No economist alive is quoted more than Daron Acemoglu. From his office at the Massachusetts Institute of Technology he produces papers at a rate that would exhaust most departments, ranges across subjects that would daunt most scholars, and in 2024 collected the Nobel prize for work on why some nations grow rich while others stay poor. If influence in economics could be measured, he would top the chart by a wide margin. And yet a curious feeling follows a close reading of his work, which is admiration for the craft mixed with doubt about the conclusion.

The doubt is not about his intelligence, which is formidable, nor his industry, which is superhuman. It is about a particular intellectual style, the habit of taking the largest questions human beings ask, why civilisations rise, how machines change lives, what makes a country free, and answering them with a model. The model is always elegant, the mathematics always clean, the empirical work always vast. The trouble is that the world these questions concern is messier than any equation can hold, and the gap shows.

The theory of everything

Consider the book that made his name beyond the profession. Written with James Robinson, it argued that the wealth of nations comes down to institutions, that countries prosper when their rules include the many and fail when they are captured by the few. It is a powerful idea, and often a true one. But pressed into service as an explanation for all of human history, it starts to strain, absorbing every case as either a confirmation or an exception, until the reader begins to suspect that a theory which explains everything may explain rather less than it claims.

The same instinct animates his later work with Simon Johnson on technology, which contends that the gains from innovation reach ordinary workers only under particular conditions, and otherwise flow upward to those who own the machines. Again the argument is valuable and, as a corrective to blind techno optimism, badly needed. Yet it too has the quality of a frame built first and fitted to the evidence after, capacious enough to accommodate almost any outcome, which is precisely what makes it hard to test.

The wager on artificial intelligence

Nowhere is the tension clearer than in his stance on artificial intelligence, where he has become the most prominent sceptic in the room. While Wall Street analysts forecast a transformation, Acemoglu has run the numbers and emerged with a strikingly modest figure, estimating that the technology will lift total factor productivity by only around half a percent over an entire decade, and that perhaps 5 percent of tasks will be worth automating in the near term. Against the euphoria this is a bracing splash of cold water, and it may well prove closer to the mark than the hype.

But the confidence of the estimate sits oddly with the uncertainty of the subject. To put a precise decimal on the productivity effect of a technology still in its infancy is to project an air of precision that the underlying knowledge cannot support. The number is derived rigorously from assumptions that are themselves guesses, and the rigor of the derivation can disguise the fragility of the inputs. One is left admiring the calculation while quietly wondering whether it measures the future or merely the premises fed into it.

Correct without being right

This is the heart of the puzzle. Acemoglu is very rarely wrong in the narrow sense. His logic holds, his data are real, his models do what he says they do. The critics who have gone after him tend to concede the correctness of each step and still come away unpersuaded by the whole, which is a strange and telling verdict. It suggests that being right about the world requires something beyond being correct within a framework, a feel for the parts of reality that refuse to be modelled, and that this is the very thing his method tends to leave out.

There is a cost to the style that goes beyond individual arguments. Because his work is so influential, its habits become the profession's habits, and a generation of economists learns that the way to address a vast human question is to reduce it to a tractable model and grind out a result. The approach yields publishable papers and clean stories, but it can crowd out the messier forms of understanding, the historical, the institutional, the frankly uncertain, that the biggest questions actually demand.

The value of the doubt

None of this is to dismiss him, which would be foolish. Acemoglu asks the right questions, more of them and more ambitiously than almost anyone, and even his less convincing answers are more interesting than most economists' certainties. His scepticism about artificial intelligence is a useful counterweight, his insistence that institutions and power shape prosperity is broadly correct, and the field is richer for his relentless output. The issue is one of calibration, of matching the confidence of the conclusion to the strength of what can really be known.

Perhaps that is the lesson buried in the odd experience of reading him. The most influential economist of the age is a reminder that influence and persuasion are not the same thing, and that the cleverest possible answer to an enormous question is sometimes less honest than admitting how much remains unknown. Acemoglu has built a magnificent machine for turning the world into equations. What his work keeps demonstrating, almost against its own intention, is how much of the world still slips through.