The Missing Pieces of the Skills Economy are the Skills and the Economy

The idea of a “skills economy” finally appears to be getting broader interest and increasing action, for many reasons. AI is certainly one. Jobs are changing faster than ever. The jobs that need doing and the skills required to do them evolve visibly over months rather than years now. And AI could enable us to create and analyze digital evidence of a person’s skills with substantially less labor (although we’re still quite early in that shift).

Other, less sexy changes are also major drivers. The high cost of college is finally biting hard enough that students choosing to go directly from high school to work is now a familiar trend to parents and friends of parents. For an increasing portion of middle-class high school students, the value of incurring substantial debt while spending four years figuring out how you’re going to pay it off when you graduate seems daunting. Meanwhile, globalism is reconfiguring itself and, in some ways, retreating a bit. Geopolitical pressures and fractures are rerouting and shortening supply chains. Oil prices and data center growth are both driving electrification. You can buy your solar panels from China, but you can’t have them installed on your roof from there. Whether it’s the US, Canada, Europe, or other parts of the world, many of these changes are hitting at the same time. People need to demonstrate, evolve, and migrate with their skills at a pace never seen before.

All of this should be positive for developing a skills economy. And it is—with one little problem. We don’t agree on what a “skills economy” is yet, and I’m not sure we know that we don’t agree. I believe there is an answer. It’s just a messy one in the way economies need to be. It’s not tidy in the way that ontologies need to be. And that changes the way we should be approaching competency standards and infrastructure. Talk about skills tends to sound like talk about widgets. Like they’re cogs in a machine.

They are not. Let’s talk about what a skill is, why we get confused, and what it would take to enable a true skills economy.

What’s a skill?

For starters, I don’t think we’re communicating about the nature of skills clearly. Take me, for example. I studied philosophy in college. When my professors mentioned the value of the degree at all, which they rarely did (at least in a positive way), they might have said something about writing. But writing what? If they meant anything specific by that, it was generally either law briefs or philosophy papers. Nobody said, “writing good blog posts.” Blogging wasn’t a thing yet. LMSs weren’t a thing either. So nobody was saying, “If you’re interested in educational technology, you’ll want to learn about writing online.”

A bit less than twenty years later, when I started my blog, I did it to explore the medium at a time when educational technology was just becoming a thing at scale. LMSs existed and were used at many universities (though not broadly or deeply yet, and not in K12 yet). Blogging was having a moment. Nobody said to me, “Blogging will help you become a social media influencer and build your personal brand for your career.” But it did those things. I wrote the way I learned to write as a philosophy student. People liked it. I developed an audience of readers I don’t know. Interestingly, e-Literate is read by very different people. I learned to write my posts so that different audiences would each find my writing understandable and resonant in ways that worked for them. I learned to think about nuances of meaning that could be read differently. Nobody told me that was a good skill to develop for prompt and context engineering with an AI. But a bit less than 20 years after I started blogging, I am discovering just that. I’m good at prompting AIs.

Something I learned as an undergraduate philosophy major in the 1980s was a durable skill. Can you name it? I can’t. “Writing” is too generic. I get the impression that some competency-based education (CBE) implementers think, “Well, skills have subskills and sub-subskills. If you articulate the full 37-level skill taxonomy, in four-part harmony, everything will be clear.” The things I learned in the 1980s didn’t have stable value that remains the same in 2026. I doubt we would even name them the same today. “Skills,” in this sense, are whatever slices of what I’ve learned that have economic value in the current context.

Let’s take a different example. Suppose a cashier time travels from 1926 to 2026. They’ve never seen an electronic cash register before. Or a bar code scanner. Or a smartphone. Or a credit card. Do you think none of their skills will transfer? Can we break apart which ones will? Maybe, to some extent. They’d probably actually be better than their modern counterparts at counting change and noticing when a bill total doesn’t look right. So maybe we could tease out the durable skills.

Even assuming you could recruit a large workforce from 1926, would the effort of mapping the skills ontology be worth it?

There are, of course, many situations when having a formal skills ontology makes sense. I would like my surgeon to be board-certified. And I would like my X-ray tech to be certified too. Most jobs aren’t like that, though, and not because of the skill level involved. In cases where skills certification rules, the deciding factor is often the cost of certifying versus the risk of not certifying. In these cases, the employer is almost never the one maintaining the skills ontology and evidence standards. That’s too hard and expensive. There’s usually a certification body. I’ll have more to say about these proxies later in the post.

