In 1985, Apple’s board fired Steve Jobs. He was 30, the company he’d founded a decade earlier had grown beyond his management capabilities, and John Sculley (the Pepsi executive Jobs himself had recruited) orchestrated his removal. Jobs spent the next twelve years in the wilderness. He started NeXT, which nearly bankrupted him. He bought Pixar, which nearly bankrupted him again before Toy Story changed everything. When Apple acquired NeXT in 1997 and Jobs returned as CEO, he was a different operator. The products that followed (iMac, iPod, iPhone) reflected not just design genius but judgement: a sense of what mattered and what didn’t, what to build and what to kill, when to move and when to wait.

Judgement may be the defining economic asset of the next decade.

The Gig Economy Gave Us Freedom. AI Gives Us Capability.

The last decade saw technology transform how we work. The gig economy (Uber, Deliveroo, TaskRabbit) democratised access to work itself. Anyone with a car or a bicycle could earn. The trade-offs were real: flexibility came at the cost of security, autonomy at the cost of protection. But the fundamental nature of the work remained unchanged. Driving is driving. Delivering is delivering. Ten thousand hours on the platform didn’t make you meaningfully more valuable.

AI changes this equation. It doesn’t just give you access to work, it gives you access to capability. A person with Claude or GPT can now write code, analyse data, produce designs, and synthesise research at a level that previously required specialists. The barrier isn’t access anymore. It’s knowing what to build, what questions to ask, when the output is good enough, which opportunities matter.

Execution is becoming abundant. Judgement is becoming more valuable than ever.

When Everyone Has Superpowers, Taste Becomes the Differentiator

The economic logic is straightforward. When the price of something falls, its complements become more valuable. AI is collapsing the cost of prediction, analysis, and execution. The complement to all three is judgement, the human capacity to decide what’s worth doing, to evaluate whether it’s been done well, to course-correct when circumstances change.

AI can generate a hundred logo options, but someone has to know which one is right. It can draft a dozen strategies, but someone has to sense which will resonate. This is what Steve Jobs meant when he talked about taste: “the ability to expose yourself to the best things humans have done and then try to bring those things into what you are doing.”

The Judgement Economy

Call it the judgement economy. For the first time, the binding constraint on what we can produce isn’t capability, it’s the wisdom to know what’s worth producing, and the discernment to know when it’s been done well.

This isn’t a rhetorical flourish. It’s a structural shift. And we can see it most clearly where AI has advanced fastest: software development.

AI coding has progressed remarkably because software already has verification infrastructure. Compilers, linters, test suites, type checkers, these tools can tell you whether code works. When AI writes code, we can check it. The verification is encoded in the tooling itself. This is why developers can use AI constantly: the guardrails already exist.

Other domains don’t have this luxury. A generated legal brief can’t be run through a compiler. A medical diagnosis has no linter. A strategic recommendation has no test suite. In these domains, the verification still lives in human judgement. Decision traces (infrastructure that captures and validates AI reasoning) are an attempt to build guardrails for fields that lack them. But we’re early. The verification infrastructure that makes AI coding so productive simply doesn’t exist yet for most knowledge work.

And here’s the deeper point: even in software, with all its verification tools, you still need judgement about what to build and whether the output is right. A test suite can tell you whether code runs. It cannot tell you whether you’re building the right product. The less time you spend executing, the more time you must spend judging and evaluating. That’s the trade-off at the heart of the judgement economy.

The Founder Paradox

Here’s the uncomfortable truth about entrepreneurship: starting a company requires a certain suspension of judgement. Jensen Huang has said that if he’d known how hard it would be to build NVIDIA, he never would have started. Every founder ignores the odds. That’s what makes them founders.

But surviving as a founder requires the opposite. Once you’re in the arena, every decision is a judgement call. Who to hire. Who to fire. Which customers to prioritise. Which features to build. Who to raise money from. When to spend and when to preserve. The founders who succeed are the ones who develop judgement fast enough to compensate for the naivety that got them started.

