In the 1960s, Barbara Minto, the first woman hired into McKinsey's consulting ranks, worked out a method for organising a piece of thinking so that anyone could follow it. She called it the Pyramid Principle. Lead with the answer, group the supporting arguments, order them so each one earns the next.
It became the closest thing the professional world has to a grammar of reasoning. For the next fifty years it was taught in consultancies, banks and business schools, and the skill it describes acquired a name.
Structured thinking. The discipline of arranging the reasoning inside your head so that it holds together and travels to other people.
For half a century, being good at it was a career. The professional who could take a mess of information and produce a clear, ordered argument was valuable precisely because most people could not.
One letter, and almost the opposite meaning
There is a second phrase, one letter away, that points in nearly the opposite direction. Structural thinking.
Structured thinking looks inward, at the order of your own reasoning. Structural thinking looks outward, at the conditions your reasoning is being asked to operate inside. One asks whether your argument is well built. The other asks whether the situation you are arguing about is what you think it is.
An example makes the gap obvious. A professional cannot keep up, and reaches for structured thinking: they organise their tasks, rank their priorities, build a cleaner system for the week.
Structural thinking asks a different question. Not whether the tasks are in the right order, but whether three people are assigning work without any of them seeing what the others have already sent.
The first is a reasoning problem. The second is structural, and no amount of personal organisation will touch it.
Structural thinking is what it takes to answer the only question that matters when a working life stops going well: is it me, or is it here?
The bet that just got called
For fifty years the safe bet was to get good at the first one. That bet has just been called.
Large language models are, among other things, structured-thinking machines. Hand one a mess of information and it will lead with the answer, group the arguments, order them cleanly, and do it in seconds. The Pyramid Principle, executed on demand, at close to zero cost.
This is not a forecast. It is the current state of the tools. The single capability the professional world spent half a century teaching as its premium skill is now the thing the machine does most easily.
What the machine cannot see
What the machine cannot do is read the structure.
A model reasons on what it is given. But structural conditions are, almost by definition, the things that are hardest to see from the inside, and therefore hardest to hand over.
The priority order that was never made explicit. The work that appears in no plan. The way effort has quietly concentrated on the two people who never say no.
You cannot paste those into a prompt, because the whole difficulty is that you cannot see them clearly enough to describe them. Memory is mood-coloured and recency-biased, and the person in the middle of the problem is the worst-placed observer of it.
And even with a perfect description, the structural question is not a reasoning task. Which of these conditions is actually binding, what can be moved and what cannot, who holds the authority to move it: that is situated judgement. It depends on standing in the room, and the machine has no standing in any room.
This is the part worth being precise about. Reasoning is a layer, and it operates on whatever it is handed. Structure is the layer that decides what it gets handed: what the constraints are, who set them, which of them are real and which are merely unquestioned.
Artificial intelligence has the reasoning layer, comprehensively. It does not have the structural layer, and it is not obvious how it ever could, because that layer is not made of information. It is made of arrangements, and an arrangement can only be read from a position inside it.
The advice pointing the wrong way
Here is where it turns strange. Faced with machines that now out-reason us, the advice on offer to professionals is to reason harder. Sharpen your structured thinking. Optimise your workflow. Manage your time better.
We are training people to get better at the one thing the machine already does better, and calling their difficulty a personal shortcoming. Meanwhile the capability that is actually becoming scarce goes untaught: the ability to look at a working environment and see its structure, to tell a badly built argument from a badly built situation, to know that the problem is not always in the thinking.
Opinion: The Skill That Does Not Commoditise
Our position is that structural thinking is the professional capability of the coming decade, and that it has been hiding in plain sight, one letter away from a skill we have spent fifty years overvaluing.
Structured thinking was never wrong. It was simply never enough, and its insufficiency did not matter while it remained scarce. Now that it is abundant, what is left is the harder and more human work: reading the conditions, not just the argument, and seeing the structure that sits on top of the reasoning and quietly decides what the reasoning is worth.
The provocation we would leave open is this. If the environments people work in can defeat capable reasoning, and we have just built machines that reason better than any of us, then the last thing that makes a professional distinctly valuable is the ability to see what the machine cannot, which is the structure of the situation itself. Whether we begin to teach that deliberately, or keep handing people a better pyramid while the building goes on falling down, is now a choice rather than an accident.
Declaration of Generative AI and AI-assisted technologies in the writing process:
The author made use of Generative AI or AI-assisted technologies in the preparation of this post.
Sources
Barbara Minto, "The Minto Pyramid Principle: Logic in Writing, Thinking and Problem Solving," first published 1987, developed from her work at McKinsey and Company in the 1960s.
The contents of this article are for informational purposes only and do not constitute professional, legal, or financial advice.
.jpg)
