The race isn't tightening. It was always close.

If you have been paying attention, you might have noticed that we are getting pretty close to this election thing actually happening. It's taking place in real life now. Joe Biden has even left his basement and, despite his earlier assertions that he would not be traveling during the general election this year, is out on the campaign trail to dispense words of wisdom — e.g., "COVID has taken this year, just since the outbreak, has taken more than 100 year [sic]." Audiences are no doubt thrilled.

The conventional wisdom concerning Biden's about-face is that the race is tightening in Donald Trump's favor. This is being attributed in part to the latter's performance at last week's Republican National Convention, which seems to me an unlikely explanation, and to the rioting in Portland, Kenosha, Wisconsin, and other cities. This is at least somewhat more plausible.

The truth is, though, that the extent to which we are seeing a shift in polls is just as likely to reflect what the actual state of the race has been for some time, which is to say, as close as any election in modern history. Trump has always benefited to some extent from what sociologists call "shy Toryism," the reluctance of voters to admit support for a candidate or party widely considered unsavory. But by how much? Which is really more likely, that Biden's support has plummeted in a state like Michigan by more than five points since public polling began for this cycle, and now stands basically where it did before five months of lockdown and rising crime and unemployment, with all of the above redounding to Trump's credit somehow, or that this was always going to be the nearest run thing you ever saw?

The polls in other likely battleground states also leave me skeptical of just-so stories about a suddenly tightening race. Yes, Trump appears to be doing better than he was in Wisconsin a month ago, but he is still down from where he was in, say, March. On the other hand, in 2016 the Real Clear Politics polling average never showed Trump coming within even five points of Hillary Clinton in the state, which he ended up winning by an exceedingly thin one-point margin. In Florida and Pennsylvania, Trump is running better than he has been since April, but Biden is still on top in both states. The only state in which he appears actually to have rebounded is North Carolina, and even there his lead is well within the margin of error.

Which is pretty much what we should expect to see between now and November. The truth is that in a year in which everything — gas prices, the numbers of miles traveled by airplane, rates of violent crime, Major League Baseball RBI totals — we measure statistically looks like an outlier, it is impossible to make hard and fast predictions about the outcome of an election. Pollsters failed miserably in 2016, not so much because they failed to predict Trump's eventual victory but because they wildly overestimated support for his opponent in states like Michigan, where as late as October 18 Clinton's lead was pegged at 12 points. The margin was always going to be much thinner than that. Everything about this year's election is going to be strange, and most of us would be grateful if a handful of failed prognostications ended up being its most obvious irregularities.

Trump's chances this year have never been as bad as some people (including a great many of his supporters) have led themselves to believe. They are still strong, and I would even go so far as to say that I expect another narrow victory. But if he does win it will not be because 100,000 people spread across five or so states changed their minds in the closing days of the campaign but because something that was set in motion long ago once again proved difficult to forecast.