A historian is writing about Thomas Cromwell, and the documentary record runs out at a moment where something clearly happened that the sources don't describe. What does she do? She doesn't make it up. She doesn't skip past the gap as if it weren't there. She writes the sentence MacCulloch literally enacts in the interview: "We may think that it is quite likely the overwhelming sense of what this is is..."1
Read that sentence aloud. It's clumsy. It has hedges stacked on hedges. May think and quite likely and overwhelming sense and what this is is. A copy editor would cut it in half. But the clumsiness is doing the work. Every hedge is marking a degree of confidence in the writing itself, so the reader can tell — at the sentence level — what the historian is sure of, what she suspects, what she's reasoning toward, and what she's inferring from absence. The reader gets the calibration alongside the claim. That is the "may have" discipline, and it is the historian's signature epistemic move.
The sentence is a piece of evidential apparatus disguised as prose. It tells the reader four things at once:
Compare to what a novelist or a confident popular writer does. They write the gap as if it were known. "Cromwell entered the room, saw Wolsey's face, and felt the old loyalty surge through him." That's a sentence with no calibration. The reader has no way to tell what's documented from what's filled in. The novelist (rightly, for fiction) doesn't owe the calibration. The historian (necessarily, for history) does. The "may have" syntax is what owes it.
MacCulloch's clearest articulation of the discipline comes through his contrast with Hilary Mantel. Mantel wrote the Wolf Hall trilogy about Cromwell. MacCulloch wrote the standard scholarly biography. They were "doing the same project," but one was a novelist and one was a historian, and the difference shows up exactly at the "may have" line.
"She was liberated from that," MacCulloch says, "because if novelists never have to say, and it sounds odd if they do, they never have to say may have. They can simply present the story and fill in the gaps."1 The novelist's freedom is exactly the discipline the historian gives up. Mantel doesn't have to mark her inferences as inferences; she can present them as scene. That's what makes the prose work as fiction.
But Mantel's gap-filling is sometimes wrong, and MacCulloch is precise about what wrong means here. He gives the Antwerp romance as the example. We know — from documentary evidence — that Cromwell had an illegitimate child. We don't know where it came from, who the mother was, what the circumstances were. Mantel constructed an Antwerp romance to account for the child. "And that isn't right. It simply isn't right."1 Not because the romance is implausible, but because there is no evidence it happened, and presenting it as scene without a "may have" pretends to a knowledge no one has.
What's striking is what Mantel did with this. She sent MacCulloch the typescript of the third volume of the trilogy. "Can you look at this and have a look at the the historicity of it? Can you point out the mistakes? Not the fictions, the mistakes."1 The distinction is the entire point. Mantel knew the Antwerp romance was fiction. She was happy to keep it. What she wanted MacCulloch to flag was places where she'd gotten the factual substrate wrong — Abbot vs. Prior of St. Bartholomew's, the kind of detail that's not a novelistic choice but a research error. MacCulloch did exactly that. "I'm not trying to change your story. All I'm saying is that actually it was not the Abbot of St. Bartholomew's here, it was the Prior of St. Bartholomew's. You want to get that right, won't you?"1 She did.
The Mantel collaboration reveals the line between the two disciplines clearly. The novelist is permitted to fill gaps with invented scene; the historian is not. But the novelist still owes the background — the factual substrate the invented scene sits inside. Mantel's discipline was to make the invented foreground rich while keeping the historical background accurate. MacCulloch's job was to police the background. Each was running a different "may have" calculus.
The discipline isn't just defensive — it also produces. The clearest example is the insight MacCulloch credits Mantel with helping him pursue: that Cromwell adored Wolsey and detested Anne Boleyn, despite Cromwell and Boleyn being usually presented as Protestant allies.
The standard story, going back to the great historians of the past, was simple. Cromwell and Boleyn were both Protestant reformers. They were on the same side. They worked together to bring about the break with Rome and the English Reformation. That's the heroic version, and it has the satisfying clean alignment of ideology and alliance.1
What MacCulloch — and Mantel, in her novelistic reading — recognized was that the standard story collapses two different things. Ideological alignment and personal loyalty are not the same vector. Cromwell was a Protestant reformer; Boleyn was a Protestant reformer; but Cromwell was besotted with Cardinal Wolsey, his first great employer, who had been brought down partly by Anne Boleyn's pressure for the marriage that Wolsey was slowing up. Cromwell adored Wolsey. He detested Boleyn for what she had done to Wolsey. The Protestant alignment was real, but it sat underneath a personal loyalty-and-grievance structure that ran the opposite direction.1
What MacCulloch says next is the principle worth filing: "we forget that people can share an ideology while still detesting each other. Because something may be more important."1
This is the substantive payoff of running "may have" discipline. If you assume that shared ideology produces alliance — that the inference from one to the other is automatic — you reconstruct the past wrongly because you've collapsed two variables into one. The discipline of marking inferences as inferences means you don't quietly fill in alliance when only ideology is documented. You write: "they shared a Protestant program, though it may be that personal loyalties pulled in different directions." The "may have" preserves the space in which the actual relationship can be reconstructed from the evidence, rather than presumed from the ideology.
