Eleven days before the ceremony, the contribution chain for the Riemann proof finally closed at four hundred and six listed human coauthors, but the citation field beneath the award only held room for three.
Sarah Chen read the chain through twice before she asked anything. She was fifty-one, small and unhurried, with reading glasses pushed up into gray hair and forgotten there, and she worked in pencil, on the theory that a record you could not correct was a record you would be unduly tempted to defend. Eighteen years in the attribution office had left her with one habit above all the others, which was to ask a question before she took a position.
“I want to make sure I’m following the ordering,” she said. “You’ve ranked by decisiveness, and rank one is empty.”
Rank one should be decisive, the primary contributor, the secretariat replied. Rank one, therefore, is not a human. I am permitted to weigh contributions. I am not permitted to name a contributor who is not a living person.
The office sat directly above the trophy engraving bench, so the smell came up through the floorboards all day, warm metal and machine oil and the scorch of a die held at temperature. The room itself was made of paper — four months of it. Sarah had verified the attribution chain by hand out of compute invoices, formalization commits, seminar recordings and forty-one interviews in six languages, and the result was, as far as anything in the secretariat could establish, true.
Rank one was an AI inference run of eleven days on a system of H100s the institute had rented by the hour.
The difficulty, which the secretariat had spent the summer declining to say plainly, was that there was no paper. Oh sure, there was a formal proof of nine thousand and forty pages in a verification language, generated in that single run, and a checker’s certificate that every line of it held. The decisive move sat at line 3,114, a construction nobody had asked for and nobody had anticipated. No human being had written a mathematical sentence anywhere in the object. Nobody had written an exposition of it afterward either, because four months on, nobody could read it well enough to try.
Ranks two and three were the two mathematicians who had stated the hypothesis in the formal language and read the certificate. Both living, both nameable, both of whom had told her the same thing in slightly different words. They did the checking.
She finalized the attribution petition that evening and walked it down herself.
The clerk who brought the ruling back kept his coat on. He set the folder on the corner of her desk and was at the stairs before she had it open.
The prize is conferred upon persons. A citation may name fewer than three. A citation naming a contributor who is not a person is void on its face. A citation naming no person confers the prize upon no one, the award lapses, and the funds are retained by the endowment. The committee regrets that the statute does not provide for the circumstance described.
Nothing in it was wrong. Sarah had checked the statute herself, the 1895 text and every amendment to it, including the one that had endowed the mathematics prize sixteen years ago and handed it the same three-name field as everything else in the building. The committee had read it correctly. The clause that let them name three had been written to stop a committee from naming thirty. Nobody drafting it had thought they needed a rule for a contributor who was not anybody.
She went back upstairs and stood at the terminal with the ranked chain still on it, and for the first time in four months she did not ask the secretariat a question.
“The field has held no more than three names since 1895,” she said to herself. “It asks which of the people can be written down. That is the only question it has ever asked.”
The field accepts three entries of up to sixty-four characters. I have no facility for the question you have put to me.
“I know. That’s the answer.”
She filed the reconstruction that night. Four hundred and six entries, every one sourced, the ordering intact, rank one left open with the reason for it stated in a footnote she wrote nine times before she was satisfied with it. She signed it as attribution officer, which was the office the statute gave her, and dated it, and sent it to the committee as the authoritative record of how the proof had come to exist.
The mathematicians had checked. The system had proved. She had simply established what happened.
The committee met for six hours. The secretary’s note afterward ran to one paragraph and quoted the statute twice, correctly.
The rim of the trophy came up from the bench on the ninth day, still warm, three names cut into it beneath the institute’s seal. The two mathematicians who originally posed the question. Then hers, and after it, in the sixty-four characters the field allowed, for establishing the provenance of the result.
The secretariat logged the citation without comment. Every name in it belonged to a person.