The New Yorker – no paywall: “…The focal point of the department was the checking library, which contained reference books such as Who’s Who in the People’s Republic of China, Debrett’s Peerage & Baronetage, and the Physicians’ Desk Reference for Herbal Medicines. (New checkers are advised that you can’t trust books—they tend not to be fact-checked. But reference works help, and endnotes are a gold mine.) The library had another relic—a metal Rolodex that Calvin Trillin has said belongs in the Smithsonian. (Under “C”: “Chomsky,” “Cher (actress),” “Congo,” “Cold Fusion.”) Every Friday, the department held a meeting in the library, where checkers discussed thorny stories and bitched about difficult writers and editors. There was a smaller library, for even more books; checkers, on especially tight deadlines, would spend the night on a cushion on the floor. Colleagues would talk for hours with the powerful and the secretive; a conversation with Julian Assange required technological methods that we were not permitted to discuss, to discourage eavesdropping. Yasmine AlSayyad got propositioned by Islamic militants. Fergus McIntosh, the department’s current head, got book recommendations from a Supreme Court Justice. Danyoung Kim would come to work in an astronaut costume, sit down, and call up Harry Reid. We probably took our jobs too seriously. This was the first Trump Administration, and the work felt urgent but doable. We talked to Cabinet members and to neo-Nazis. We’d sometimes get threatened, and that only inflated our self-importance. We were, as the writer and former checker David Kirkpatrick has put it, “intoxicated by our own busyness.” The writer had already engaged in the charm and betrayal inherent in reporting. We were in the harm-reduction business….

Checkers talk to virtually all sources in a piece, named and unnamed. They also contact people who are mentioned, even glancingly, whom the writer didn’t already speak to, and many people not mentioned in the piece at all. Checkers don’t read out quotes or seek approval. Sources can’t make changes. They can flag errors, provide context and evidence. The checker then discusses the points of contention with the writer and the editor. It’s an intentionally adversarial process, like a court proceeding. You want to see every side’s best case. The editor makes the final call. In a sense, the checker is re-reporting a piece, probing for weak spots, reaching a hand across the gulf of misunderstanding. The checker also asks questions that, in any other situation, might prompt the respondent to wonder if she was experiencing a brain aneurysm. “Does the Swedish Chef have a unibrow?” “He actually has two separate eyebrows that come close together above his nose.” Could a peccary chase a human up a tree? Certainly if it’s a white-lipped peccary, which is the size of a small bear and prone to stampede. Zadie Smith once received a call regarding whether, years earlier, at Ian McEwan’s birthday party, a butterfly landed on her knee. When a Talk piece by Tad Friend described the singer Art Garfunkel waving his arms around, the checker asked Garfunkel to confirm that he had two arms…”