On a night where so many polls, models and expectations were obliterated, there was at least one prediction that remained intact by the morning: that this presidential election would probably end up hinging on a few states in the rust belt.
What happens over the coming hours remains unclear. Indeed, there is a reasonable chance that the final tally ends up taking weeks.
But one thing is certainly clear. The pollsters who predicted that lightning couldn’t strike twice – that their confident probabilities about a landslide victory wouldn’t be proved wrong for a second time in a row – have egg on their faces all over again.
And, whoever ends up occupying the Oval Office in January will preside over a country which remains deeply divided, where those divisions are writ in the electoral college and in Congress.
Most pundits presumed this was the moment where the Democrats would seize control of both the executive and the legislature, winning both the presidency and the Senate back from the Republicans.
Even if Joe Biden wins a narrow victory in the presidential election, he looks unlikely to take the Senate.
Even if he wins, the cult of Donald Trump has not ended. The American people came out and voted for their president in numbers far greater than anyone had anticipated.
They defied both the media and the pollsters who assumed they would punish the president for his handling of COVID-19.
Quite where things departed from the script written by the pollsters is hard to say.
Was it the Latino voters in Miami who abandoned Joe Biden in favour of Donald Trump? Was it the white non-college educated voters who turned out to redouble their support for the president, rather than turning back to the Democrats?
Was it the suburban women who didn’t quite follow the script they were expected to? Was it the fact that no one could quite anticipate the impact of early voting?
In time we will find out. It is still way too early to take any lessons from this evening and morning – save that we should remain deeply sceptical about firmly held assumptions when it comes to elections.
To make one enormous polling error may be regarded as a misfortune; to make two in a row looks like carelessness.