I have been watching these lines for months:

The gap is very steady, which means that every time Trump moves up, so does Biden–at the expense of undecideds–and vice-versa. By way of contrast, this is how the 2012 election looked:

Granted, the scale is even tighter in the 2012 graph than in 2020, and the length of the trend is longer. Still, we saw points when the lines converged, crossed, or moved apart.

Likewise in 2016:

Or 2008, according to Gallup’s polling:

Henry Enten does better and compares Biden/Trump 2020 polling to historical trends going back to the beginning of polling. He says, “The steadiness in the polls is record breaking. Biden’s advantage is the steadiest in a race with an incumbent running since at least 1944.” Enten adds: “all the [national] polls taken since the beginning of 2019 have [Biden] up 6 points.” He means an average of six points, but that has a small standard deviation. The Real Clear Politics chart, reproduced above, suggests that the full range has been 4-10 points.

The consistency of this gap is noteworthy in an extraordinarily tumultuous period, marked by impeachment, a competitive Democratic primary campaign, a global pandemic, and the worst economic decline since 1929. Donald Trump’s own support has risen and fallen–although only within a five-point range. Biden has had his own ups and downs: near defeat in the primary, a serious accusation of sexual impropriety. Yet the gap between the candidates has been virtually unchanged since December.

Polls this far out are not necessarily very predictive. National polls don’t map exactly onto Electoral College outcomes. Polls conducted now include all adults or self-described “registered voters”; actual voters will be a subset of those. Turnout in 2020 is particularly hard to predict given the practicalities of voting in what may still be a pandemic.

But all these caveats are about whether the graph foretells the result in November 2020. Even if it doesn’t, it tells a very interesting story about now. As Matthew Continetti writes: 

It is not foolish to suppose that these world-shaking events would affect the presidential election. On the contrary: One would expect a dramatic swing toward either the incumbent or the challenger. But look at the polls. Not only has there been no big shift. There has been no shift. … Neither good nor bad news has an effect.

I think that almost all Americans have formed such firm and well-anchored beliefs about both Trump and Biden that even epoch-making events don’t shift us. We already have enough information to judge these men, whatever the news throws at us. By six percentage points, we like Biden more than Trump.

(By the way, Biden also leads in battleground states’ averages: by 3 points in WI and NC, by 4 in NV and FL, by 7 in PA and NH, and by 8 in Michigan. That would bode well for an Electoral College win, if we want to get into forecasting November.)