Walk into any leadership offsite and you will still hear decisiveness celebrated as the executive virtue — the leader who makes the tough call, who decides under pressure. For a hundred years that instinct was correct. The organization existed to funnel information upward to a small number of people who made the consequential calls, and the scarcest resource in the building was good judgment at the top. That resource is being quietly distributed. Analytics pushed decisions outward years ago. Now networks of human and AI contributors can analyze, draft, execute, and even decide across a vast surface of work. What they cannot do — what no model and no network can do for itself — is determine which problem is worth solving and frame it well enough to be solved. That has become the scarce resource. And it is now the executive’s real job.

The Century of the Decisive Executive.

The organizational pyramid was a decision-routing machine. Information flowed up, decisions flowed down, and the people at the top were valuable precisely because they sat where the consequential choices were made. We built an entire culture of leadership around that architecture — the celebration of judgment, gut, decisiveness, the willingness to make the call when others hesitated. It was the right culture for the structure.

But the structure is changing underneath the culture. The decision itself — the act of choosing between options once they are framed — is no longer the bottleneck it once was. Decision support, analytics, and now AI networks can surface options, model tradeoffs, and execute choices at a speed and scale no executive committee can match. The organization no longer needs to funnel every consequential choice to a single apex node. What it needs is something the apex node is uniquely positioned to provide, and that the network cannot necessarily generate for itself.

What Just Became Scarce.

There’s a pattern that repeats every time something valuable becomes cheap and plentiful: the value doesn’t disappear — it moves to whatever is still rare. When computing power became cheap and abundant, the rare thing became human attention; suddenly every app on earth was fighting for your eyes. When information became free and infinite, the rare thing became knowing what actually matters; anyone can look something up, but knowing which facts deserve your trust became the skill. The pattern is reliable: abundance in one place creates scarcity right next to it.

It’s happening again. Getting work done — analyzed, drafted, decided, executed — is becoming cheap and abundant, because networks of people and AI working together can now produce finished work at remarkable speed. So what becomes rare and valuable? The thing that has to happen before any of that work starts: a clearly defined problem. The bottleneck is no longer “can we get this done?” It is “are we sure this is the right thing to do?”

Here’s why that matters so much. A network of specialized contributors — human and digital — is extraordinary at solving a well-framed problem. But point it at a vague one and it does something worse than fail. It produces fast, polished, confident, professional-looking work that solves the wrong problem. And because the output looks finished, nobody questions it until the damage is done — the error is expensive to find and even more expensive to undo.

The network amplifies whatever you hand it. Hand it clarity and it multiplies clarity. Hand it confusion and it multiplies confusion — at scale, at speed, and with total confidence.

The Blast Radius of a Badly Defined Problem.

Here is what changes the stakes. In the old model, a poorly framed problem handed to one analyst wasted one analyst’s week. The damage was contained by the size of the resource working on it. In the network model, a poorly framed problem is handed simultaneously to six agents, three analysts, and two managers across an initiative — and every one of them executes against the flawed framing in parallel. The waste is no longer one analyst’s week. It is the entire network’s capacity, spent at speed, producing a confident answer to a question nobody should have asked.

The Anatomy of a Well-Defined Problem.

If problem definition is the new core skill, it deserves to be treated as a discipline with components, not a vague gift some leaders have. A well-defined problem — one a network of humans and agents can actually solve together — contains six things.

Notice that not one of these six is a technical skill. They are judgment, diagnosis, organizational knowledge, and the ability to make implicit things explicit — the most human capabilities in the building, and the hardest to automate.

What It Looks Like in Practice.

The difference between a poorly defined and a well-defined problem is not a matter of length or polish. It is a matter of whether a network can act on it without guessing.

This Skill Is Not New.

The reassuring part of this shift is that problem definition is among the most studied disciplines in management — it has simply never been the headline. Design thinking puts “Define” at the center of its process and tells practitioners to fall in love with the problem, not the solution. Six Sigma’s DMAIC cycle begins with Define, because a misdefined problem corrupts everything downstream. Organizational development has always led with diagnosis — the action-research tradition insists you understand the system before you intervene in it. And Peter Drucker spent a career warning that the truly dangerous error in management is not the wrong answer but the wrong question, pursued efficiently.

What AI changes is not the existence of this skill but its position. Problem definition moves from a background virtue — the quiet competence of unusually effective leaders — to the central, measurable, developable capability that determines whether an organization’s growing network of human and digital contributors produces value or expensive noise.

What This Means for Every Seat at the Table.

This is not an AI-team concern. The shift from decision-making to problem definition lands on every executive role — and each one inherits a specific version of it.

Problem Definition Is Developable.

Because this is a discipline and not a gift, it can be assessed and built. The quality of the problems a leader frames is observable — in the rework rate of the teams and networks they direct, in how often work has to be redone because it solved the wrong thing, in whether their initiatives start with a diagnosis or a vague ambition. It can be developed through the traditions that already own it: design thinking, the OD diagnostic tradition, strategic-questioning practice, and structured framing methods. And it can be measured through the same evidence chain that governs the rest of the mixed workforce — from the problem defined, to the network’s output, to the business outcome, to the value created.

What to Do First.