What’s happened over the past 12 hours in the presidential contest is nothing unusual or surprising. Initial results in key states showed President Trump with a lead. As absentee and mail-in ballots are being counted, though, his leads have narrowed or vanished, a function of those votes being stronger for Democratic nominee Joe Biden.

This was all sufficiently predictable that Trump has for weeks telegraphed his strategy as those votes were counted: dismiss the changes in the total as evidence of some unspecified fraud. And sure enough, that’s the claim Trump made during a brief appearance at the White House early Wednesday morning.

Pro-Trump Twitter has characteristically seized upon Trump’s rhetoric, searching eagerly for evidence, however sketchy, that Trump’s fraud assertions were warranted. So we end up with things like this:

At some point, it seems, Decision Desk HQ (from which those images are taken) updated its data for Michigan, showing a sudden jump for Biden. Trump, predictably, retweeted the claim.

What happened? Nothing nefarious.

Decision Desk HQ, like other outlets including The Washington Post, uses complicated data systems to power its election-night graphics. There’s no one there (or here) who types the results in; the maps are powered by an automated data-update process. So if there’s a problem with the data, there’s a problem with the map.

“It was a simple error from the state-created file that we ingested,” the group’s Scott Tranter explained over email. “It was caught and corrected (by the state). This happens.”

It does. Another such error was spotted earlier Tuesday evening by writer Dan McLaughlin, in fact. Such errors crop up and are then resolved. (Here’s another example, from 2016.) In this case, the state fixed the error within minutes, and the data was soon correct. No phantom vote-rigging, just good old-fashioned human error on a busy night in Michigan. The Republican operative who tweeted the initial claim deleted it and admitted his error.

But again, it’s because of how the votes are counted that such theories can emerge in the first place. Because Michigan (and Wisconsin and Pennsylvania) decided to wait until Election Day to count votes that had been submitted and received at times weeks before Tuesday, those votes were added only belatedly. (This decision was a function of the Republican-led legislatures in each state.) And since they favored Biden in general — and since more-populous places have more such votes and tend to lean Democratic — it becomes easy to layer a pall of nefariousness over the proceedings.

The Washington Post’s Philip Bump analyzes how Democratic presidential nominee Joe Biden flipped key states blue in 2020 and what it means for his campaign. (The Washington Post)

This is an important point, so let’s emphasize it — maybe even with some bolded text.

Vote-counting doesn’t reflect actual changes in the race.

We’ve been over this any number of times. My favorite analogy is to imagine an NFL reporter filling in his audience on the results of the Super Bowl by randomly announcing various touchdowns and field goals. The outcome will be the same for the listener (assuming he announces all of the points scored), but the way in which the listener perceives how the game went will depend on the order in which the points are announced.

To use a more to-the-point analogy, imagine a state with four counties, Alpha, Beta, Gamma and Delta. Between the four of them, they have 24 voting precincts of varying size. In this state, as in the United States, there are two parties, Orange and Purple.

In the example we’ll consider, the Orange candidate wins the election by 10 points, 55 percent to 45 percent. She does so on the strength of support in absentee ballots, 74 percent of which she won. The Purple candidate won the vote on Election Day, though, taking 61 percent of that vote.

Let’s assume that votes are counted in a fashion roughly similar to the counting in Michigan. Two precincts of in-person Election Day votes each hour, with those precincts’ absentee votes tallied two hours later. (If you’re curious, there’s a big table of the data we used for this at the bottom of this article.)

Purple starts off with two-thirds of the vote. That lead narrows until Orange overtakes him at 11 p.m.

Granted, there weren’t a lot of precincts in, so reputable news agencies wouldn’t have assumed a Purple victory. But you can see how a narrative of “Purple’s lead is eroding” can be formed. It’s hard to avoid! We used it ourselves, above.

But of course Purple was never winning. The results at 8 p.m. were that Orange was the choice of 55 percent of voters. It just took hours before that choice was made obvious.

Notice what happens, though, if we count all of the mail-in votes at the front end. The state had them all in-hand after all, so why not?

Suddenly, Orange’s win is never in doubt.

Yes, it’s overstated at the beginning, but what’s important here is the perception. In one scenario, Purple seems like he is watching his lead slip away. In the other, Orange is seen as the clear winner from start to finish. Which is more accurate?

Had Michigan (and Pennsylvania and Wisconsin) simply counted absentee ballots in advance, there would be no perception that Trump’s position was softening. There would be no untrue conspiracy theories about Biden being given votes. We would know now what we’ll know in a few hours or days: who voters in Michigan actually supported for president.

It looks like that wasn’t Trump. Which is why Trump wants to convince everyone that the very normal process of counting votes is, instead, some grand conspiracy organized against him.

Here is the data we used:

This article was updated with Mackowiak’s correction.