I met two interviewers from a distant land,
who said: twelve feet from the camera stood,
and half the words were lost before they landed—
after Shelley, Ozymandias
Nobody ever hired a good chef by asking them to recite the supplier's batch codes. Yet a surprising amount of this industry, interviews included, runs on exactly that: who knows the most code names, who's read the most leaked documentation, who can name the update before it's been explained. Meanwhile the question that actually matters is much older and much plainer. Does it make sense? Would it work? Does it taste right?
Emily Hall's recent Times piece, "It's more brutal than ever", follows experienced professionals in their forties and fifties who can't get work however many applications they send. The ONS figures it cites put 5.3% of working-age men as looking for work between May and July, against 4.5% of women. One recruiter quoted in it, Rich Howell, says "brilliant people are being filtered out of the application process for all the wrong reasons."
The best line on why came from a LinkedIn comment underneath the post, not from the article. One commenter, who spent fifteen months looking for work after redundancy, wrote that part of the problem is hiring managers with only a few years' experience in total, often dwarfed by what an older candidate brings. "While some can, many just can't evaluate or appreciate the value in that experience." Which means, he went on, that they can't see the difference between the older candidate and a younger one with less experience and lower wage expectations, and believe both are equally capable of doing the job.
That's a more useful explanation than most, because it doesn't need anyone to be afraid or malicious. It only needs someone who can't tell the two candidates apart.
The kitchen test
Think about how a good restaurant kitchen works. The head chef can taste. A line cook who is better than them at fish, or pastry, or sauces, is an obvious asset, because the chef can tell within a service what that cook is worth. The best kitchens tend to be run by people who hired cooks better than themselves at particular things, then built the menu around them.
Now put a head chef in the kitchen who has never tasted a properly made sauce. A cheaper cook turns out something acceptable and the chef, honestly, can't tell it from the real thing. Hiring the cheaper one isn't cynical. From where they're standing it's a sensible decision, and it's the wrong one.
Hiring is the same test. Whether you can recognise a better chef depends on whether you've cooked enough yourself to know what better looks like, and a surprising number of people are making the call without having done that.
Why it's worse in search
A kitchen has an advantage most knowledge work doesn't: the food goes out and the verdict arrives within the hour. SEO, and a lot of marketing, has no such feedback loop. Results lag by months, they're tangled up with a dozen other causes, and the reporting is often built to flatter. A hiring manager can't taste the work, so they fall back on proxies: the job title, the tool names on the CV, the cost. On those three measures, a cheaper candidate with a similar-looking CV is easy to mistake for an equivalent one.
It also explains something the earlier version of this piece got slightly wrong. Interview feedback that is vague, with no example attached ("not quite the right fit", "wasn't as strategic as we needed"), can look like a room defending itself. Sometimes it is. But it's just as likely to be a room that couldn't evaluate the answer well enough to give a specific reason. You can't cite an example of a gap you couldn't see in the first place.
Fear and familiarity, as well
The inability to evaluate isn't the only mechanism, and the research behind the others is worth keeping in view. Organisational psychologists call one of them status threat: when someone's competence implicitly questions the standing of whoever is evaluating them, the natural response isn't always to reward it. Often it's to protect the hierarchy and score the threat down. Boris Groysberg's research into star performers found something similar from another angle: highly capable people moving into established teams struggle to be absorbed, not because their ability disappears, but because the structure resists capability that wasn't grown in-house.
The Peter Principle adds a plainer version: people get promoted until they reach a role they're not quite equal to, then stay there. A manager sitting at their own ceiling has an understandable reason not to hire someone who might show where that ceiling is. And alongside that sits the older, blunter pattern, jobs for the boys. Sociologists call the mechanism homophily, a tendency documented in a landmark 2001 review by McPherson, Smith-Lovin and Cook for people to trust, favour and hire whoever most resembles themselves.
Four different routes, then, and they all end up in the same place: the experienced, expensive, genuinely capable candidate who loses out, usually without anyone in the room having decided to do that to them.
