Four years ago this month, it seemed likely that Hillary Clinton would win the presidential election. Clinton’s national lead over Donald Trump had repeatedly widened and narrowed, suggesting that, depending on where the waveform truncated on Election Day, her victory would similarly be either wide or narrow. State polling, though, consistently showed that she had enough of a lead in enough states that she’d manage to win both the popular and electoral votes.

That she didn’t, that Trump beat the odds, has become the political equivalent of dividing by zero, an act deemed so outside of expectations that it negates any presumptions about what to expect this time around. Now-President Trump uses a version of that argument all the time: Look what happened in 2016; no one should take the polls at face value.

In reality, Trump’s win was not some black-swan event. In the last week before the 2016 election, the race tightened, with undecided voters beginning to break for Trump. His position as the race ended was the strongest he’d seen over the course of the campaign, probably a function of people running out of time to choose a candidate and choosing him. The national polls correctly predicted that Clinton would have a narrow popular-vote lead but, as FiveThirtyEight wrote four days before Election Day, the closeness of the race meant a very real chance that Trump might still win. As voting concluded, Trump’s chance of winning was, according to FiveThirtyEight’s analysis, slightly better than the odds that you’d get heads twice in a row if you flipped a coin.

Trump beat the odds — but the odds weren’t as bad as the narrative has subsequently suggested.

All of this background is meant to illustrate how odd it was for Axios to present the 2020 election as being considered a fait accompli by pundits and media observers.

“It feels like August of 2016 all over again,” an article published Wednesday read. “Polls show Donald Trump losing big. Pundits proclaim he can’t win. Reporters sneer at Trump voters on Twitter and cable.”

It’s useless to rebut anecdotalism with anecdotes, but this thesis does seem hard to justify. I lamentably spend a decent amount of time reading pundits and talking to other reporters, and the sense of what’s happening is, if anything, the opposite of what’s presented above. Despite former vice president Joe Biden’s national lead, observers are generally extremely skeptical of that lead, probably unduly so. The pervasive sense of inevitability is that 2020 will fail to behave as expected, something this year has demonstrated over and over again. Biden’s up eight points? Then this will be the year that he wins every vote in California and loses enough states by 10 votes that Trump grabs the electoral college 483 to 55.

So let’s instead look at what the polls actually say. Specifically, let’s compare the current RealClearPolitics polling average in 10 swing states/possible swing states with the past two elections. Oh, by the way: That “possible swing states” formulation is itself an artifact of the 2016 race. At this point four years ago, it didn’t seem like the contest would come down to Michigan, Pennsylvania and Wisconsin, where Clinton led by eight, seven and five points, respectively. But it did.

On these graphs, the red line shows Biden’s margin relative to that of Trump. The lower the better for the former vice president.

What we see is a race that looks, in many places, like the race that was underway four years ago. The most recent RealClearPolitics averages in each of these 10 states has Biden outperforming Clinton in only three. On average, though, Biden’s margins are only 1.9 points different from where Clinton was. That includes Iowa and Ohio, two states where there hasn’t been much polling of late.

On the surface, this seems like bad news for Biden, right? Well, it’s not where he’d want to be, but his current position in some averages is informed by the addition of a number of polls that haven’t been particularly strong for him.

It’s worth remembering, too, that the errors in state polling in 2016 have prompted a lot of scrutiny of how those pollsters viewed the likely universe of voters. You’ll notice that in a number of the states above, the final result (the gray circle) was above the final polling average in 2016 — meaning that the result was more favorable to Trump than the polling average suggested. That shift wasn’t captured in the polling. But also notice that in 2012, a lot of the actual results were lower than the final poll averages, meaning better for Barack Obama. Obama, you’ll remember, was supposed to be facing a close contest that year but ended up winning fairly easily. Four years later, the polls were off in the other direction; perhaps, in some cases, because pollsters didn’t want to again underestimate Democratic support.

It may be the case that the state polls in which Biden is performing better than Clinton at the moment have been better calibrated to be less likely to overstate his support. Biden running even with Clinton in a poll less likely to overstate Democratic support would mean Biden’s in a better position. But this is speculation and hard for various reasons to measure.

There are other reasons for pessimism by Trump’s team. If, for example, Biden holds leads in Florida and Arizona, he doesn’t need to pick up any of the Midwestern states that Trump picked off last time. Trump can lose a small state, but he can’t lose both of those big ones.

Nationally, Biden is still outperforming Clinton by about three points. That’s in part because, at this point in 2016, her lead over Trump was going through one of its narrowing phases following the party conventions. (The effect of the conventions may be more muted this year.) By contrast, Biden’s lead has been relatively stable.

In FiveThirtyEight’s polling average, that stability is even more pronounced.

We’ve noted before that Biden has maintained consistent advantages over Trump in polling. His lowest level of support in RealClearPolitics’s average has always been higher than Trump’s highest level of support. Trump’s approval ratings have been unusually stagnant and unusually low. He’s lost the advantage of being preferred by voters skeptical of both candidates.

Can Biden win? He can and, if this were a normal year and Election Day were tomorrow, it seems quite likely that he would win. Can Trump win? Of course! The numbers are good for Biden, but they’re not guaranteed-to-win good, even if the election were tomorrow.

Maybe there are pundits who feel firmly that the election is over. If so, my recommendation would be that you ignore those pundits. After all, not only do they not know what they’re talking about, they are apparently the only people in these United States who are unaware of what happened in November 2016.