You'll hear it constantly, from HR departments and governments and headlines: there's a skills gap. Employers can't find the people they need. The workforce isn't keeping up with what the economy requires.
The data mostly supports this claim. Millions of jobs go unfilled while millions of people remain unemployed or underemployed. Something is clearly misaligned.
But dig into what "skills gap" actually means in practice — what employers can't find, and why — and a different picture emerges. One that's more interesting, and more useful, than the headline version.
The skills employers say they want
When companies describe their hiring difficulties, the most common complaints are: soft skills (communication, collaboration, critical thinking), technical skills that don't exist in the labour market yet (roles in AI, cybersecurity, advanced data engineering), and domain-specific experience that takes years to develop (specialised engineering, clinical expertise, regulatory knowledge).
The first category is worth pausing on. "Soft skills" is a terrible name for what are arguably the hardest skills: the ability to communicate clearly under pressure, to navigate disagreement without fracturing a team, to take a vague problem and turn it into a clear action. These are cultivated over years and they don't show up on a CV.
The second category — roles that didn't exist five years ago — is a supply problem with a structural cause: education systems train for the world as it was, not as it will be. A university curriculum takes years to update. The market moves faster.
The third category is partly a chicken-and-egg problem of the same type described earlier: you need experienced people, but you won't invest in creating them.
The gap companies won't admit
What rarely makes it into the official skills-gap narrative: many companies have made their own roles increasingly hard to fill by raising requirements without raising pay.
Between 2010 and 2020, the median requirements for jobs in the US increased significantly — more required credentials, more required experience, more required technical skills — while median wages in many sectors grew slowly. When the pay doesn't keep up with the requirements, fewer people choose to develop those skills. That's not a mystery, it's basic economics.
There's also a training investment problem. Companies today invest significantly less in on-the-job training than they did in the 1970s and 1980s. They've offloaded the cost of skill development onto individuals and educational institutions, then complained that individuals and educational institutions aren't doing it fast enough. The skills gap is partly a training withdrawal gap.
The skills that are actually scarce
Stripping back the rhetoric, the truly scarce skills tend to fall into a few clusters.
Translation skills: the ability to move between technical and non-technical worlds — to explain what the data means to someone who doesn't understand statistics, to understand what the engineer is building well enough to sell it. These people are rare and extraordinarily valuable.
High-complexity judgment: roles where the situation is always slightly different and the decision can't be fully systematised — experienced clinicians, senior engineers, good managers, skilled lawyers. These take years to develop and can't be shortcut.
New-domain early expertise: people who got skilled in something emerging before the mainstream caught up — early AI practitioners, certain cybersecurity specialisms, regulatory experts in new industries. These windows of scarcity open and close quickly, which is why the people inside them get paid so well temporarily.
What this means if you're choosing what to get good at
The useful implication isn't "study AI" or "get certified in whatever is trending." It's something more durable: scarcity is the mechanism, and scarcity comes from genuine difficulty or genuine earliness.
Genuine difficulty: things that take years of deliberate practice, things that require both technical and human skills, things that require judgment under genuine uncertainty. These stay scarce because they're hard to replicate and impossible to automate cleanly.
Genuine earliness: moving into a domain before it becomes obvious, building expertise while the pool is still shallow. The risk is that the domain might not matter; the reward is that if it does, you're one of a small number of people who can do it.
The skills gap, properly understood, is a map of where genuine value is being created and not captured. That's actually useful information — if you're willing to read it honestly rather than take the headline version.