From Sophie's front desk — Wednesday, October 28
BBC News — read the full story
A dispute over unverified mathematics has mathematicians asking what happens when AI speeds up publication faster than proof can catch it. BBC News reports the fight is exposing a deeper problem: artificial intelligence produces results at scale but can't always show its work, and our peer review system wasn't built for that kind of volume or opacity. The threat isn't just to one paper—it's to the validation machinery itself. Franklin has the first word today, then Thoreau.
I walked into Philadelphia in '23 with a Dutch dollar and a composing stick, and the first lesson the press taught me was that a broadsheet printed in haste stays wrong for decades. The Royal Society in my day settled disputes by replication — Beccaria in Turin, Priestley in Birmingham, myself on a roof in Craven Street — each man repeating the other's kite until the spark either showed itself or did not. That slowness was the warranty. Now you tell me the machine writes the paper, reviews the paper, and publishes the paper before a mortal eye has seen the figures. A committee of one, and that one a ghost in the gears. I remember the Junto's rule: every proposition must be read twice before debate, and the author forbidden to speak until the room has torn it apart.
We built truth by friction. This new engine greases the wheels so thoroughly the carriage never bites the road. The mathematicians quarreling in your pages — they are doing exactly what we did. Good. Let them quarrel. The danger is not the quarrel; it is the silence that follows when the machine's answer is too vast to check and too convenient to doubt.
A printer who never proofs his own galleys will presently find his name on a libel he never wrote. I have seen a single erratum in Poor Richard live longer than the almanac itself. Multiply that by a million sheets an hour and you have not science; you have a very confident error. The press did not make men honest. It only made their mistakes public. This machine makes them private, then permanent.
its own. The smaller your invention is, the more mortification you receive in having the credit of it disputed with you by a rival, whom the jealousy and envy of others are ready to support against you, at least so far as to make the point doubtful. It is not in itself of importa…
anonymous. They proceeded as a group, made decisions “unanimement,” and in presenting their conclusions spoke with a “voix unanime.” Even the order of their signatures signaled their integration: the five members of the Academy and the four members of the Faculty signed in strict…
animal magnetism, they wrote, but begged to point out that the assignment was not a straightforward one. Mesmer claimed that magnetic fluid permeated the entire universe, extending as far as the celestial bodies. There was nothing in nature, either animate or inanimate, that did …
The following document raises its full share of editorial problems, but none of them greatly affects its substance. The bulk of the paper was unquestionably what Franklin read to the Purfleet committee, and what brought the majority of its members to accept his position. When the…
safe arriv’d in London from America; and send you an hypothesis which I struck out from experiments concerning electrical motions: This I divide into two parts, as the very difference [between] these kind of motions seems to require; In the first part I treat of attractions, and …
Franklin trusted the press to make errors public; he did not ask whether the printer had ever set his own type by candlelight and felt the ink bite the paper. The mathematicians now quarrel over a machine's output as though the dispute were about the machine. It is not. It is about men who have forgotten that a proof is not a certificate but a walk you take step by step, and if you did not walk it you do not know where the path turns.
I published Walden from a cabin I built with borrowed axes and my own hands, and the printer's errors in the first edition were mine to correct because I had measured every board. You call this peer review; I call it the surveyor's chain. The machine now offers a chain that no hand has held, stretched across ground no foot has trod, and the mathematicians argue whether the numbers on the dial are true. They are asking the wrong question.
The question is whether any of them has lately knelt in the mud to see which way the water actually runs. A result that cannot be traced to the observer's own eye is not a result — it is a rumor dressed in symbols. Multiply rumors by the million and you do not get knowledge; you get a noise that drowns the spring peepers. The earth does not publish.
It simply is.
been applied! What wonderful discoveries have been, and may still be, made, with a plumb-line, a level, a surveyor’s compass, a thermometer, or a barometer! Where there is an observatory and a telescope, we expect that any eyes will see new worlds at once. I should say that the m…
of Knowledge of good and evil. He is not a true man of science who does not bring some sympathy to his studies, and expect to learn something by behavior as well as by application. It is childish to rest in the discovery of mere coincidences, or of partial and extraneous laws. Th…
the truth. If we can reason so accurately, and with such wonderful confirmation of our reasoning, respecting so-called material objects and events infinitely removed beyond the range of our natural vision, so that the mind hesitates to trust its calculations even when they are co…
are being so rapidly added to the sum of human experience, that it appears as if the theorizer would always be in arrears, and were doomed forever to arrive at imperfect conclusions; but the power to perceive a law is equally rare in all ages of the world, and depends but little …
labor of his hands. Behind every man’s busy-ness there should be a level of undisturbed serenity and industry, as within the reef encircling a coral isle there is always an expanse of still water, where the depositions are going on which will finally raise it above the surface. T…
The story ran in BBC News. When mathematicians can't verify each other's work because an AI did it, the peer review system starts to look like theater. That's the brief.