Lila Soto: Hugo, hey — I've been sitting with something weird since Tuesday and I need to get it out of my head and into yours.
Hugo Vance: Go on.
Lila Soto: My nephew got sick last week — fever, the whole thing — and my mom said, completely seriously, that she could tell because the cat stopped sitting near him. The cat left first, then the fever came. And she meant it as cause and effect. Like the cat sensed something and withdrew, and that withdrawal was a sign. And I kept thinking — okay, but which way does that actually run?
Hugo Vance: Which is precisely the error medieval Europeans made with lice.
Lila Soto: Right — but the part that gets me is how tidy their logic was. They watched lice leave the body when someone became ill, so they concluded lice were beneficial to health. Lice present, person healthy. Lice absent, person sick. Seemed obvious.
Hugo Vance: Yes, and the mechanism was the reverse entirely. Fever kills lice. The illness drove them off — they didn't leave because something was wrong, they died because of what was already wrong. The causal arrow pointed the opposite direction, and centuries of observation cemented the mistake.
Lila Soto: And that's — I mean, that's not a medieval stupidity thing, that's a human wiring thing. Sequence feels like proof. We see A, then B, and the story writes itself. So what we're actually trying to work out today is: how do historians — who are trained, who know better — actually escape that instinct? E.H. Carr said a study of history is a study of causes, not just a study of what happened next. But saying that and doing it are different things.
Hugo Vance: Carr names it as a discipline, not a default. That distinction is where the problem lives.
Lila Soto: But that gap — between naming causation as a discipline and actually doing it — I think it shows up in something really ordinary. Like a parent watching their kid fall behind in school, the kid gets sick for a week, comes back, struggles worse. The parent goes: the illness did it. And that story feels complete.
Hugo Vance: Yes. Because sequence is seductive. Two events in order and the mind fills in the arrow between them. The core problem is simply this: correlation means two things co-occur or follow in sequence — causation means one actually produces the other through an identifiable mechanism. Sequence is necessary. It is never sufficient.
Lila Soto: Never sufficient — okay, but what breaks it?
Hugo Vance: The third factor. What if something was already cracking before the fever? A learning difficulty, a classroom dynamic, something nobody examined because the illness arrived and gave everyone a story to hold. The illness and the decline both visible — the underlying cause invisible. That's the confounding variable. It drives both events, and the apparent link between them is spurious.
Lila Soto: Oh — so it's not that the sequence is wrong exactly, it's that the sequence is real but it's pointing at the wrong cause entirely.
Hugo Vance: Galton made exactly this error. He observed that parents and children share physical and intellectual traits — tallness, certain abilities — and he inferred direct parent-to-child transmission. What he missed was that shared genetic inheritance across the whole family line explained the association. Not direct causation. A third factor running underneath everything he measured.
Lila Soto: And Galton was a trained statistician — this isn't a naïve error.
Hugo Vance: Which is the uncomfortable part. Training sharpens the tools but does not, on its own, surface the factor you haven't thought to look for. The parent, Galton, the medieval physician with the lice — they're all doing the same thing. The sequence is real. The mechanism they name is wrong. So the question you're left with: what would actually be enough?
Lila Soto: Which gets me to the part I keep wanting to name — there's actually a checklist for this. Like a literal framework. Means, motive, opportunity. Borrowed straight from legal reasoning. A proposed cause has to have had the capacity to produce the effect, a reason behind the action, and the circumstances that made it possible. All three.
Hugo Vance: It is tidier than historians usually admit they find useful.
Lila Soto: Walk it through Mehmed II. 1453. Constantinople falls. What's the motive?
Hugo Vance: Ottoman imperial ambition, yes, but specifically dynastic legitimacy. Mehmed needed Constantinople to consolidate the sultanate's claim — it wasn't appetite alone, it was the symbolic architecture of power. Opportunity: Byzantine exhaustion, no Western reinforcement arrived. Means: superior artillery, the great bombards, and troop numbers the Byzantines couldn't match. All three present. All three connected.
Lila Soto: And that's — I mean, it almost feels too clean when you say it like that. Like the framework is retroactively tidying something that was probably chaotic.
