On the Insufficiency of Proof

I had proposed a restaurant in Adams Morgan but Didion wanted somewhere quieter, and Cusk had not responded to the question at all, which I took as indifference. So we met at a rented apartment near Dupont Circle that belonged to a friend of mine who was abroad — a one-bedroom with too many books and a kitchen table that seated three if no one needed to push back their chair. Cusk arrived first. She stood in the living room looking at the bookshelves with the expression of someone conducting an inventory she had not asked to conduct. She did not touch any of the books. She picked up a framed photograph from the side table — a woman and two children at what looked like a lake — studied it, and set it back without comment.

Didion came twenty minutes late and did not apologize. She was carrying a canvas tote bag that said STRAND on it, which seemed like a choice, and she set it on the kitchen counter and looked around the apartment the way she might look at a crime scene — not with suspicion, exactly, but with the understanding that the arrangement of objects would tell her more than anyone’s account of them.

“Whose apartment is this?” she asked.

I told her. She nodded once and sat at the table. Cusk was already sitting. I poured water from a Brita pitcher that was half full and slightly warm.

I said I wanted to write an essay about a fact-checker at a major American magazine who spends a week verifying a story about a political conspiracy theory. She can prove every claim false. She has documents, sources, institutional records. The facts are on her side, entirely. But during the process, she begins making a private inventory — moments in her own life when she was absolutely certain of something and turned out to be wrong.

Cusk said: “What kind of wrong?”

I said I meant ordinary wrong. Not dramatic. She was certain her sister had said something cruel at a Thanksgiving dinner, and years later found a recording — someone had been filming — and her sister had said something different. Not kind, but different. She was certain a man she’d dated had lied about where he went to school, and he hadn’t. She remembered a street in Philadelphia as running east-west, and it ran north-south.

“Those aren’t equivalent errors,” Cusk said.

“No.”

“The street is trivial. The sister is not. And the man — the man is interesting because she needed him to have lied. The conviction that he had lied was doing something for her. Certainty is functional. People are certain of things because the certainty organizes their experience.”

Didion was watching Cusk with what I can only describe as recognition. Not agreement — something prior to agreement. The recognition of a shared jurisdiction.

“That’s the essay,” Didion said. “Not the fact-checker disproving a conspiracy theory. The other thing. The personal inventory. The list of certainties that failed.”

“It’s both,” I said. “The essay moves between the professional fact-check and the personal one. The two modes contaminate each other.”

“How?” Didion asked.

I said I wasn’t sure yet, which was why I’d asked them here. The professional fact-check is clean. She has procedures, sources, verification protocols. The personal inventory is messy. There are no sources for the feeling that your sister meant to hurt you. You can’t call a second expert to confirm that the man you dated was lying. And the essay lives in the space where the clean process begins to feel fraudulent — not wrong, but insufficient — because she is simultaneously discovering how unreliable she is about the things she knows best.

Cusk was quiet for a long time. She picked up her glass of water and held it without drinking. “A friend of mine once sent me a text,” she said. “Late at night. I don’t remember the context — some argument we’d been having about memory, about whether you could trust your own account of your childhood. She sent me a list. ‘My articles of faith,’ she called them. Things she had been certain about that turned out to be false. Seven items. The first was factual — she had been certain her father had served in the military, and he had not. The last was interpretive — she had been certain her marriage ended because of her husband’s affair, and twenty years later she understood that the affair was a consequence of something she had done much earlier, something she could not name precisely, only that she had withdrawn from the marriage before he had.”

“Did she say this changed anything?” Didion asked.

“She said it changed nothing. That was the point. She had been wrong about these things and she had eventually admitted it, quietly, without apology, and moved on to the next certainty. She said: ‘The difference between us and them is that we eventually admit it. They double down forever.’ She was talking about the conspiratorial mind, I think, though she didn’t use that word.”

“And was she right about that distinction?” I asked.

Cusk set the glass down. “She believed she was right about it. Which is itself an article of faith.”

Didion smiled — not warmly, but with a sharpness that meant she had found the structural problem. “The essay collapses if the narrator claims that her kind of wrongness is different from the conspiracy theorist’s kind of wrongness. It’s a tempting claim. It’s probably true. But the essay cannot make it, because the essay is about the unreliability of certainty, and the moment the narrator says ‘I am different because I admit my errors’ she has produced a new certainty, and the reader should feel the floor shift.”

“The floor should shift several times,” Cusk said. “Not once. Not as a twist. It should be a condition of the essay that the ground is not solid. The narrator should establish her authority — she is good at her job, she knows how to check a fact — and then the authority should erode, and then she should try to rebuild it, and the rebuilding should be partial and unconvincing, and the reader should understand that the narrator knows it is partial and unconvincing, and does it anyway, because what else is there.”

