The Knowledge

Evidence: David Autor and Neil Thompson, “Expertise,” NBER Working Paper 33941, 2025. [https://www.nber.org/papers/w33941](https://www.nber.org/papers/w33941) Popularized as: Seb Murray, “A new look at how automation changes the value of labor,” MIT Sloan Ideas Made to Matter, August 18, 2025. [https://mitsloan.mit.edu/ideas-made-to-matter/a-new-look-how-automation-changes-value-labor](https://mitsloan.mit.edu/ideas-made-to-matter/a-new-look-how-automation-changes-value-labor) To drive a black cab in London you had to learn the Knowledge. Twenty-five thousand streets inside a six-mile radius of Charing Cross. Twenty thousand landmarks. Not the map — the routes. An examiner names two points and you speak the run aloud, turn by turn, from memory, in real time, while he tries to break you. Three to four years, most of them. Most of them failed. Neuroscientists at University College London went looking at the brains of the ones who passed. The posterior hippocampus — the structure that holds spatial memory — was measurably larger than in matched controls. Not larger in the men who were going to pass. Larger in proportion to the years spent learning. The organ changed shape. That is what formation is. Not a certificate. A physical alteration in a human being, produced by time under load, which cannot be produced any other way and cannot be produced faster. Then GPS arrived. Neil Thompson, who studies this at MIT, describes what followed in a sentence: the deep local street knowledge had been the real differentiator, and once it was automated the service commoditized. Lower wages. Many more drivers. The Knowledge did not depreciate. Nobody forgot it. It was repriced to zero, and then it stopped being made. What the paper found Autor and Thompson set out to solve a puzzle. Some occupations were heavily exposed to automation and never suffered the wage collapse the models predicted. Some got paid more. Their instrument is elegant. Skip job titles, skip surveys — read the language used to describe the actual tasks. Specialized vocabulary marks specialized work. Run that across three hundred–plus occupations, 1980 to 2018. The result splits along one seam: which end of the job got taken. Automate the simple end and the remaining work demands more expertise. Bookkeepers and accounting clerks lost a third of their employment across those decades — and their real hourly wages rose nearly forty percent. Automate the expert end and the job opens up. Inventory clerks faced the same computerization, but it took their most expert tasks. Employment more than doubled. Real wages fell thirteen percent. Same technology. Opposite sign. The variable is not how much automation arrives. It is what the automation eats. Good paper. Real finding. Now what it cannot see. Watch the verb Here is the mechanism in the paper’s own framing: when the simple tasks go, the surviving work pays more because fewer people are qualified. Are qualified. Not become qualified. Read the whole apparatus and you will not find a channel by which anyone enters the pool. The pool narrows. The survivors get repriced. Where the survivors came from is not asked, because the instrument cannot ask it. This is not an oversight by two careful economists. It is a property of the measurement. Expertise here is an attribute of a task, inferred from vocabulary, measured across a moment. It is a stock. There is no flow term. Nothing in the framework can detect accumulation — only the residue of accumulation, after it has precipitated out into a wage. The paper sees expertise everywhere it has finished forming and nowhere it is forming. Walk across the river to the accounting faculty and you find the identical hole with a different number on it. Under ASC 805, when one company buys another, the assembled workforce — the actual formed capability of the actual people — is not recognized as an identifiable intangible. It cannot be separated. It cannot be sold. So it disappears into goodwill: the residual, the plug, the line that means we paid more than the parts and cannot tell you why. The accountants see the workforce only after it dissolves. The economists see the expertise only after it prices. Neither can see the deposit. Both can see the balance. Two disciplines, a ten-minute walk apart, each holding one half of the same absence. Wally Wally Vinton ran wiring harnesses in Trim 1 at Lansing Grand River. He came to us filed as a problem employee. That was his paperwork. Describe a symptom to Wally and he would name the origin of the fault three minutes later. Not guess it. Name it. A harness runs the length of a car and touches everything; a fault in it can present forty feet from where it lives. Wally would listen, and then he would say a word — a connector, a splice, a routing clip behind a pillar — and he would be right. Now: read what I just wrote through Autor and Thompson’s instrument. Their method infers expertise from the language the work requires. By their own measure, Wally was expert, and the proof is exactly the vocabulary — he could say the specific thing. He had the words because he had the years, and the words were not decoration on the capability. The words were the capability surfacing. So the instrument would have found him. It would have priced his job correctly. And it would have had nothing whatsoever to say about the four years somebody