You don’t often see the number 57 percent in national, late-presidential-election polling. Maybe when measuring support among some demographic group, but not overall. The last time a presidential candidate received at least 57 percent of the actual vote was in 1984, when Ronald Reagan’s reelection bid earned him more than 500 electoral votes. Things like that don’t really happen anymore.

And yet here was CNN and its polling partners at SSRS, releasing a poll Tuesday morning showing Democratic presidential nominee Joe Biden leading President Trump by 16 points, with Biden earning the support of 57 percent of respondents. Now, in hyper-polarized 2020, CNN’s poll has Biden getting nearly 6 in 10 votes nationally. To call this surprising is an understatement.

It didn’t take long for people to dump cold water on enthusiastic Biden supporters. In mid-October 2016, after all, CBS News released a poll showing Hillary Clinton leading Trump by 14 points — and we know how that turned out. The central lesson from 2016 for many observers was that things can change dramatically and rapidly, and nothing should be taken for granted.

Which is true, in the broadest sense of that advice. But it is also important to recognize that the 2020 race has not generally mirrored the race in 2016. It’s important to recognize that Trump is now in a much worse position than he was four years ago, parallels aside.

The story of the 2016 race was wild swings in support. That Clinton led Trump was consistent, but the margin of that lead varied widely. In the last 150 days of the contest, the range of margins between Clinton and Trump spread across more than 10 percentage points. It was a sine curve, with Clinton seeing wider and narrower leads over the course of the year. The race ended with a narrower lead — narrow enough that Trump wrung an electoral college victory out of the contest.

By contrast, the 2020 race has been remarkably stable. Biden’s lead has never been less than six points or more than 10 points. It’s just a steady lead, as unaffected by the wild events of the year as the Rock of Gibraltar is obviously unaffected by a storm in the Mediterranean.

(Most of the polls and charts in this article use FiveThirtyEight’s much-appreciated historic and current polling average data.)

The stability of the margins in the 2020 race has been remarkable, even in the context of the past five elections. In 2000, 2004, 2008, 2012 and 2016, there was at some point a tie in the polling average in the last 150 days of the race.

Not this year. This year, there hasn’t even been a moment in which Biden’s lead was as narrow as five points.

Margins can be deceptive. That CBS News poll showing Clinton with a 14-point lead isn’t really comparable with the new CNN poll in part because her support in it was 52 percent. The issue for Trump wasn’t that Clinton was outpacing him, it was that 38 percent of voters planned to support him in that moment. Clinton had a big lead but relatively modest support.

Biden’s 57 percent support is much more robust. While CNN’s poll is currently an outlier (though, of course, that could change), the gap between Biden’s support and Trump’s is consistent with past polls from this cycle.

If we look at the average levels of support for the candidates in FiveThirtyEight’s data, we see a massive gap between Trump and Biden in 2020. Normally, the range of support seen by the candidates has some overlap, meaning that one candidate’s highest level of support is higher than the other candidate’s lowest level of support. In 2020, that hasn’t been the case.

In FiveThirtyEight’s average, Biden’s never been below 49.5 percent support in the last 150 days of the race, and Trump’s never been above 43.6 — a nearly six-point gap.

So what happened in 2016? Well, for one thing, there were more solid third-party candidates. For another, there was a much bigger pool of undecided voters. The two races in which we’ve seen recent splits between the popular vote and the electoral vote were in 2000 and 2016, years with stronger third-party candidates and more undecided voters late in the election.

In 2020, that’s not the case. Measuring the level of third-party support or undecided voters in the FiveThirtyEight average, we see that about 6 to 7 percent of voters fall into one of those categories. Most voters have made up their minds — as would make sense, given how polarized views of Trump are — and there’s not much of an untapped pool of voters for either campaign to tap into.

Trump trails in FiveThirtyEight’s average by more than eight points right now and only a bit more than 6 percent of the electorate is undecided or leaning third-party. That’s a problem for the incumbent.

Another problem is that the undecided voters are probably less likely to give him the benefit of the doubt than they were four years ago. One metric I’ve looked at frequently is how voters who dislike both candidates plan to vote. In 2016, Trump won that group on Election Day by 17 points. That was enough to make the difference in the three states that gave him his electoral college lead. This year, though, Biden has consistently led Trump among voters who view both him and Trump unfavorably — and led him by more than 20 points.

Here, too, though, there’s a difference from 2016: Biden’s favorability with voters is substantially better than was Clinton’s four years ago.

That captures one part of a bigger story of how 2016 and 2020 differ: that there have been remarkable swings in how demographic groups view Trump. For example, voters 65 and older have preferred the Republican to the Democrat in each of the past four presidential elections. It’s a group that votes heavily and made up an important part of Trump’s 2016 support.

Polling this year (including from The Washington Post and our partners at ABC News) show Biden with an advantage among older voters. He led by five points in our recent poll, which had Biden up 10 points overall. In CNN’s new poll, Biden leads among older voters by 21 points.

And then there’s support among women.

Women have consistently voted more heavily for Democratic candidates in recent elections, but not by the margins we’re seeing in current polls. In the Post-ABC poll, women prefer Biden by 31 points. In CNN’s poll, the spread is 34 points. In 2016, Clinton won women by 13 points. And women make up more of the electorate than do men.

Now we get to the giant asterisk that’s been floating over 2020 polling like the Goodyear blimp: National polls don’t matter in a contest decided in state-to-state fights. And that’s obviously true, as both Clinton and Trump would be happy to inform you.

Trump’s position in state polling is better than his national position, even relative to 2016. Biden still leads in nearly every swing state, as we’ve written, but his leads might be subject to polling error, which could mean narrower victories than expected or repeat Trump victories.

Last month, we created this interactive, using current averages to compare state and national polling this year with years past. Explore. (On the interactive below, the arrows at right show the actual final vote margin.)

Show average, including

Your browser cannot display this graph.

The problem with having a blimp sitting over your head is that things which might in reality be pretty clear become shadowy. If every polling average is right on the money, no average changes before the election, and voters are able to cast votes as though there wasn’t a pandemic ensnaring the country, Trump would lose the 2020 election in dramatic fashion.

But none of those three things can be assumed. What we can say, with confidence, is that Biden is in much better position now than Clinton was four years ago. No one who watched 2016, though, should be willing to assume much more than that.

The Fix’s Aaron Blake analyzes how the first presidential debate and President Trump’s positive coronavirus test could impact the final month of the campaign. (The Washington Post)