Onpode
Cover art for The structural limit of asking 'what if' — how historians weigh plausibility without evidence

The structural limit of asking 'what if' — how historians weigh plausibility without evidence

November 22, 2026 · 11 min

Marcus Vale & Ben Okonkwo

Every causal claim in history secretly contains a counterfactual: 'had X not happened, Y would have differed.' Niall Ferguson's 1997 Virtual History rehabilitated this method, while Tetlock and Belkin set formal rules — change one variable, hold others fixed, make assumptions explicit — but passing those rules proves plausibility, not truth.

Counterfactual history—also called virtual history—is a form of historiography that asks "what if?" by altering a specified past event, decision, or condition and reasoning through plausible alternative outcomes.

0:0011:21
Get the next episode on History →

Follow it free — new episodes land in your feed.

Or make your own — any topic, in minutes

More Onpode episodes on History →

About this episode

There's a move every historian, investor, and post-mortem facilitator makes without noticing: to say X caused Y, you have to imagine a world where X didn't happen. Counterfactual history just makes that move explicit — and then argues about whether it's honest. The episode traces where the method came from (Weber's 'objective possibility,' Thucydides reasoning about Athens), why serious historians like E.H. Carr and E.P. Thompson attacked it, and why that attack turns out to be self-defeating. If you name a cause, you've already committed to a counterfactual. The only question is whether you show your assumptions. The Sarajevo case is the anchor: a driver takes a wrong turn, stalls in front of an assassin, and a war begins. Was the assassination the cause, or just the last spark in a pile already stacked? The episode doesn't resolve that — it shows why the question is the point. Isolating the event from the structural forces (alliances, militarism, nationalism) may simply be impossible, and no amount of rigor gets you a confirmed answer. What the best-practice rules do instead — the Tetlock and Belkin criteria — is push a story toward the plausible end of a gradient. Not true, not false. Consistent enough to argue about. That's the honest ceiling, and the episode is frank about it. Worth the eleven minutes if you've ever written a post-mortem and wondered who was making up the ending.

Frequently asked

What is counterfactual history and how do historians use it?

Counterfactual history alters one specified past event and reasons through what plausibly follows, in order to test whether that event was genuinely causal or merely coincidental. Historians use it because every causal claim — 'X caused Y' — implicitly assumes that without X, Y would not have occurred, so the method makes that hidden assumption explicit.

What are Tetlock and Belkin's rules for counterfactual history?

Philip Tetlock and Aaron Belkin's best-practice rules for historical counterfactuals require changing only one variable, holding everything else fixed, and maintaining three consistencies: theoretical (no contradiction of good theory), empirical (no contradiction of known facts), and statistical (no wildly improbable premise). These rules test plausibility, not truth — they define a gradient, not a pass/fail threshold.

Why did E.H. Carr call counterfactual history a parlour game?

E.H. Carr dismissed counterfactual history as a 'parlour game' in his 1961 book What Is History?, on the grounds that history only ran once and alternate outcomes can never be confirmed or refuted. Critics point out, however, that Carr still named causes in his own work, which implicitly relies on the same counterfactual logic he rejected.

Did the assassination of Franz Ferdinand actually cause World War One?

Whether the 1914 assassination of Archduke Franz Ferdinand caused World War One is precisely what counterfactual analysis makes difficult to settle. The driver's wrong turn stalled the car in front of Gavrilo Princip, making the shooting highly contingent. But interlocking alliances, militarism, and nationalism were already in place, so removing the assassination may not have removed the war.

What is Max Weber's 'objective possibility' test in historical analysis?

Max Weber's 'objective possibility' test, developed in the early twentieth century, asks what options were realistically available to decision-makers given their actual knowledge at the time — not options visible only in hindsight. A cause is then weighted by how much the outcome would change if that option had been chosen differently, filtering out wish-fulfillment scenarios.

Grounded in 12 sources
Motives and merits of counterfactual histories of science - ScienceDirect ↗ · sciencedirect.com
Full article: Counterfactual thinking & nuclear risk in the digital age: The role of uncertainty, complexity, chance, and human psychology ↗ · tandfonline.com
Everything You Need to Know About Historical Counterfactuals – Futures of Science (by Veli Virmajoki) ↗ · blogit.utu.fi
What If… Historians Were Honest About Counterfactuals? ↗ · contingentmagazine.org
Counterfactual History with Niall Ferguson - Centre for the Study of Governance & Society ↗ · csgs.kcl.ac.uk
Counterfactual Thought Experiments in World Politics: Logical, Methodological, and Psychological Perspectives 9780691215075 - DOKUMEN.PUB ↗ · dokumen.pub
Counterfactual history - Wikipedia ↗ · en.wikipedia.org
A tragedy I know too well: The assassination of Franz Ferdinand ... ↗ · gardenofmemory.net
Teaching What Didn’t Happen ↗ · insidehighered.com
Forum 15 on “Symposium on Counterfactual Analysis” - H-Diplo|RJISSF ↗ · issforum.org
Review Articles WHAT'S SO DIFFERENT ABOUT A ... ↗ · library.fes.de
Counterfactual History — Definition & Meaning ↗ · modeldiplomat.com
Read transcript

Ben Okonkwo: Marcus, hey, you ready? I'm handing you one claim and you tell me if it's true: every time you say something caused something, you're secretly imagining a world where it didn't happen.

