Tetlock and Counter-factual History
By Bryan Caplan
I’ve long been fond of counter-factual history. Like: If Prinzip hadn’t assassinated Archduke Franz Ferdinand, North Korea wouldn’t be a Communist dictatorship. Or: If Lenin had died in early 1917, the Communists wouldn’t have taken over Russia, and the Nazis probably wouldn’t have taken over Germany.
In many cases, I freely admit that counter-factual history is hard and hazy. I make no confident claims about what the world would be like today if Christianity had never arisen. In other cases, however, I think counter-factual history is pretty easy and clear – especially if we’re only discussing the following decade or two rather than the entire subsequent history of mankind. If Julius Caesar hadn’t been assassinated in 44 B.C., it’s very likely he would have remained the leader of the Roman Empire for another 5-10 years, and the civil war that followed his death would not have occurred. Would the Roman Empire have lasted longer or shorter? Beats me.
Question: How can I reconcile my qualified confidence in counter-factual history with my admiration for Philip Tetlock’s Expert Political Judgment? Doesn’t he expose such pontification as pure conceit?
I think not. Contrary to populist readers, Tetlock’s work doesn’t show that political experts don’t know what they’re talking about. What it shows, rather, is that experts are overconfident on topics that are controversial among experts. Tetlock freely admits that naive extrapolation of existing patterns is often highly predictive. But he deliberately refrains from posing such questions to experts because they’re too easy. When designing his survey, Tetlock explicitly picked question to “Pass
the ‘don’t bother me too often with dumb questions’ test.”
When I make confident claims about counter-factual history, I’m normally doing this easy extrapolation. If Prinzip hadn’t killed the Austrian Archduke, it’s very likely that Europe would have remained at peace for at least a few more years. Sure, another international incident could have struck the very next day. But international incidents shocking enough to spark major wars very rarely occur.
Similarly, if Lenin had died in early 1917, we have strong specific evidence that his followers wouldn’t have violently overthrown the Kerensky government. Sure, another radical party could have tried Lenin’s approach in his stead. But events like Lenin’s coup aren’t just very rare in general. They’re so rare that I’ve studied this critical period in Russian history for decades without hearing about any other party that seriously pondered a Lenin-style coup against Kerensky. Of course, given a few more decades, other horrific events might have ruined Russia, clouding long-run prediction. But it’s still probable that the Russian Civil War that followed World War I wouldn’t have happened if Lenin had died the day Kerensky gained power.
In sum, while Tetlock’s work does highlight serious pitfalls of counter-factual history, there’s no reason to abandon the subject. As long as you (a) explore what would have happened if potent-but-rare events had gone otherwise, (b) extrapolate from long-run trends, and (c) discuss short- to medium-run consequences, your efforts are not in vain.