We see this pattern constantly. First-time founders bring a raw energy that can be extraordinary, they’re unconstrained by legacy thinking, willing to try things that experienced operators would dismiss. But they also spend months reinventing wheels, wander down blind alleys, and too often scale prematurely. Flush with venture capital, they build teams before they’ve found product-market fit. The cash runs out before the learning compounds.

Experienced founders operate differently. Second and third-time entrepreneurs show an almost obsessive focus on product-market fit before scaling. They’re more judicious with capital. They sequence decisions more effectively. They know, from hard-earned experience, what matters and what doesn’t.

Jobs in 1985 was a first-time founder whose company had outgrown his judgement. Jobs in 1997 was something else entirely. The wilderness years weren’t wasted, they were where his judgement developed.

The Optimism: Judgement Is Learnable

The doom version of this story is simple: AI favours people who already have good judgement, and everyone else falls behind. The rich get richer. The talented accelerate. The gap widens.

But judgement isn’t fixed. Unlike raw intelligence or technical virtuosity, judgement develops through experience, feedback, pattern recognition, and, critically, failure. And AI is making all four dramatically more accessible.

Consider what it now costs to try something. A decade ago, building a software product required a team, capital, months of development. Today, a solo founder with AI tools can ship in weeks. The startup costs have collapsed, average small business formation now costs around $3,000. Solopreneurs contribute $1.7 trillion to the US economy. Solo-founded startups have risen from 22% to 38% of all new companies in less than a decade.

Each attempt is a judgement-development cycle. What did I build? Did it work? What would I do differently? When the cost of trying falls, the velocity of learning increases. The founder who builds and fails three times in two years may have started with less judgement than the careful planner, but all else equal, they’ll have acquired far more by the end. Experience is the raw material from which judgement is forged.

If judgement develops through practice, the question becomes: what kind of cultures produce people who are practiced at exercising it?

What Europe Does Well (And Could Do Better)

Judgement cannot be taught in a classroom. It must be exercised, repeatedly, with real stakes.

Some educational traditions understand this better than others. The Nordic educational philosophy of frihed under ansvar (freedom with responsibility)embeds judgement training from early childhood. Danish children commonly ride public transport alone at age 8. Finnish children cross main roads unaccompanied at the same age. It’s graduated autonomy: real decisions, real consequences. The stakes are genuinely high (a child can be hurt) but cultures that embrace this approach believe the developmental benefit justifies the risk.

The results track. Estonia produces 4.6 times the European average in startups per capita. Sweden, despite having among the highest failure-stigma in Europe, generates more unicorn founders per million inhabitants than almost anywhere else. The pattern suggests that the ability to exercise good judgement may matter more than comfort with failure.

But Europe’s challenge isn’t about which countries are “good” at developing judgement. It’s about what constrains judgement everywhere: educational systems that prioritise compliance over autonomy, failure stigma that prevents learning cycles, and attractive alternatives to entrepreneurship that absorb would-be founders into stable employment. These constraints exist in varying degrees across the continent, and they’re all addressable.

The opportunity is to recognise that Europe already has traditions of independence and rigour that develop judgement. The question is whether those traditions can be extended, and whether the people who develop good judgement are then given room to use it.

The View From Here

Two narratives dominate the current discourse. The AI doomers see mass unemployment, capability concentrated in machines, humans rendered redundant. The demography optimists see labour scarcity offsetting any displacement: as workers retire, the robots merely fill gaps.

Both miss the structural shift underneath.

AI is not simply replacing human work or filling demographic holes. It is changing the composition of what humans do. Execution (the tasks that could be specified, repeated, scaled) migrates to machines. Judgement (the capacity to decide, evaluate, course-correct) becomes the human contribution.

This is neither utopian nor dystopian. It is a transformation in which some people thrive and others struggle, depending on their ability to develop and exercise judgement. The optimistic case isn’t that everyone wins automatically. It’s that the path to developing judgement is more accessible than ever: more attempts possible, faster feedback loops, lower cost of failure.

Jobs spent twelve years in the wilderness before he understood what mattered. The judgement economy compresses that timeline. The founders building for it don’t need a decade of exile, but they do need the reps, the failures, the pattern recognition that only comes from doing the work. Our job is to find and back them.