Marcus Vale: Hold on. That sounds clever, and I distrust clever. Prove it with something real.

Ben Okonkwo: Okay, picture a failed launch. Someone says the missed date sank us. That only works if, on time, you'd have won. Take that away and you've just listed two things that happened in a row.

Marcus Vale: Huh. Okay, that lands. So how does a historian use it?

Ben Okonkwo: Counterfactual history, sometimes called virtual history: alter one specified event, reason through what plausibly follows, and that shows whether it was causal or just coincidental. The famous test is Sarajevo, 1914, the assassination of Archduke Franz Ferdinand. Did it cause the war, or was it just the spark in a pile already stacked?

Marcus Vale: And the skeptics? There must be skeptics.

Ben Okonkwo: Plenty. E.H. Carr called it a 'parlour game' in What Is History? in 1961. Yet he leans on that same logic whenever he names a cause. The wrinkle is that history only ran once, so nobody can rerun it.

Marcus Vale: Which means dismissing it is self-defeating. You can't call it a parlour game and then explain why anything happened. Basically it's what every VC does at a post-mortem. "We'd have survived if we'd raised in Q3." Untestable. We say it anyway.

Ben Okonkwo: Right, and that's the analytical school's whole argument. Any causal claim carries a subjunctive inside it. Had X been absent, the outcome would've differed. The method just makes you say that part out loud.

Marcus Vale: Where did this come from? Someone must have formalized it.

Ben Okonkwo: Older than you'd think. Thucydides reasons this way, about what Athens might have done otherwise. Then Max Weber, early twentieth century, gives it a tool: objective possibility. At a decision point, you ask what was actually open to the people there, and weigh a cause by how much the outcome changes when you remove it.

Marcus Vale: Objective. Meaning not just what we wish had been open.

Ben Okonkwo: Hm, yes. Options a person in that room could really have taken, given what they knew. Not hindsight options.

Marcus Vale: Okay. Which is the Excite story. Passed on buying Google in 1999, reportedly for around a million dollars. Everyone says idiots. But would Google have thrived inside Excite? Probably not. That's a hindsight trophy.

Ben Okonkwo: That's Weber's test failing in real time. Was buying it objectively possible, and would the outcome you're mourning have survived the purchase? Your trophy assumes yes to both. I'd want evidence for either.

Marcus Vale: And the modern revival?

Ben Okonkwo: Two markers. Winston Churchill, 1931, an essay called "If Lee Had Not Won the Battle of Gettysburg," in a collection, If It Had Happened Otherwise. Early, and frankly a bit of a game. Then Niall Ferguson edits Virtual History in 1997, arguing counterfactuals are unavoidable whenever you assign causes. That's the rehabilitation.

Marcus Vale: Define contingency for me. Plain terms.

Ben Okonkwo: The outcome depended on choices that could plausibly have varied. It wasn't forced by structure. And the counterfactual shows it by how fragile the path looks. If one change unravels everything, it was never inevitable.

Marcus Vale: But if it's that useful, why did Carr hate it? And E.P. Thompson. Somebody serious called it nonsense.

Ben Okonkwo: Thompson called it 'unhistorical nonsense.' Richard J. Evans, 2014, wrote that 'what if' is a waste of time. And their core claim is fair: history is what happened, not what might have. The alternate outcome never occurred, so no counterfactual can be confirmed or refuted. However careful you are.

Marcus Vale: Permanent, though? Look, I'd push on that.

Ben Okonkwo: Go on.

Marcus Vale: Tech reruns history constantly. An A/B test is a counterfactual with a control group. Hundreds of startups hit the same market at once. Frankly, a portfolio is a counterfactual engine. Untestability is overblown for anyone who has to decide something.

Ben Okonkwo: Hm, fair, for decisions. Hold it loosely, though. Some historical questions get partial tests from natural experiments or models. But Sarajevo has no control group. June 28, 1914, the first bomb fails. Then the driver takes a wrong turn onto Franz Joseph Street and stalls while reversing, right in front of Gavrilo Princip.

Marcus Vale: He stalled. In front of him.

Ben Okonkwo: Right. So does the war still come? Alliances, militarism, nationalism, all interacting. If those forces were already stacked, the wrong turn is just the spark. But if you can't isolate the assassination from them, calling it 'the' cause gets unstable. And nobody can run it twice.

Marcus Vale: So the wrong turn is evidence of fragility, or irrelevance. You can't tell which.

Ben Okonkwo: That's the problem. And here's the part I'd actually worry about. If nothing checks the story, the author picks the ending. People build alternate histories that prove what they already believed, politically or theoretically. That's the real danger. Not flippancy. Wish-fulfillment.

Marcus Vale: Give me the office version.