The pipeline problem underneath it
There's a second, related fear worth naming plainly, because it makes the first one worse. A lot of the current conversation about AI and entry-level work isn't really about junior roles at all. It's about where senior competence comes from in the first place. A learning-and-development survey published by D2L this May found 58% of respondents concerned that shrinking entry-level hiring because of AI could leave a real shortage of qualified senior leaders within five years. Separately, an IDC survey carried out on behalf of Deel found 71% of enterprises already reporting it's getting harder to recruit and develop future leaders, because the on-the-job learning pathway junior roles used to provide is quietly disappearing. The same survey found 66% expect to slow entry-level hiring further, while 91% say entry-level roles are already changing or disappearing because of AI.
Worth being fair to the other side of this, because it's a genuinely live, unsettled argument. Research published by the Strada Education Foundation this year found nearly three times as many senior talent leaders expecting AI to increase entry-level hiring in 2026 as to decrease it, and the clearest pattern across the wider data so far looks more like a shift in what entry-level work involves than its outright disappearance. Nobody serious is claiming this is settled.
But if even a version of the concern is right, it connects straight back to the kitchen. Fewer people get the years at the stove that teach them what good tastes like, which means fewer people able to judge it later. An industry that sidelines its experienced people while also thinning the route to becoming experienced is closing the same door from both ends.
This isn't just a story about hiring rooms
It's the same shape as something much older on this site's own subject. In 1999, the Cluetrain Manifesto put it in five words: hyperlinks subvert hierarchy. The promise of the early open web was exactly that. A smaller site, a genuine expert, an outsider with nothing but a better answer, could out-rank an institution simply by being more useful and more genuinely linked to. Authority routed around whoever used to gatekeep it. For a while, more or less, it actually worked that way.
Then it got exploited (link farms, reciprocal schemes, paid placement dressed up as organic authority) and once it was exploited, it got managed. Not fixed. Managed: rankings data locked further behind paywalled and partner-gated access, backlink signal made progressively harder and more expensive to read, each new closure justified by the abuse it was supposedly stopping. Some of that really was necessary. But the net effect is the same one this piece keeps circling: a mechanism that once rewarded genuine competence over comfortable familiarity quietly learns to protect whoever's already sitting in the room, whether the room is a hiring panel or a search index.
The word for it, when it's said out loud
Usually this mechanism stays quiet, which is most of why it's hard to point at. Occasionally it gets a name. Michael Lewis's new book Blockers: Rebels in the Deep State, extracted this month in the Guardian, reports that staff inside Elon Musk's Department of Government Efficiency used exactly one word for the career civil servants they removed from US federal agencies in 2025: blockers. Not incompetent. Not wrong. Blocking: standing between a young, loyal team and what it had already decided to do.
Worth being precise about what's being drawn on here rather than taking a side on. Whatever the merits of DOGE's broader mission, a genuinely contested political question this piece has no need to settle, the word itself is a clean example of the move described above, just said plainly instead of left unspoken. Reporting from Lewis's book describes career staff who understood procurement law, taxpayer data protections and the institutional reasoning behind them, reclassified as personal obstacles rather than people whose knowledge was simply inconvenient to a decision already made. It's the kitchen test failed in the loudest possible way: not "I can't tell what this person is worth", but "what this person knows is the problem, and the problem gets removed."
How good kitchens fix it
None of this is fixed law. Chefs have a simple answer to the can't-taste problem, which is the trial shift: you don't hire someone on their CV, you put them on the line for a day and watch what comes out. Orchestras arrived at a version of the same idea. A widely cited 2000 study by Claudia Goldin and Cecilia Rouse found that auditioning behind a physical screen, so the panel could only hear the playing, shifted who got hired, though how large the effect was has been argued over since. The principle is the part that holds. Evaluate the work itself, and the proxies lose their grip.
An institution that does nothing equivalent doesn't stay neutral by default. It keeps hiring its own reflection, gets a little more staid and a little less capable of surprising itself each time, and rarely notices, because nothing about staleness announces itself.
What you do with this, from the other side
You can't argue someone into seeing what they can't see, and trying tends to confirm the story the room had already settled on. What you can do is stop treating one room's verdict as a verdict on your ability, and make the work easy to taste. Bring the thing you built rather than the description of it. Offer the task, the audit, the walkthrough. Put it in a form that can be judged by someone who has to rely on evidence rather than instinct, because that is the one kind of evaluation that doesn't depend on the evaluator having cooked as much as you have.
And if you're ever the one deciding, hire the people who are better than you at something. Then listen to what they say about the menu.