Foundations, not surface
Every job posting mixes two kinds of requirements and treats them as one kind. That single confusion explains most bad hiring decisions I have watched, from both sides of the table.
The first kind is surface. React instead of Vue. This cloud instead of that one. Postgres instead of MySQL. A particular vertical, a particular ORM, a particular deployment pipeline. Surface is real, it is not fake knowledge, and someone who already has the layer underneath it picks it up in days or weeks. Not because they are brilliant. Because the concepts transfer and only the names changed.
The second kind is foundations. Has shipped something to production and then lived with it breaking. Can sit with a user and hear what they actually mean instead of what they said. Can look at data and form a view, rather than waiting for someone to hand them the conclusion. Can hold a whole system in their head well enough to know which part will bend first. Foundations take years, and they are only acquired by doing the thing, repeatedly, with consequences attached.
A hiring filter that cannot tell these apart will reject people who have every foundation and none of this quarter's surface, and accept people with the exact surface and no foundation under it. Both errors are expensive. The second one is the one that quietly ruins teams, because it does not look like a mistake until month four.
Here is the question the requirement list is failing to ask. Not "has this person done exactly this" and not the lazy opposite, "can this person learn anything". The real question is: how far is it from what they have done to what this needs, and is that distance crossable at the speed this role actually demands?
I have a personal stake in the answer, so let me put my evidence on the table. I learned to code on the way. I learned design on the way. I learned AI engineering on the way, in public, while shipping with it. I once learned enough real estate appraisal to build a product in it. I wrote about building a personalization engine in Python, a language I do not know, that roughly tripled conversion. At no point was I the person who already had the surface. Every single time, I was the person who had the foundation and closed the distance while the thing was being built.
That is not a boast, it is a claim about what the years actually bought. They did not buy a list of technologies. They bought the ability to arrive somewhere unfamiliar and be useful quickly, which is a different asset, and a more durable one, because it does not expire when the stack rotates.
Now the part where this argument gets abused, because I have seen it used as a licence and it deserves a limit.
"I can learn it" is not a universal answer. There is a real difference between "I know Vue and not React" and "I have never shipped anything to production." The first is a gap and it closes in a week. The second is not a gap, it is a wall, and no amount of enthusiasm walks through it. The same for never having talked to a customer, never having owned an outcome, never having maintained something after the launch party. Those are foundations, and someone who lacks them lacks the thing that makes the rest transferable. Treating a missing foundation as a learnable surface is exactly as wrong as treating a missing surface as a disqualification. It is the same error, pointed the other way.
So the honest form of the claim is narrow: foundations are prerequisites, surface is logistics. Filter hard on the first. Stop filtering on the second.
This matters most at the early stage, and the reasoning is not sentimental. An early company does not know where it is going yet. The product will move, probably more than once. Hiring someone for a deep match to today's stack is a bet that the direction holds, and the direction almost never holds. What survives a pivot is not the person who knew the framework. It is the person who can be dropped into the new thing and be productive before anyone has written the documentation.
There is a version of this for candidates too, and it is the more useful half. If you are missing surface, say so plainly and show the distance instead of hiding it. "I have not used React, I shipped this in Vue, and here is the thing I learned in a fortnight last year" is a stronger sentence than any confident vagueness, because it hands over a piece of evidence the reader can check. Learning speed is not a personality trait you get to assert. It is a track record, and either you have one or you are asking to be believed.
Which is really the whole point. Every CV is a list of surfaces, because surfaces are easy to write down. The thing worth hiring is underneath, harder to name, and almost never in the requirements.