Signalocracy vs. Meritocracy
Meet Dana. Dana is made up, but you've met her. She is the best analyst at her company. She finds the errors nobody else finds, and she trains the new hires. She has no degree, she says "um" a lot in meetings, and she buys her work clothes on sale.
Now meet Chip. Chip is also made up, and you've met him too. He's a perfectly fine analyst. He also has a degree from a famous university, a firm handshake, and an uncanny ability to say "let's double-click on that" with a straight face.
A new director job opens up. The hiring committee has never met either of them.
Chip gets the job.
Was it merit? No. Was it nepotism? Also no. Nobody on the committee is Chip's uncle. Something else happened here, and I think we need a new word for it.
We live in a signalocracy. Your status, your position, and your ability to make money depend largely on what you can signal to other people and to institutions.
We Want a Meritocracy
Most of us want to live in a meritocracy, where the best people rise because they're the best. Fun fact: the word comes from a warning. The sociologist Michael Young popularized it in his 1958 book The Rise of the Meritocracy, a dystopian satire. Young spent his later years dismayed that people had taken his warning as a to-do list.
Still, the basic impulse is a good one. If you need brain surgery, you want the best surgeon, not the mayor's nephew.
We also know we don't have a pure meritocracy. Talent isn't handed out in a vacuum. As I argued in Motivate, environment does much of the work we credit to character. Luck does a lot of the rest.
But We Don't Live in a Nepotocracy Either
The cynical view says it's all connections. There's some truth to that, and parents with money and pull really do pass advantages to their kids. But pure nepotism can't explain how most people get most jobs. There are too many positions and too few uncles.
Here is the real problem: most of the time, the people making decisions about you are strangers.
Merit is invisible to strangers. They can't watch you work for three years before they hire you. They can't see inside your head. So they do what any sensible person does when they can't observe something directly. They look for clues.
Welcome to the Signalocracy
The economist Michael Spence won a share of the Nobel Prize in part for his work on this idea. In his model of the job market, an employer can't see a worker's ability, so the worker invests in a signal, like education. The signal can work even if the education itself teaches nothing useful, because it's easier for able people to get.
Once you see signals, you see them everywhere:
- Degrees: where you went matters as much as what you learned
- Clothes: the watch, the shoes, the $400 hoodie that looks like a $20 hoodie
- Speech: your accent, your grammar, your vocabulary, and whether you say "summer" as a verb
- Places: the neighborhood, the gym, the club, the conference
- Achievements: the marathon, the book, the award
- Jobs: the logos on your résumé
None of these are merit. They are claims about merit.
Is This Bad?
Not so fast, Bucko. Let's start with the case for the signalocracy, because it's a good one.
Signals let strangers trust each other. Without them, we'd be stuck hiring our cousins. A degree can carry a kid from nowhere past a gatekeeper who would never have met her family. For all its faults, a signalocracy is a huge improvement over an aristocracy, because many signals can be earned.
In that sense, the signalocracy is a sort of meritocracy. It's just a meritocracy viewed through a smudged window. I think of it as an imperfect epistemic layer sitting on top of an imperfect meritocracy. Reality is unequal, and then our knowledge of reality is distorted on top of that.
And that smudged window has real problems.
Some signals are noise. In a famous experiment, Marianne Bertrand and Sendhil Mullainathan sent fictitious résumés to employers and randomly assigned the names. Applicants with white-sounding names got about 50 percent more callbacks than identical applicants with Black-sounding names. A name tells you nothing about how well someone can do a job. The signal still worked, and it did harm.
Some signals are for sale. Raj Chetty, David Deming, and John Friedman found that children from top 1 percent families were about twice as likely to attend an Ivy-Plus college as middle-class kids with comparable test scores. Here's the part that should make us indignant. The things driving that advantage were legacy preferences, non-academic ratings, and athletic recruitment, and those things weren't associated with better outcomes after college. Test scores and academic ratings were. In other words, the signals that favored the rich weren't the ones that told you much.
Some signals measure sameness. Lauren Rivera's Pedigree studied hiring at elite banks, law firms, and consulting firms. Over half of the evaluators she interviewed ranked cultural fit as their top criterion at the interview stage. Fit often came down to whether a candidate shared their background and hobbies. Apparently, squash is a leadership skill.
Some signals are wasteful. When a signal gets valuable, everybody has to buy it. Then it stops separating anybody, so everybody has to buy the next one. That's how a bachelor's degree turns into a master's degree, with a lot of tuition paid along the way.
Some signals are rationed. This one is my personal hobbyhorse. Harvard works a lot like a Supreme hoodie: it's valuable partly because it's scarce. And when existing institutions control who may issue a signal, you get a cartel. That's the story I tell in The Caging of the American Mind.
Put all of this together and you get what I call the Signal Gap: the distance between what a person can do and what the world can see. Chip's signals run ahead of his ability. Dana's run far behind hers.
The Signal Gap isn't just inefficient. It's a moral problem. Every talented person who gets passed over because they couldn't afford the right signals is a case of avoidable misery, and avoidable misery is the kind we're obligated to do something about.
How to Improve the Signalocracy
We can't abolish signals. Strangers will always need clues. So let's stop asking how to get rid of the signalocracy and start asking how to make it better.
Here's a simple test. A good signal should be TRUE:
- Tied to the work: does it measure something the job actually requires?
- Reliable: is it hard to fake, and would it mean the same thing tomorrow?
- Universally available: can a talented person get it without being rich or connected?
- Examined: has anyone checked whether it predicts anything?
With that test in hand, here are seven ways to improve things.
- Test the work itself. Use work samples, portfolios, and auditions. Many orchestras audition musicians behind a screen, because what matters is the sound. If you're hiring a programmer, watch them program.
- Audit your signals. Compare what you screened for with how people actually performed. If a signal doesn't predict success, drop it. This costs almost nothing.
- Hide the noise. Take names, addresses, and photos off the first round of applications. If it can't help you make the decision, it can only hurt.
- Lower the price of good signals. Make exams and credentials cheap, or free for those who can't pay. A good signal that costs $80,000 fails the U in TRUE.
- Break up signal monopolies. Let qualified people start new colleges, credentials, and certifications, and let them compete on whether their graduates can actually do the work. Full disclosure: I run a new university, so I have a dog in this fight. Judge the argument accordingly.
- Send more information. "Has a bachelor's degree" is one bit of data about four years of a life. Show what the person built, what they can do, and what their teachers and colleagues observed.
- Stop overpaying for tiny differences. When many applicants are clearly qualified, consider a lottery among them. Why make teenagers spend years in an arms race over distinctions that predict nothing?
AI makes this urgent. For decades, a polished cover letter was a decent signal of care and writing ability. Now it takes ten seconds and costs nothing. I am AI-curious, and I think this may be a blessing in disguise. When cheap signals collapse, we'll have to go looking for better ones.
What You Can Do Tomorrow
If you hire people, pick one signal you lean on and ask whether it's TRUE. Then replace it with a short, paid work sample and see who shows up.
If you're a Dana, don't wait to be discovered. Make your work visible. Write it down, publish it, and ask for the credit you've earned. It isn't fair that you have to, but it works.
If you're a Chip, and honestly most of us are at least a little bit Chip, go find a Dana and put your signal behind her. Say her name in the rooms she isn't in.
We'll never see each other perfectly. But we can clean the window.