Ben Okonkwo: An engineering manager runs a post-mortem on a failed launch. One engineer says, ship two weeks earlier and we win. Another says the competitor's funding made it inevitable. Both untestable. So she writes each claim's assumptions on the board: customers would've switched, the rival wouldn't have reacted. Then everyone sees which assumption carries the whole story.

Marcus Vale: So the output isn't an answer. It's a list of bets you can finally argue about.

Ben Okonkwo: Yes. Gavriel D. Rosenfeld took on the flippant reputation and built a typology of approaches. But the field's real answer was rules, and the rules crack in a few specific places. Ferguson's own example is one.

Marcus Vale: Ferguson broke his own rules? Okay, now I want that.

Ben Okonkwo: Right, and that's why the field wrote rules. If the author picks the ending, you need something that stops the author. Philip Tetlock and Aaron Belkin set out the best-practice version. Marcus, you explain it. You've seen the whiteboard version.

Marcus Vale: Sure. Plain terms. Change one decision. Hold everything else as fixed as you can. Stay consistent with what's established. And make your assumptions visible. Minimal rewrite. Basically, one variable, not five.

Ben Okonkwo: Close. They add three consistencies: theoretical, empirical, statistical. Your story shouldn't contradict good theory, known facts, or what we know about how often such things happen. Okay, now the catch. Does passing those make the story true?

Marcus Vale: Presumably it makes it believable. Which isn't the same.

Ben Okonkwo: That's the gap. The criteria test consistency, not truth. A counterfactual can be perfectly consistent and still wrong. And there's no pass or fail. It's a plausibility continuum, a gradient. The rules push you toward the plausible end, but nobody gets a stamp.

Marcus Vale: So it's a credit score, not a verdict. Fine. Now Ferguson. Show me the break.

Ben Okonkwo: Virtual History, 1997. His German military spending counterfactual. Critics flag it as internally inconsistent, because by his own account the premise was politically impossible. I mean— he's saying, imagine Germany had spent differently, while also telling you it couldn't have. The leading rehabilitator broke the rule he helped popularize.

Marcus Vale: That's a founder violating his own playbook. Wow.

Ben Okonkwo: Wry, I know. But Cleopatra's nose is the deeper crack. Small contingent event, enormous consequence. Think of the stalled car in Sarajevo. Tiny change, continental war. Minimal rewrite says change little. But if the change is small and the effect is huge, the rule fails exactly where the interesting cases live.

Marcus Vale: So you can be disciplined or interesting. Not both.

Ben Okonkwo: Mostly. And notably, Tetlock himself moved on, from imagining to scoring, with forecasting tournaments, where predictions get graded against outcomes. That's the honest admission: if you want a check, you need a future that actually arrives.

Marcus Vale: Tetlock moved to scoring. Fine. But the past still matters for decisions. Is there a case where the what-if is used before anything happens?

Ben Okonkwo: Cuban Missile Crisis. Scholars take the documented evidence on what Kennedy wanted, what Khrushchev wanted, what the advisers actually argued for, and ask a hard question. If they'd chosen differently at a given point, would that probably have meant nuclear war?

Marcus Vale: So that's not trophy hunting. That's risk assessment.

Ben Okonkwo: Right. And the standard shifts, I think. Explaining the past, you're asking what likely happened in a world you can't rerun. Planning forward, you don't need the one true alternate history. You need a spread of scenarios, with the assumptions written next to each, so you know which one to watch.

Marcus Vale: Look, that's a pre-mortem. Write the what-if before the outcome. Does history have an analog, or is this new?

Ben Okonkwo: Hm. Not a clean one that I can name. The Cuban case is closest, because the evidence is rich enough to constrain the scenarios. Even there, I'd hold it loosely. Nobody knows the true odds of war. We know which assumptions about Khrushchev's aims carried the weight.

Marcus Vale: Kodak, then. Steve Sasson builds a digital camera in 1975. Death by technology, everyone says. Inevitable.

Ben Okonkwo: That's technological fetishism. Technology gets its own agency, so the outcome looks like physics. Counterfactuals pull that apart. Ask what if management had chosen differently, and the outcome starts to look like politics, hubris, accident. Contingency, not destiny.

Marcus Vale: And structural versus contingent for Kodak? Can't test that either.

Ben Okonkwo: No. Same wall as Sarajevo. But you can lay out what each story assumes, and then argue about the assumption instead of the ending.

Marcus Vale: So I'll concede it. The tension's permanent. Portfolios help me, they don't cure it. The practical move is the whiteboard. Write down what you assumed, in advance, so the post-mortem can't rewrite it.

Ben Okonkwo: That's what clicked for me, actually. At the start I gave you a claim about secret imagined worlds. Now it sounds less like a trick and more like a habit. Every cause you name is an argument, and the honest version shows its assumptions so someone can challenge them.

Marcus Vale: Funny. You opened with the missed launch date, and I distrusted it for being clever. I'd still bet on whoever wrote the assumption down first.

Ben Okonkwo: I'd take that bet, with the assumption written on the board. Thanks for pushing on it. The stalled car is going to bother me for a while.

The structural limit of asking 'what if' — how historians weigh plausibility without evidence · Onpode