Why skills can be confusing

There are at least three different ways we can mean “skills” or “competencies,” and we tend to mix them up. The first is the way I’ve been talking about. A “skill” is a thing that has contextual value. If I can program COBOL, that’s not valuable to the vast majority of the world and very valuable to a tiny number of organizations. Are COBOL programming skillls things with value? It depends on whether you need to maintain COBOL code.

The second meaning of “skill” can look the same but works substantially differently. Institutions have skill definitions that are more about their own accountability and differentiation. Are they meeting governmental standards of education? Are they meeting their own? The states of Georgia, North Carolina, and South Carolina have converted their state-mandated K12 competencies into the 1EdTech Competencies and Academic Standards Exchange (CASE) format. They can now ask, “Does this OER lesson from the neighboring state address our competency? If not, what needs to be adjusted?” While the intention is that these skills have value to individual students, the mandate is to ensure all students are taught them. Skills, in this meaning, are about institutional accountability.

Here’s another skills question that academic institutions need to answer: Should this student get transfer credit for taking a course with a similar name to one of ours at a different institution?

Credit transfer is actually a huge potential win for digital credentials in general and skill definitions in particular. Colleges and universities burn huge amounts of faculty and staff time looking over syllabi and course descriptions to decide, “Should this student get transfer credit.” Unfortunately, academia is more effective at adding up lost tuition dollars from granting transfer credit than they are at accounting for productivity costs of having skilled employees do work that software could cut by maybe 80%. (And to be fair, many companies are bad at this too, which does add to the uptake friction for the skills economy.) The kinds of skills valued in the school-to-school economy are different.

The third kind of skill is important but a lot more elusive than we pretend. It’s a fundamental building block for learning a topic like math or language. The evidence-backed framework that I keep going back to is Ken Koedinger’s Knowledge Components (KCs). These are small pieces that may be psychologically real—meaning they’re actually represented in your head in roughly the way we talk about them—and that build on each other. You can’t add mixed fractions until you can add fractions. KCs are actually hard work to identify, and while they can add up to some kinds of skills in either the academic or workforce economies, they are not the same. They are important and related, though. If you need to learn one KC before you can learn the next, and a collection of those KCs together amount to learning something of value…isn’t that essential to education?

Technical standards for skill definitions can help us here too—and it matters. We can’t know if students are making progress without a definition of what they are supposed to be progressing toward. Learning activities are impossible to interpret unless you know the learning goal.

How economies work (and how they don’t)

I often hear that the hard part of getting the skills economy to scale is getting skills infrastructure adopted by employers and talent management systems.

It is not.

The calculation is simple: Employers will adopt—and demand—skills infrastructure when it provably reduces costs or increases profits by more than it costs to purchase and implement. While I am not one to claim that companies are perfectly rational, they tend to be more rational than not about money. Economies work less well when either the value is low or it’s hard to assess. Difficulty to assess can really bite. As I write this, the uncertain insurance risk of oil tankers passing through the Strait of Hormuz during a military conflict is enough to halt the flow of tankers. Uncertainty kills markets and makes products nonviable. The same is true for skills. Workers have skills. Employers need skills. What’s been missing is the ability to assess the match cheaply enough for the transaction to happen.

Consider the case of buying, selling, or building a house in a rural area with thin comps. If you’re trying to value a two-bedroom apartment in a large building with many nearly identical apartments, in a city with many similar apartment buildings, the value is often straightforward. “Fifteen other apartments similar to yours have sold for between $X and $Y in the past year.” In a rural area, it’s harder. “This is a passive house.” Sorry, but there are no passive houses in the area, and you paid a lot for that insulation. You didn’t build in a place where people are buying passive houses. “But I built a huge, beautiful garage.” Actually, around your area, many people don’t have garages and won’t pay extra for one. “That’s BS. If I can just find somebody who values the house like I do, I can get what it’s worth.”

Yes. But not in the way you mean. In an economy, your house is worth whatever somebody is willing to pay you for it. If you’re lucky, you’ll find that person. But possibly not for the reason you thought.

My wife and I bought the last house we owned because it had a mother-in-law apartment that we needed for—wait for it—my mother-in-law. The sellers were sure the studio space in the barn was the thing that would earn them their price.