In the Cromwell-Boleyn case, "Hilary nerved me to pursue it," MacCulloch says.1 Mantel's novelistic reading — her character-level sense of how Cromwell felt about Boleyn — gave MacCulloch the courage to dig through the sources looking for documentary confirmation of what the standard historians had collapsed. The novelist's intuition served as a hypothesis-generator the historian could then verify with primary evidence. That's a methodological insight worth holding onto: the "may have" discipline is what lets the historian honor the novelist's intuition as hypothesis rather than dismiss it as fiction. Without the discipline, the historian has only two options — adopt the intuition uncritically (sliding toward fiction) or reject it on principle (losing the lead). With the discipline, she has a third: treat it as a hypothesis worth checking, and write the resulting claim with the appropriate "may have" if the evidence supports it.
The discipline is operational, not philosophical. The historian writes hedges that match the evidence. The vocabulary roughly maps to confidence levels:
A reader running through MacCulloch's prose can use the hedge vocabulary as a confidence dial. The "may haves" are not authorial weakness; they're calibration. The reader who learns to read them gains access to the historian's epistemic state, not just to the historian's conclusions.
This is the operational difference between disciplined history and confident popular history. Confident popular history uses the same vocabulary — "Cromwell evidently," "appears to have," "we may think" — but uses it as rhetorical garnish, not as calibration. The phrases mean the same thing whether the evidence is strong or weak. The disciplined historian uses the phrases to map the evidence's strength onto the sentence. The reader who can tell the two apart is reading historiography rather than entertainment.
The discipline reads as a writing-style convention, but it makes a substantive claim about what historical knowledge is. The "may have" syntax presumes that historical claims have gradations of certainty and that the gradations can be communicated in prose. Both presumptions are real. The first is widely accepted in epistemology; the second is craft.
Tension: the discipline can be performed without being run. A historian can sprinkle "may have" through the prose as decoration while still essentially asserting confident reconstructions. The phrases lose their calibration function and become noise. MacCulloch's discipline requires the hedges to match the actual evidential state, which requires the historian to have done the evidential work first. The phrases without the work are worse than no phrases — they're false calibration.
A second tension: heavy hedging slows the prose. Readers trained on confident popular history experience disciplined "may have" prose as evasive or wishy-washy. The historian has to choose between the rhetorical cost (slower, less satisfying prose) and the epistemic cost (overstated confidence). MacCulloch chooses the rhetorical cost. Not all historians do; the trade is real.
A third tension lives in the Mantel relationship itself. If novelistic gap-filling sometimes produces character-readings that the historian then confirms by archival work, then fiction is sometimes the leading edge of historiographic insight. That sits uneasily with the discipline. It implies that historians without access to novelists' intuitions are systematically missing certain hypotheses. What's the structural workaround? Cross-discipline conversation, evidently — but the workaround relies on personal contact and is not formally part of the discipline.
Open questions:
Sit down to make a Tetlock-style forecast and you'll feel the same discipline operating in a different medium. You attach a probability — 35%, 60%, whatever — to your prediction. The number is the calibration. It tells your future self and other readers exactly how much weight you put on the claim. MacCulloch's "may have" does the same work in prose. The hedges aren't decoration; they're calibration showing up as syntax. Tetlock uses numbers because his domain admits of them. MacCulloch uses words because his domain mostly doesn't. Different medium, identical move: attach epistemic state to claim so the reader can weight accordingly.
Then read Trivers and learn why this is so hard. Humans evolved to project confidence — overstating certainty had selection advantages in coalition-building and persuasion, so the default cognitive program is calibrated against the move MacCulloch's discipline requires. The "may have" syntax is fighting the body's grain. You can feel the resistance when you try to write it honestly. Confident claims feel natural; hedged claims feel weak. That feeling is the evolved program objecting. Knowing it's evolved doesn't dissolve the resistance — but it reframes the discipline from "writing style" to evolved-bias corrective.
