How close did we come to a second Trump term?

This is one of those counter-factuals that historians and political scientists will probably be kicking around a lot in years to come.

Right now the standard narrative is that the 2020 presidential election wasn’t really close in the end, which is superficially plausible given that Biden won by more than seven million real votes and 74 electoral votes.

But:

Only three state elections were really close — that is, decided by considerably less than a one percent margin: Arizona, Georgia, and Wisconsin. Biden won all of them (The only semi-close state presidential election that Trump won was in North Carolina, but the margin there — 1.3% — was quite a bit more than double the average percentage margin in the three really close states).

When an election ends up being decided by a lot less than one percent of the vote, that means the outcome was determined by marginal factors that to a great extent were random. To put it another way, if the election had been held a couple of days earlier or later, or if the weather had been significantly different, etc., these three states could just as easily have gone for Trump. They were coin flips. Those coins all flipping the other way would have resulted in a 269-269 tie, which means the House would have then elected Trump, thanks to our increasingly ridiculous Constitution.

Now on one level that still doesn’t mean the election was terribly close as a statistical matter, because TECHNICALLY (a critical caveat, see below) Trump would have had to have won all three states, and the odds of him doing that if the election had been a couple of days later or whatever were only 12.5%, assuming plausibly enough that the various shifting random marginal factors would have produced statistically independent results in the three crucial states.

But:

While an 8-1 shot of liberal democracy going completely bye-bye is scary enough by itself, there’s a further major complicating factor here: The odds are good that Trump would have managed to steal the election if he had won just two of the three coin flip states. This is the crucial counterfactual point. Would Republican legislators have refused to throw the election to Trump if a single state could have swung the outcome? (The game theoretical situation alters completely when the expected payoff of such a move goes from very low, which is what it is when you have a three-legislature collective action problem, to extremely high, when success is basically assured by one state flipping.) The answer is unknowable but let’s be thankful we didn’t get a chance to find out . . . this time.