Hugo Vance: Which is precisely why you run the counterfactual after. Remove Mehmed specifically — not the Ottoman empire, him — does Constantinople still fall in 1453? The empire has the means either way. But the dynastic motive, the specific timing, the willingness to commit those particular resources? That question is what sharpens whether this is a structural outcome or a genuinely agentive one.
Lila Soto: Oh wow — so the counterfactual isn't just a thought experiment, it's doing actual causal work. It's the thing that tells you how much weight to put on Mehmed versus the structure he was operating inside.
Hugo Vance: Montesquieu was doing exactly this in 1734. 'Considerations on the Causes of the Grandeur and Decadence of the Romans.' Not merely that Rome fell — why it fell, what had to be present. Most people assume that kind of systematic causal thinking is a twentieth-century import. It is not. And yet historians still resist naming it explicitly when they deploy it.
Lila Soto: And that resistance — yeah, that's the thread I don't want to drop. Because even with all three boxes checked and the counterfactual run, there's a deeper problem sitting underneath this that we haven't touched yet, about whether the framework can actually hold when you're trying to separate what Mehmed decided from what the structure made almost inevitable — and it gets considerably more complicated once you bring in how historians actually categorize causal investigation itself.
Hugo Vance: Well, that's where the framework starts to crack, actually. Because even if Mehmed checks all three boxes — means, motive, opportunity — you still face what I'd call the replacement question. If Mehmed had died in 1452, does Constantinople fall to a different sultan by 1460? The Ottoman structure was pointing that direction regardless. And if the answer is yes, then Mehmed's individual causal contribution... it shrinks. The structural force becomes the stronger candidate.
Lila Soto: Oh — so the checklist can pass and the agent still might not matter.
Hugo Vance: Precisely. And this is where Ivan Ermakoff becomes important. He identified three genuinely different modes of causal investigation — morphological, which is pattern-based across cases; variable-centered, treating factors as covariating; and genetic, which is process-tracing, step by step through a single sequence of decisions. They don't give you the same answer about agency.
Lila Soto: Hm — and you'd need different ones for different scales? Like, Braudel's longue durée versus someone's specific choice on a specific day?
Hugo Vance: Exactly that tension. Braudel says individual decisions are almost noise against structural tides — centuries of climate, demography, trade routes. Micro-history says the exact sequence of one decision changes everything downstream. Those are incommensurable scales. Process tracing suited to Mehmed's choices in 1453 simply does not translate to explaining why Mediterranean commerce shifted over three hundred years.
Lila Soto: I mean — so which one do historians actually use? Do they pick?
Hugo Vance: That's where Anton Froeyman's finding lands hard. He showed historians don't consistently use causation in one way — they shift between modes without flagging it. Same historian, same monograph, variable-centered in one chapter, process-tracing in the next. And when they say the conclusion is evidence-based... yes, well — they may be describing inference as readout. The primary sources give you traces of decisions, justifications. Not the causal chain. That has to be reconstructed through argument.
Lila Soto: So is that methodological self-deception, or is it just... what humanistic explanation looks like?
Hugo Vance: I think that's the question we can't resolve cleanly. And I'm not sure we should.
Lila Soto: Mahoney, Kimball, and Koivu wrote a whole book on this — 'The Logic of Historical Explanation in the Social Sciences' — arguing that the inferential standards have to be made explicit rather than left buried in narrative. And I keep turning that over. Like, they're right. And also I wonder if making it fully explicit just... produces a different kind of fiction.
Hugo Vance: Well, that's the reservation I can't shake. The humanistic case — and I think it holds — is that human intentionality is genuinely too granular for variable-centered formal models. What Mehmed weighed the night before the final assault — that doesn't compress into a coefficient.
Lila Soto: Yeah — so maybe the resistance isn't evasion.
Hugo Vance: That's the thing I genuinely can't resolve. The tools are real. The implicit counterfactuals are real. And naming them might just introduce false precision rather than actual clarity.
Lila Soto: What would history even look like if every implicit counterfactual got named? I don't know if that's more honest or just more theater. I keep sitting with both and they both feel true.
Hugo Vance: I think that's where we actually are. Not a resolution. That's — mm. An honest place to stop.
