You're a Lemon

No matter what your mama says, you're not special. Neither is your neighbor who got the job you wanted, or the four hundred other people who applied alongside you. Don't take this as an insult; I'm sure all of you are talented in your own right. Rather, this is an observation about a broader problem in the job market, and, surprise, surprise, it's about to get worse.

For most of history, the market ran on the assumption that employers could tell people apart. Unfortunately, they can't anymore.

Take a look at the numbers, and you'll quickly see that things don't really make sense. Applications per job is at an all-time high, and job openings are still sitting at around 7.6 million. Yet somehow, new grads are complaining they can't get hired anywhere, while companies swear that talent is nowhere to be found. And, of course, everyone freaks out and screams, "AI is coming for our jobs!"

I'm here to tell you, boys and girls, that this narrative is wrong. The jobs still exist. What is fading away is the market's ability to tell who's good. At the root, we don't have a job shortage problem, we have a signal problem.

We're All Lemons Now

There's a famous idea in economics about used cars. When someone tries to sell a car, they, as a seller, know if the car is a peach (good) or a lemon (bad). The buyer can't tell, so they would only pay the average price of a peach and a lemon. However, nobody with a real peach would want to sell at the average price, so they walk. Now that the peaches are gone, the buyer notices, pays less, and more good cars leave the market.

This applies to all markets with some sort of information asymmetry. In the labor market, similarly, you know if you're good but the employer doesn't. I'll leave it as an exercise for the reader to find the analogy there.

There's only one escape hatch to not get priced like the average of everyone in the pile. The strategy is to show some sort of signal that you're not a lemon. That's why the degree, the resume, the cover letter, the portfolio, and the interview exist.

The Death of Modern Signals

The catch is that signals aren't random. You can't just say you're good without having anything to prove it. Talk is universal and cheap, so it tells the employer nada. So the only way to separate yourself is to do something that is harder for a bad candidate to do than a good one. Economists call this a signal.

The degree is a classic example of a signal. The reality is: a degree might teach you absolutely nothing (certainly true for all my dear friends) and still work, as long as it's harder for a weak person (lemon) to get than a strong (peach) one. There's actually concrete evidence that signaling is real. It is a fact that finishing a degree pays a lot more than doing almost all of it and not finishing. A shittily laminated piece of paper alone jumps your salary.

Every job-search tool ever built is just a machine that sends and/or reads these signals. The last company to fix the signal was LinkedIn. Its real innovation isn't the job board, it was innovating a new signal when the last signal broke (job aggregators). LinkedIn made your identity, your network, your reputation a credible differentiation. For a while, it worked.

It is now broken in two ways. First, LinkedIn drifted, and now makes almost all its money helping recruiters, not job-seekers. Obviously, the product serves the person who pays. Two, "Easy apply" (more on this later) and gamed endorsements. The last real attempt to fix the signal collapsed too.

AI Job Apocalypse?

A signal only works if it's expensive for the wrong people to replicate. AI made certain signals as cheap as fentanyl. Everyone and their moms now have a tailored resume and cover letter: they're instant and identical across everyone who uses them. The thing that used to separate a strong candidate from a weak one costs nothing now, and thus separates nobody.

In the jargon of economics, the market went from separating, where good and bad look different so the employer can tell, to pooling, where everyone looks the same so the employer is back to guessing the average. We're all lemons!

Why People Are Stupid

Ready for the most stupid part? Think about the job-search advice that everyone has been giving you: "apply to more jobs!"

Think about what this actually does. Every extra application makes the pile bigger and the signal weaker. This is collectively suicidal behavior. As total volume rises, the information in any single application falls towards zero.

Now apply this to the tools you see on Instagram reels that auto-apply to hundreds of jobs on your behalf. They took the single dumbest behavior in a broken market and automated it. The auto-appliers are pouring gasoline on the exact fire it tries to put out.

So what do employers do when resumes stop meaning anything? They retreat to the signals that are still expensive for AI to fake: verifiable prior experience in the field, and someone they trust vouching for you. This is exactly why "entry-level" jobs now ask for years of experience, and why your idiot friend is getting the job through nepotism and warm intros. This is a sign that the market is fleeing towards the last signals that still mean something.

But there's a trap they're not seeing here. Where does prior experience come from? Entry-level jobs, right? That's literally the point of entry-level jobs. It was never really about cheap labor, it was the place a novice turned into someone who knows things by doing them. In other words, entry-level jobs are by definition an investment. Without entry-level roles, there will be no senior-level people in a couple of years. Yes, Claude probably codes better than the average intern. But who's doing the system architecture in a couple of years if you tell the interns to fuck off? (I, for one, would probably not trust Claude to do that.)

The market is now demanding proof of experience while trying to eliminate the only place experience gets made. This will not hold. Entry-level roles will come back. And when it does return, there needs to be a type of signal.

The Pursuit of a Universal Signal

Volume doesn't work. Prior experience won't hold. Nepotism != peach. The fix is a new signal layer. One built on the things that AI can't cheaply fake. Moreover, it has to be built for the candidate, not sold to the recruiter.

Somebody is going to build it. The market yearns for it. The old machine is out of gas and everyone can feel it even if the reason is still ambiguous.

You're not special. But you're not a lemon either. The problem was never you. It was that nobody could tell the difference. Fix that, and everything else follows.