On Thursday, Terence Tao published a short piece on his blog titled “We’re gonna need a lot more mathematicians.” By Saturday morning it was on the front page of Hacker News with 324 points and 426 comments. That is the entire news peg: a title, and a very large number of people arguing about a title.
I’m taking the headline at face value, because the headline is doing all the work. If there were an itemized inventory behind it — job listings, an estimate of the gap — I would have led with the inventory instead. There isn’t one. That’s the story.
426 Comments, No Inventory
The interesting number in that thread is not 426. It’s zero: the count of published, credible estimates of how many additional mathematicians the current AI buildout actually requires. Not from the companies doing the hiring. Not from the federal agencies that underwrite graduate stipends. Not from the departments that would have to supply the people.
A shortage claim without a demand curve is not a forecast. It’s a mood — and the mood has been remarkably useful to everyone in the room. If you have spent two years telling donors your discipline is more essential than ever, a headline like Thursday’s is manna. If you run a research organization telling investors that the hardest problems are finally tractable, a mathematician shortage is a lovely way to explain why the timeline keeps slipping. Neither party has any incentive to publish the actual figure, because the actual figure is almost certainly smaller than the rhetoric and larger than the current budget line.
Judgment Got Scarce. The Degree Didn’t.
Here is where the reflexive conservative reading goes wrong. The standard move on the right is to treat “we need more mathematicians” as a plea for more subsidized graduate students — a jobs program for people who write letters to the editor of a discipline’s newsletter. That’s half right. The other half is that the scarce good in this market is not mathematical labor. It’s mathematical judgment: knowing which problem matters, and knowing whether the thing the machine produced is a proof or merely a very confident paragraph.
That skill got more valuable, not less. Which means you cannot simultaneously believe that AI has eaten mathematics and that a math degree is a dead end. Pick one. The uncomfortable truth is narrower and points in two directions at once: the ability went up in price, and the credential went down. Those are different assets. Everyone in this argument keeps talking as though they were the same one.
“We’re not short of people who can do the work,” a postdoc in applied math told me, waiting out a delayed flight at SFO after a conference let out early. “We’re short of people who can tell whether the work is right — and we pay those people like students while everyone else bills them out like consultants.”
She has a point that costs money to act on.
The Doctorate Is Priced Off a Scarcity That Just Left
The academic pipeline isn’t priced off the value of mathematics. It’s priced off the scarcity of mathematical labor. That is the whole economic logic of the low-paid apprenticeship: you accept the thin stipend years because the alternative — walking into industry without the credential — used to be worse. When the field’s scarce resource was people who could do the work, hoarding them in graduate programs at subsistence wages was rational for the institution and tolerable for the student.
If judgment can now be sold by the week rather than certified once by the decade, the rational move for a good 26-year-old is to skip the dissertation and invoice somebody. That is a genuine crisis for departments, and it will not be solved by declaring that we need more mathematicians. It gets solved by paying the ones we have — a hard thing to say in a profession whose prestige system, from the arXiv to the tenure track, is built on the premise that the work is its own reward.
The people who need to hear “we need more” are the ones who would have to fund the more. They are currently reading the same headline as a reason to hire fewer full-time faculty and more contractors.
Whoever Publishes the Number Wins
So watch what happens next, because the fight over this headline isn’t about mathematics. It’s about which interest group gets to define the shortage.
Departments want it read as a shortage of graduates, which is a block-grant argument. Research organizations want it read as a shortage of rare geniuses, which is a hiring-leverage argument and a handy reason to keep the compensation conversation small. Both framings are self-serving, and both are probably wrong about the magnitude.
The right’s error is treating the headline as a punchline about useless degrees — a joke that will look silly if verification work turns out to be the best-paid technical job of the decade. The academy’s error is treating it as an appropriations request for the one part of the pipeline the technology has just made optional. If your answer to a demand shock is “fund the middlemen harder,” you haven’t understood the shock.
Four hundred and twenty-six people spent a weekend arguing about whether the profession has a future. Meanwhile the actual number — the one figure that would settle it, that would tell a 19-year-old whether to take the scholarship and a department whether to keep the seminar — remains unpublished. It will stay unpublished. It is the only fact in this entire discussion that nobody in a position to know benefits from sharing.