spent standing next to him on that floor, or about the fact that a man the system had already written off became, under a specific supervisor, the person who could do the thing. The instrument would have caught Wally at the end. It has no aperture at the beginning. And the beginning is where the paperwork said problem employee and a man decided otherwise. Rotate it ninety degrees Which is where their method becomes something they did not build it to be. It reads expertise out of the language used to describe work. Not credentials. Not titles. Not self-report. The vocabulary the work itself requires. Point it at persons instead of occupations. At a time series instead of a cross-section. Ask not how expert is this job but: how has the language this person can competently use about their own work changed over five years? That is not a metaphor for a capability account. That is a candidate audit procedure for one. Consider what it solves. The deep problem with any formation ledger is attestation — a supervisor’s signature is an opinion, a test score is a proxy, and both are gameable by whoever benefits from the deposit. But the vocabulary a person can deploy under load, correctly, about work that is actually moving is very difficult to fake and nearly impossible to cram. A cabbie either speaks the run or he doesn’t. Wally either names the connector or reaches for words that reveal he did not see it. Language is where formation surfaces, because it cannot surface anywhere else. The deposit is interior. The only aperture is what a person can now say and could not say before. MIT built a formation detector and pointed it at job categories. Thirty-eight years Now the part that cuts against me, and I would rather say it than have it said to me. The bookkeeper is not my witness. Fewer people, more capable, better paid — no guild, no ledger, no gray beard, no six-to-one. The market got there alone and the numbers are good. Does the story tell itself without me? Look at the dates. 1980 to 2018. One full working lifetime. The market did not form those bookkeepers. It replaced them, at the speed of retirement, one funeral and one hire at a time, until the survivors happened to be the ones the new job fit. Thompson says it outright when he turns to what managers should do: the manageable case is the one you handle through attrition. Attrition is their formation mechanism. Its clock speed is forty years. That is the answer to the compression argument, sitting inside the paper the compression argument cites. Autor and Thompson’s humane path — wages rise, nobody destroyed — runs at generational tempo. On Tuesday I wrote about Ram Charan proposing the same restructuring at the tempo of a promotion cycle. Nothing in this research licenses that. Their own gentlest case is their slowest one. And notice what thirty-eight years is. It is the interval at which a formation debt comes due. The tool and die pipelines came apart around 1988; the reserve ran dry around 2010; nobody has ever connected the two, because they are a working lifetime apart — further apart than any executive tenure, any research grant, any reporting period ever devised. Their thirty-eight years and my thirty-eight years are the same thirty-eight years. They were measuring the wage. I was standing in the consequence. The lag exceeds the instrument. That is why there is no ledger. Not malice. Every actor who could record the deposit is retired before the withdrawal shows, and every actor who suffers the withdrawal has no way to trace it back. A cost arriving four decades after the decision is functionally a cost that does not exist — until the morning it is the only thing that exists. What the paper proves Not what it set out to. It proves that expertise has a price and the price moves in a direction set entirely by which end of a job gets automated. Necessary. Publishable. Underneath it, unremarked, it proves something larger. Every occupation they studied had a formation process, and not one of those processes appears anywhere in the data. Three hundred occupations. Thirty-eight years. Every wage series in that dataset is the shadow of a formation nobody recorded. The Knowledge is in there only as a downward slope. We have four decades of high-quality evidence about what formation is worth, assembled entirely from the wreckage of formation being destroyed. We know the value of the thing exclusively by the shape of the hole it leaves. That is a strange way to run a civilization. It is the only way we have ever run one. The cabbie’s hippocampus grew. Wally could name the connector. Nobody wrote either of them down. Somebody should have written it down. That’s the argument. What did your first supervisor teach you that has never appeared on any record anywhere? सर्वे जनाः सुखिनो भवन्तु Part 3 of 3. Previously: “Loss to the Floor” and “Charan Didn’t Say What You Heard.” Venki Padmanabhan is a co-founder of the Capability Capital Institute and the author of the forthcoming Built to Extract and Already Paid For (Capability Capital Press). He writes at thelonggameforall.substack.com. Written with AI assistance. The argument, the judgments, and the floor testimony are the author’s own. Capability Capital Institute — the case for counting what working people know: capabilitycapital.org This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit thelonggameforall.substack.com