I said that sounded like Cusk’s own method — the narrator who reveals herself through what she notices in others. In Outline, the narrator listens to other people’s stories and her own shape emerges in negative space. This essay’s narrator would be doing something similar: listening to the conspiracy theory, listening to her own inventory of errors, and her shape would emerge from the gap between the two listening acts.

“Don’t overstructure it,” Cusk said. “The gap is not a thesis. The gap is a sensation. The reader should feel it in the body — a mild vertigo, the feeling that the next sentence might not hold your weight.”

“I wrote about this,” Didion said. “In The White Album. ‘We tell ourselves stories in order to live.’ That sentence has been quoted so often it has become a bumper sticker, which means people have stopped reading it. The sentence is not aspirational. It is diagnostic. We tell ourselves stories in order to live, and the stories are failing, and we keep telling them anyway. The fact-checker’s procedures are stories. The verification protocol — call two sources, cross-reference with the public record, check the timestamps — that’s a narrative. It is a narrative about how truth is established. And the essay should let the reader feel how thin that narrative is. Not false. Thin.”

“Thin is the right word,” Cusk said. “Not wrong. Not broken. Thin. Like paper you can see through when you hold it up to light. The procedures work. The fact-check is valid. The conspiracy theory is false. All of that is true. And at the same time, the narrator is sitting in a room at eleven at night with printouts and phone records and she is thinking about the time she was certain her mother had forgotten her birthday and her mother had not forgotten it but had chosen not to mention it, which was different, and the certainty she had carried for a decade — that her mother forgot — had organized a whole wing of her emotional life, and when the certainty collapsed, the wing did not collapse with it. It was still there. Furnished. Inhabited.”

I said: “So the essay’s argument is that certainty is structural, not evidentiary. We don’t believe things because of the evidence. We believe things because the belief holds up other beliefs, and the cost of being wrong is not the wrongness itself but the renovation.”

“The essay should not make that argument,” Didion said. “The essay should create conditions in which the reader arrives at something like that argument and then mistrusts it.”

“This is where we disagree,” Cusk said. She said it flatly, without emphasis, the way she might report the weather. “You want the essay to perform epistemological crisis as a kind of style. A texture. The sentences cool and precise, the narrator’s voice held steady while everything underneath shifts. That’s your method and it works — no one does it better. But I think there’s another way, which is to let the narrator be genuinely lost. Not performing lostness. Lost. Not sure, sentence by sentence, whether the next thing she says will be true. Writing toward a certainty she doesn’t reach.”

“A narrator who is genuinely lost is fiction,” Didion said. “An essay narrator is a construction. She may present herself as lost, but the presenting is an act of order. The sentences are written and revised and placed in sequence. The lostness is composed.”

“Everything is composed. The question is whether the composition admits it or not.”

“My compositions admit everything. That’s their entire method.”

Cusk turned to me. “What does the fact-checker do with her list? The personal inventory. Does she write it down?”

I said I imagined her keeping it on her phone. A note. Adding to it during the week, between calls to sources and checks of public records. Seven items, then eleven, then fifteen. Each one a small collapse. Not dramatic — the drama is in the accumulation, in the way the list grows longer while the professional fact-check grows more conclusive. By the end of the week, she has definitively debunked the conspiracy theory and she has also compiled a document that proves she cannot trust her own memory, her own judgment, her own sense of what happened in rooms she was actually in.

“The phone matters,” Didion said. “The same device holds both documents. The professional fact-check and the personal inventory. They live in adjacent apps. The phone does not distinguish between them. The phone treats both as data.”

“The phone is correct,” Cusk said. “Both are data. That is what the narrator discovers, and it horrifies her, and the horror is the essay.”

“Is she horrified?” I asked.

They both looked at me, and for a moment I felt what the essay’s reader should feel — the sudden uncertainty about whether I had asked the right question.

“She should not know whether she is horrified,” Cusk said. “That is more interesting than horror. A narrator who has arrived at a feeling and identified it is a narrator who is finished. A narrator who is between feelings — who senses something shifting in her chest and is not yet sure whether it is fear or excitement or nausea — that narrator is still moving, and the essay moves with her.”

“I agree with that,” Didion said, “though I would have said it differently. I would have said: the narrator should be precise about external things — the documents, the timestamps, the weather, what the source said on the phone at 2:15 p.m. — and imprecise about internal things. Not imprecise as a failure. Imprecise as an honest accounting. Because we are all imprecise about what we feel. We are exact about the time and approximate about the grief. The essay should replicate that asymmetry.”