Value assessment proxies

Now, there can be disagreements about value in this situation, which is one reason why appraisers exist. They’re particularly important in building new houses. The prospective homeowner says, “I want to build a passive house with a giant garage in this rural area.” The appraiser tells the bank, “This house costs X more than buyers in this area typically pay for because they don’t pay for fancy garages or passive houses.” The bank says, “I’ll pay you your building costs minus X.” Appraisers have to justify their assessments with evidence and precedents. That’s one kind of proxy function for assessing value.

Degrees—particularly from prestige universities—and high-stakes industry certifications handle the value assessment burden in different ways. Some firms say, “We only hire from Ivy League colleges, because we know we’ll get smart people we can train.” Notice: No skill evaluation. Likewise, some middle-skills jobs without formal industry certifications often see degree inflation: “Well, the person taking this job doesn’t need a BA, but if I require one, I can be more confident that the person has or can learn the skills I need.” Degrees are often proxies for holistic capabilities of a person that largely bypass the skills question. Not always, and not equally. A pharmacy or engineering degree may convey more about economic skills than a literature degree. But even with these degrees, there can be considerable differences regarding what was taught and what was learned.

Certifications are also proxies, though they generally have specific verification behind them. A phlebotomy certification tells a hospital, “I can hire this person, because they are certified. I don’t have to check.” Most colleges and universities do not certify in this way for most degrees (although that is changing, and it’s less common in the US than elsewhere).

AIs can now play this proxy role, a little like appraisers reading rough comps in a market with variable inventory. But only with the right infrastructure. First, they need a skill certification in a form they can read. “The skill being certified is depreciation for bookkeeping, by which we mean the following….” 1EdTech, my employer, has a stardard called “CASE” that does exactly this. It’s a skill definition that stands on its own and travels. To do the job well, the AI also needs evidence. “Here’s the work the person did that shows they mastered the skill being certified.” This is supported by the Open Badge specification, which supports CASE. So you can put together a little package that says, “Here’s the skill we think the person mastered, and here’s why we think they did.”

The employer, or the transfer school, or whoever, can show the package to the AI and prompt, “I’m looking for evidence the person can do this thing Y that I need [where Y is similar but not identical to X]. How confident can I be from this package that this person can do Y?” The AI can spit out an answer. The quality of the answer will depend partly on the quality of the information in the package. For example, if the package contains a series of ordered learning events related to the skill, with evidence of where the person struggled and how they improved over time, the AI can infer more about a skill assertion than it can from a summary paragraph (or no evidence at all). The organization or individual vouching for the competency counts too. That’s part of the evidence. But it doesn’t have to be the sole evidence anymore.

This is a massive shift, particularly at a time when the signals from résumés and from LinkedIn are being devalued by AI. (AI giveth, and AI taketh away.) When I interview a candidate for a job, I don’t think about the job description. “Must be a team player.” I write the job description partly to attract the kind of person who might be more likely to answer well when I ask them interview questions about collaboration skills. If an AI could point me to evidence that person has already shown about a skill, I might be satisfied with that, or I might ask better questions that elicit more revealing answers. Either way, that’s valuable to me as a manager, and it would be valuable to my organization.

We have the standards infrastructure and raw AI capability to do this today. The main barrier is that schools are not yet issuing this kind of evidence or thinking of skills this way.

Getting to the point

An economy enables people to agree on the value of something so they can transact for mutual benefit. Skills are worth, economically speaking, whatever someone will pay for them. Not what they should be worth, not what they cost to acquire, not what a taxonomy says they are, but what someone will pay. The participants in the transaction determine the value. Other kinds of value exist and matter—fulfillment, citizenship, what I can teach my children—but those aren’t the same as economic value, and conflating the two has been part of our confusion.

What a skills economy needs, then, isn’t better definitions of what individual skills “really” are. It needs infrastructure that lets two parties point at the same thing, with attached evidence to claims about it, and make a judgment about whether the match is close enough for the transaction at hand. Skills, in this sense, are semantic infrastructure you can reason over, not entities you can discover. A skills statement is a claim that something useful can be done, paired with a specification of what evidence for the claim should look like. When enough of those exist, and when using them adds more value than the cost of adopting them, a skills economy emerges. Not before.