Today, I Wouldn't Hire My Younger Self

I put my younger self through a 2026 hiring process. I failed the screening. What started as a joke exercise left me with a question the industry has no good answer to.

A few days ago I wrote that the career path I took no longer exists. I started with the jobs nobody wanted. Add photos to a page. Publish a news item. A client ran out of mailbox space? Increase their quota by a gigabyte.

Since publishing that piece, one thought kept nagging at me. If that path is gone, what exactly did we lose along with it?

I decided to check in the simplest way possible.

I sat down on the other side of the table and put my younger self through the hiring process. The version of me at the start of my career, applying for a junior frontend developer position in 2026.

The screening

Let's start with the paperwork.

CV: one entry, customer support and minor website fixes. Not a single technology from the job posting. No framework.

Portfolio: none. GitHub: none. Side projects: none.

Verifiable skills: adding photos to a CMS, publishing news items, increasing mailbox quotas.

Now open any junior posting from this year. React or Next.js, TypeScript, Git, personal projects a plus, portfolio link required.

The result is obvious. I'm out.

Not at the technical interview. Not at the take-home task. At the initial screening stage, before anyone even talks to me.

For a moment you can laugh at this. It was easier to get into the industry back then, right? Except that's not true. It wasn't easier. It was hard in a different way. And that difference is the whole point.

What the CV could not show

That kid had nothing on paper. But he had one trait no document has ever been able to show.

He wasn't afraid to poke around in someone else's production system. And when he broke something (and he did), he sat with it until he understood why. Not because anyone told him to. Because he couldn't do it any other way.

Everything I know today grew out of that one habit. Not out of frameworks; those changed every few years. Not out of specific technologies; I no longer use most of them. Out of the habit of never letting something work without knowing why it works.

And here is the thing I only understood while running this experiment.

The old entry path was an accidental filter for exactly that trait.

The boring, manual work weeded out people who didn't want to understand. Slicing Photoshop templates, rewriting code between frameworks, fixing other people's websites. You couldn't get through it by taking shortcuts, because there were no shortcuts. The only way to stop struggling was to understand what you were doing.

Nobody designed that filter. Nobody wrote it into any hiring process. It worked as a side effect of what the job looked like.

Which is why nobody noticed when it disappeared.

The signal that stopped being a signal

Today's junior candidate shows up with a portfolio I couldn't have built in a year back then. Three Next.js apps. Deployment on Vercel. Tests. A polished README with badges.

Ten years ago, that portfolio carried information. It said: this person spent hundreds of hours on this, fought through documentation, got stuck and got unstuck, set up an environment from scratch. The artifact itself was proof of the process.

Today the same artifact is the output of a single session with an agent.

I'm not writing this to take a shot at today's juniors. Some of them genuinely went through that process and genuinely understand their code. The problem is that by looking at the portfolio, you can no longer tell them apart. The portfolios look identical from the outside.

In my previous piece I wrote that review is infrastructure nobody rebuilt after adopting AI. Generating code became cheap, verifying it stayed expensive, and the difference was paid by the people accountable for quality.

Junior hiring follows the same pattern, just one level higher.

Producing the signal "I can code" became cheap. Verifying whether that signal reflects real understanding remained expensive. And hiring processes still measure the cheap part: repositories, projects, lists of technologies.

We are filtering on artifacts from an era when artifacts meant something.

The old accidental filter vanished together with the old path. Nobody built a new one. What's left is a hole that lets the wrong people through in both directions. In come candidates with impressive generated portfolios and zero need to understand. Out go today's versions of that kid with the photos and the mailbox quota.

What I would look for

I don't hire juniors. I don't have a ready-made process to show you, and I'm not going to pretend I do.

What I have is a hypothesis. If I ever had to hire a junior, I would stop looking at what they built. I would start looking for evidence of how they respond when they don't understand something. Because that is the trait the old filter caught by accident.

Three signals that I believe survived the change of era:

A hypothesis before a prompt. Broken code on the table and one question: what is happening here? I don't care whether the candidate finds the bug. I care what they do in the first two minutes, before they reach for AI. Do they form their own hypothesis and try to test it, or do they delegate the thinking right away.

"Why does this work" on their own code. A random fragment from the candidate's portfolio and a request to explain it line by line. Not to catch them using AI, using AI is fine. To check whether the code they put their name on has actually been understood, not just generated.

A story of breaking something. Tell me about a time you took something down. A candidate who has never broken anything has never touched anything. A candidate who broke something and can tell you what it taught them about the system has exactly the habit I'm looking for.

Notice that none of these signals measures knowledge. AI has made knowledge itself a much weaker differentiator. All three measure the same thing: whether the gap between "it works" and "I understand why it works" makes this person uncomfortable.

That kid with the photos and the quota would not pass today's screening. But he would pass these three tests without any preparation. Because they don't check where you are on the road.

They check whether you are walking it at all.

A question, not an answer

I don't know how to scale these three signals to a process that handles 300 applications for a single opening. Honestly: I don't. It may not be possible, and artifact-based screening may stay with us, growing blinder every year.

What I do know is what this experiment says about the cost.

The price of entering this profession did not drop to zero. It changed currency: from hours of boring work to the self-discipline of understanding. The problem is that hiring still prices candidates in the old currency. It counts artifacts that can now be generated, and it cannot see the one trait that can't be.

Today, I wouldn't hire my younger self. And until someone builds a filter to replace the one we lost, we will keep rejecting the kind of people who built this industry in the first place.

I don't know how today's hiring processes are supposed to identify that kid.

I do know that every team that passes him over will be worse off for it.