# Counterfactuals **The model:** Reasoning about what *would have* happened if some condition had been different. "If the U.S. had not entered World War I" or "if Comey had not sent the October 28th letter" or "if Florida had been recounted in 2000" are counterfactual claims — they assert something about a world that didn't happen but is being compared to the world that did. Counterfactuals are not optional decoration on causal reasoning; they are *the substance* of causal reasoning. To say "X caused Y" is implicitly to say "if X had not occurred, Y would not have occurred (or would have been different)." Without the counterfactual structure, causal claims reduce to mere correlation. Judea Pearl's influential work on causality (the "do-calculus" and the structural-equations approach) makes this explicit: causation is the difference between observing X and intervening to set X. ## The Mechanism **The closest-possible-world frame.** A counterfactual is true if, in the *closest possible world* where the condition is different, the outcome is different. "Closest" is vague — it requires holding most things constant while varying the relevant condition. The work of counterfactual reasoning is keeping the held-constants honest and not allowing motivated reasoning to vary things that shouldn't vary. **Multiple counterfactuals can all be true.** Different "closest worlds" answer different questions. "If Comey hadn't sent the letter" closest-worlds disagree on whether Trump still wins (depends on what else changes). The honest practice is to specify which counterfactual you mean and which assumptions you're making. **Counterfactual reasoning is unavoidable.** Even the most empirically careful claims rest on counterfactual support. Studies, randomized trials, regression analyses, historical narratives — all are devices for estimating what would have happened in counterfactual conditions. Pretending you can avoid counterfactuals is just doing them sloppily. **The error is overclaiming precision.** Counterfactual claims are inherently uncertain — you cannot rerun history. The error is asserting a counterfactual outcome with the confidence of an observed one. "If we hadn't passed Reaganomics, the economy would have collapsed" is a counterfactual; the speaker has no observation to support it. So is "if we *had* passed single-payer healthcare in 2009, costs would now be lower." Both claims may be true, but neither is observation. ## KB Applications ### "But For" Reasoning in Election Counterfactuals The 2016 election generated an entire genre of counterfactual claims: "but for Comey's letter, Clinton wins"; "but for Russian interference, Clinton wins"; "but for the [[Electoral College]], Clinton wins." Each is a different counterfactual with different assumptions and different evidentiary support. The Electoral College claim is well-supported (Clinton won the popular vote by 2.9 million); the Comey-letter claim has econometric support but contested causal identification; the Russian-interference claim has the weakest causal evidence because the interventions and the effects are hardest to isolate. Useful analysis distinguishes which counterfactuals are well-identified and which are speculative. ### Originalism's Implicit Counterfactual [[Originalism]] is in part a counterfactual exercise: what would the framers have decided about *this* contemporary case if they had been confronted with it? The counterfactual is unavoidable because the framers did not in fact confront most contemporary cases. The honest objection is not that originalism uses counterfactuals (everyone does) but that originalism's counterfactuals are typically *underdetermined* — multiple framer-counterfactuals are equally consistent with the historical record, and the choice between them tracks the interpreter's prior preferences. The counterfactual machinery hides the discretion. ### What If MAGA Loses 2024 Decisively? A useful exercise: counterfactual histories of the 2020-2025 period under a Trump-loses-2024 scenario. The result helps clarify which features of the current situation are durable (policy infrastructure built up by Heritage and others) and which are contingent on specific personnel (executive-branch chaos around Trump personally). The counterfactual frame disciplines analysis that would otherwise blur structural and personal causation. ### Counterfactuals and Historical Memory Right-wing historical narratives frequently invoke counterfactuals: "if we hadn't had the [[Great Society]]" (welfare-state critiques); "if the South had been allowed to manage Reconstruction" ([[Lost Cause]] variant); "if the 1960s had not happened" (cultural-decline narrative). Each is a counterfactual claim with substantial work to do. Naming them as counterfactuals — and asking what evidence would distinguish them from competing counterfactuals — is the analytical move. ## Practicing Counterfactual Reasoning 1. **Specify the counterfactual.** "What if X had been different?" needs both *which* X and *which world otherwise stays held constant*. Vague counterfactuals are arguments for whatever you wanted to argue anyway. 2. **Acknowledge the uncertainty.** Counterfactual claims are intrinsically less certain than observed claims. Confidence-level honesty separates serious analysis from advocacy. 3. **Use counterfactuals to test causal claims.** When someone says "X caused Y," ask: "and if not-X, what would Y have been?" If they can't answer, the causal claim is undercooked. 4. **Apply symmetrically.** Don't apply tight counterfactual rigor to your opponents' historical narratives and lazy counterfactual reasoning to your own. The diagnostic is bipartisan. ## See Also - Cause and Effect — counterfactuals are how causation gets analyzed - [[Correlation is Not Causation]] — the core counterfactual lesson - [[Inversion]] — the structural cousin - [[Second-Order Thinking]] — extends counterfactual reasoning forward in time