Bring Hughes into the room and the same syntax starts looking suspicious. In Hughes' frame, when a suspect hedges, the hedging is a tell — characteristic of people constructing rather than recalling. Read MacCulloch's prose through Hughes' eyes and the historian sounds shifty: she hedges constantly. But that's exactly the mistake the Hughes-frame can make. The historian's hedging is systematic and self-disclosed, not strategic and concealed. Same surface syntax, opposite function. Cover-hedging hides the writer's uncertainty; calibration-hedging exposes it. The distinction tells you something important — many behaviors are virtuous or suspect depending on whether they're transparent and consistent, or selective and strategic. The same is true of the move you're trying to learn here.
Now read the Indic Pramāṇa schools and see the apparatus formalized. They built explicit categories — direct perception, inference, scriptural authority, analogy — and required claim-makers to mark which kind of knowing produced each claim. MacCulloch's hedge-vocabulary is a folk version of that taxonomy. He doesn't categorize his "may haves" by epistemic source; he just hedges to a degree calibrated by the evidence. Where the formal taxonomy would tag "this is inference from absence," MacCulloch writes "we may think... the overwhelming sense of this is..." Same epistemic work, less explicit machinery. A historian who learned the Pramāṇa categories would probably produce sharper hedging by tagging the sources of each claim. The cost is some narrative flow. The gain is a reader who can distinguish inference from direct attestation at a glance.
The discipline connects to forecasting calibration, to the deception-detection literature, and to the Buddhist analysis of self-grasping. The shape of the network is: historiographic hedging as a folk version of a more general epistemic discipline that shows up under different names in different domains.
Behavioral Mechanics: Deception Detection — Hughes' framework treats hedging as a possible tell. MacCulloch's discipline turns the same syntactic move into a virtue by making it systematic and self-disclosed. The handshake produces a meta-principle: the same surface behavior (hedging) functions as deception-signal in one context and epistemic-virtue in another, distinguished by whether it's being used as cover or as calibration. That distinction generalizes: many behaviors are virtuous or suspect depending on whether they're transparent and consistent, or selective and strategic. The frame applies far beyond historiography.
Psychology: Evolutionary Psychology Hub — Trivers' self-deception literature and Buss's error-management theory explain why disciplined calibration is hard. Humans evolved to overstate their confidence because confident projection had coalition-building advantages; underconfident projection looked weak. The "may have" discipline fights against that evolved grain. The handshake produces the insight that historiographic discipline is a cognitive corrective — it's deliberately running counter to the default cognitive program. That reframes the discipline from "writing style" to "epistemic counterweight against evolved overstatement."
Eastern Spirituality: Buddhist Philosophy Hub — the Pramāṇa schools developed formal taxonomies of valid cognition (perception, inference, scripture, analogy) that mark which kind of knowing generated a given claim. MacCulloch's hedge-vocabulary is a folk version of the same project. The handshake produces a development question: could historiographic discipline be sharpened by adopting more explicit typology — marking, for instance, whether a "may have" comes from direct documentary inference, cross-source triangulation, or pattern-matching from analogous cases? The Indic schools have already worked out the taxonomy; historiography could adopt it.
Cross-Domain: Superforecasting as Discipline (if exists) or the Tetlock corpus generally — superforecasting calibration is the modern numerical version of MacCulloch's prose calibration. The handshake produces a practical implication: a historian who learned to attach numerical probabilities to her reconstructions (even informally, even just for her own use) would likely produce sharper hedging in the prose. The numerical discipline disciplines the linguistic discipline.
Psychology: Somatic-Trauma Theory Hub — the somatic literature names a bodily-felt phenomenon that is the somatic correlate of "may have." When you assert something you don't fully know, there's a felt-sense interruption — a slight hesitation in the body, a withdrawal from the smooth conviction the assertion would carry if it were true. The "may have" syntax is what that felt-sense looks like when it gets converted to language. The handshake produces a practical training implication: the historian who is somatically literate — who can detect her own felt-sense interruptions while writing — has a built-in calibration apparatus that the historian without somatic literacy lacks. The prose hedging tracks the body's hedging; the body's hedging tracks the evidential state more accurately than the conscious mind, which is constantly being pulled toward false confidence by Trivers' self-deception dynamics. Somatic literacy is therefore upstream of the writing discipline, not separate from it.
The Sharpest Implication
The discipline applies to your own claims about your own life. You make assertions about your past — what motivated a decision, what someone meant when they said something, why a relationship ended — with the same confident syntax most people use, with no "may have" anywhere in the sentence. But the evidential state behind those claims is often weaker than what MacCulloch operates with for the 16th century. You have one witness (you), with a stake in the version, working from incomplete memory. The "may have" discipline applied to autobiographical claim-making would change almost every sentence you have ever spoken about your own past. The third-wire reading is that most personal narrative is undisciplined historiography about one source — and the people who are most insufferable about their own pasts are the ones who refuse the hedging that the evidential state demands.
Generative Questions