private sector businesses and institutions have.
If academia uses tests, what do sports teams use for evaluating merit? Well they too have tests and auditions on the junior levels, but at the high level they have sabermetrics aka sports analytics. In fact as sports has grown as an economic sector, so too have teams embraced sports analytics to greater extents. When in the 2000s it was a fringe and innovative idea to find talent based on hard numbers rather than subjective scouting evaluations, it is now standard practice across all big sports leagues.
Are either of these perfect? Of course not. But some tests, as imperfect as they may be, are better than no tests. Some meritocracy is preferable and beats subjective feelings and personal connections. Don't take my word for it, look at Moneyball and see how Billy Beane beat the industry.
But What About Hiring?
Right right. We can make students take tests, and we have such large data volume on athletes to make merit-based judgments on them for sports leagues, but what can we do in the corporate world where hundreds and sometimes thousands of nameless candidates apply for positions?
To be clear, this is a recent phenomenon. We did not have thousands of applicants 20 years ago for positions. The level of competitiveness in today's labor market has skyrocketed due to the wide availability of the internet allowing candidates to rapidly apply for positions across different geographies.
Meanwhile our hiring practices for the most part have remained the same. Candidates are asked to upload a 1 or 2 page resume and a cover letter, maybe have a link to their portfolio, and answer some very basic questions. Then when someone passes the first filter, they are invited to a couple of rounds of interviews where a few questions are asked. In the end, the position is usually given to someone who has those connections or big names on their resume to act as a de-risk. Of course that's what would happen. Of course given such little information, the safest candidate is going to be the one that has some prior connection and where vibes are going to triumph.
This can be improved. We could be asking for more tests. We can look at credentials that have high failure rates to obtain. We could defer to accomplishments and results that are individualized such as achievements in hobbyist competitions. Remember, some test is better than no test.
Artificial Intelligence
But also, perhaps this is a good use case for AI. What if we remove the human emotions and connections and have algorithms and AI try to evaluate merit. This will only work of course if the artificial intelligence has been programmed to value merits, rather than be imbued with same connections-based values of the biased institution.
If AI is going to ruin the labor market, perhaps it'd be good to at least have it also marginally improve the processes that it's helped break in the first place. For these reasons, I actually do support the use of AI in recruitment. I am not an institution so I do actually care about fairness. I believe in my capabilities, but I can't get pass people favoring their friends and connections. Here is an opportunity that serves both sides. Institutions can hire better talent, and we get to face a fairer hiring process.
Comments