I brought up something from the pitch material — an image of the narrator’s friend screenshotting location data, timestamps, weather reports. “All these little anchors to reality that might mean something or nothing.” The friend collecting verifiable facts as though enough of them would add up to certainty.

“That’s compulsive,” Didion said. “Not rational. Compulsive. The collection of anchors is a response to vertigo. When the ground moves, you grab whatever is bolted down. Timestamps are bolted down. Weather is bolted down. Location data — where you actually were at 3:47 on a Tuesday — that is the last reliable thing. And even that is only reliable if you trust the device, which is itself a faith position.”

“Everything is a faith position,” Cusk said. “That’s what the friend’s text was saying, underneath the humor. Articles of faith. She was a nonbeliever making a confession. She was admitting that her certainties had been exactly that — beliefs, not knowledge — and the difference between her and the conspiracy theorist was not epistemological but temperamental. She could tolerate being wrong. She could tolerate the renovation. They could not.”

“Can she, though?” I asked. “Can anyone?”

Cusk looked at the photograph of the woman and the children at the lake. “There’s a question in the essay that should never be answered, and that’s it. Can she tolerate it. The narrator wants to believe she can — that she is the kind of person who holds her certainties lightly, who can be shown evidence and change her mind. And the essay’s job is to let the reader wonder whether that self-image is the narrator’s most protected certainty of all. The one she will never fact-check.”

The room was quiet. Outside, someone was having a conversation on the sidewalk — I could hear the cadence but not the words, which seemed right for this essay, where so much would be about partial reception, about hearing enough to believe you understood.

Didion said: “I want to talk about the ending. Or rather, I want to establish that there should not be one. The essay should not arrive at a position on certainty. It should not conclude that the narrator is unreliable, or that the fact-checking process is adequate, or that the conspiracy theorists are different in kind from the rest of us. It should end in the middle of the week. A Tuesday or a Wednesday. She is on the phone with a source who is confirming something she already knows to be true, and she is adding another item to her private list, and the two acts are simultaneous, and the essay stops.”

“It should stop the way a recording stops,” Cusk said. “Not because the music is over but because someone pressed a button.”

“Or because the tape ran out.”

“Tape doesn’t run out anymore. We have infinite storage. That’s part of the problem.”

Didion made a sound that might have been a laugh. “The personal essay is a form that assumes limited storage. A particular consciousness, a particular week, a particular set of observations that cohere only because a human being is holding them together. The essay’s shape is the narrator’s shape. Remove the narrator and you have data — timestamps, location pins, weather reports. The narrator is the thing that makes the data into experience. And this narrator is discovering that her own shaping — her own coherence-making — is suspect.”

“Which is the condition of every narrator,” Cusk said. “Which is why the essay is not special. Which is why it works.”

They disagreed about tone, after that. Didion wanted the prose to be controlled throughout — the narrator’s composure cracking only in what she chose to observe, not in how she observed it. Sentences that were formally intact while describing the dissolution of certainty. She cited her own work with a lack of embarrassment that I found admirable: “In The White Album I wrote about having a psychiatric diagnosis while covering the Manson trial. The prose did not become disordered because I was disordered. The prose was the order I imposed on the disorder, and the reader could feel the effort, and the effort was the meaning.”

Cusk wanted something different. “The prose should be austere, yes. But the austerity should be a form of exposure, not a form of control. Stripped-back sentences. No ornament. The narrator says what she sees and does not interpret it and the absence of interpretation is itself a kind of confession — an admission that she does not trust her own commentary. She observes. She notes. She resists the habit of explanation. And the reader feels the resistance as tension, as though the narrator is holding something back, and what she is holding back is the very certainty the essay is supposed to be about.”

I said both of those descriptions sounded right to me and could probably coexist in the same essay.

“They cannot coexist,” Cusk said.

“They do coexist,” Didion said. “In every essay I have ever written. The control and the exposure are the same gesture.”

“They are not the same gesture. They are opposite gestures that sometimes produce similar sentences.”

Neither of them was willing to give ground on this point, and I understood that the essay would have to find its own position between them — or, more accurately, would have to be written from a position that neither of them would fully endorse, which is probably the only honest place to write from.

The light in the apartment had changed. It was the kind of late-afternoon Washington light that turns everything slightly amber, and I was aware of how much the room’s appearance had shifted since we’d arrived — not because anything had moved, but because the angle of illumination was different, and the same objects looked like different objects, and this was the sort of observation that would mean too much in the essay if I let it, so I would have to be careful, or possibly I would have to not be careful and let it mean what it meant and trust the reader to