Thinking through the various ways in which the 2020 presidential election may unfold, I thought one unlikely but feasible possibility demanded attention now, and wrote it up at New York.

Trump remains one of the most unpopular presidents ever. His party is unpopular, too, as evidenced by Republicans’ poor performance in the 2018 midterms. Democratic voter enthusiasm seems high, fed by Trump’s daily antics. And the reasonably high odds that voters will eject him from the position he improbably won in 2016 while losing the popular vote are the very reason so many candidates want to run against him.

That’s the “glass half full” way of looking at the landscape. But there’s a “glass half empty” take that’s plausible as well.

After what should have been a calamitous stretch in which he shut down the government for an unpopular border wall, declared a nonexistent national emergency, and underwent a whole new round of high-profile airings of his alleged 2016 sins, Trump’s approval ratings have bounced back to the low-to-mid 40s levels that appear to be his long-term floor. Lest we forget, according to Gallup, his favorability number just before winning the presidency in 2016 was 36 percent (61 percent of respondents gave him an unfavorable rating).

And speaking of 2016, Republicans arguably retain an Electoral College advantage. The initial 2020 presidential battleground map from Larry Sabato’s Crystal Ball showed Republicans with 248 electoral votes and Democrats with 244, with 46 in the “toss-up” category. Kyle Kondik explained Trump’s strong position:

“[T]he president’s base-first strategy could again deliver him the White House, thanks in large part to his strength in the nation’s one remaining true swing region, the Midwest. He’s an incumbent, and incumbents are historically harder to defeat (although it may be that incumbency means less up and down the ticket in an era defined by party polarization). Still, Crystal Ball Senior Columnist Alan Abramowitz’s well-regarded presidential ‘Time for Change’ model, which projects the two-party presidential vote, currently projects Trump with 51.4% of the vote based on the most recent measures of presidential approval and quarterly GDP growth (the model’s official projection is based off those figures in the summer of 2020). Arguably, the state of the economy is the most important factor: If perceptions of its strength remain decent, the president could win another term. If there is a recession, his odds likely drop precipitously. Meanwhile, it’s not a given that the Democratic nominee can consolidate the votes of Trump disapprovers, particularly if a third party candidate (Howard Schultz?) eats into the anti-Trump vote.”

Making the 2020 Democratic effort even more problematic is growing evidence that some targets thought to be ripe may be more resistant to Democrats, and more supportive of Trump, than previously thought. Ron Brownstein has some new Gallup data suggesting the incumbent’s resilience in Sunbelt states where many Democrats have espied strong trends in their favor:

“Across the potentially competitive Sun Belt states, Trump’s position among whites is consistently much stronger. In particular, his support among non-college-educated whites was much higher than it was in the Rust Belt: Gallup found that he drew positive job ratings from 73 percent of these voters in Georgia, 67 percent in North Carolina, 66 percent in Texas, and 61 percent in Florida. Likewise, among college-educated whites, Trump ran well above his Rust Belt numbers in all four states.”

And that’s against a generic Democratic opponent, as opposed to the flesh-and-blood candidate who might not, as Hillary Clinton demonstrated, be as strong as his or her party hoped, after a billion-dollars-or-so of attacks from the Trump campaign and its social media/Fox News allies.

As Brownstein notes, these numbers (and the intensely pro-Trump white Evangelical voters they reflect) mean that Democrats will probably need to mobilize nonwhite voters at extraordinary levels to win Sunbelt states Trump carried in 2016. It’s not clear the kind of candidate who can exploit Rust Belt opportunities can do that. And Trump has residual areas of strength in the greater Midwest as well….

Again, this is a glass-half-empty look at how 2020 is shaping up for Democrats. But it raises an important question about how Democrats — at both the elite and grassroots level — react to the real possibility of a second Trump win, particularly if it grows more plausible as 2020 approaches. Will they calmly resolve to unite behind whoever emerges from the abattoir of the nominating process, based on their popularity among Democratic primary voters? Or will they panic and become obsessed about “electability” as opposed to any other candidate quality?

They probably shouldn’t. As my my colleague Eric Levitz has argued, “electability” is a slippery concept that often involves bad-faith efforts to tear down other candidates based on selective deployment of limited evidence. But if it looks like Trump is in a relatively good position, it may be difficult for Democrats to think about anything other than electability, as I noted late last year:

“For most Democrats, the prospect of a second Trump term in the White House is an existential threat, whereas in 2016 his initial election was a bad but implausible nightmare. A second Trump term would not only drive progressives wild with frustration and fear: It could tangibly mean enough additional Supreme Court decisions to guarantee an end to abortion rights and other cherished constitutional protections, along with a federal judiciary skewed to the right for a generation and enough backsliding on critical challenges like inequality and climate change to darken every American’s future.”

Faced with that nightmare — and ineradicable memories of that shocking Election Night in 2016 — will progressive journalists and Democratic activists neurotically look at horse-race polls every other hour and adjust their views of presidential aspirants accordingly? It’s entirely possible. So in addition to developing an exciting agenda and raising money and figuring out where on a complex primary and caucus map to deploy candidate time and other resources, 2020 Democrats need to develop, update, and document a strong case that they are a good bet to beat Trump.