If it happens, ending the electoral college will become as fixed and zealous a Democratic credo as abortion on demand. They’re already 80 percent of the way towards making it party orthodoxy after 2000 and 2016. To lose twice in a row to a character like Trump despite winning more overall votes would convert it into bedrock liberal dogma.

How does Trump lose by five million votes and win a second term? Simple, argues analyst Dave Wasserman. Although the country is changing demographically in ways that benefit Democrats electorally, those changes happen to be concentrated in states that are either already safely blue or so safely red — for the moment — that there’s little chance of flipping them in 2020. The two obvious examples are California and Texas. The Democratic nominee’s going to run up the score in Cali next fall, winning the state by many millions of votes (particularly, perhaps, if state native Kamala Harris is the nominee). And Democrats are likely to make gains in Texas as the electorate there turns more Latino, further padding the nominee’s national popular-vote total. But unless there are enough gains to actually turn Texas blue, which is unlikely, so what? Dems will enjoy a windfall of popular votes in both states and it won’t matter a lick in terms of electoral votes.

Where they need gains, of course, is in the Rust Belt. But there’s cause for pessimism, says Wasserman:

Democrats’ potential inefficiencies aren’t limited to California and Texas: The list of the nation’s top 15 fastest-diversifying states also includes the sizable yet safely blue states of New York, New Jersey, Massachusetts, Maryland, Washington and Oregon.

Meanwhile, demographic transformation isn’t nearly as rapid in the narrow band of states that are best-positioned to decide the Electoral College — a factor that seriously aids Trump.

In 2016, Trump’s victory hinged on three Great Lakes states he won by less than a point: Michigan (0.2 percent), Pennsylvania (0.7 percent) and Wisconsin (0.8 percent). All three of these aging, relatively white states have some of the nation’s highest shares of white voters without college degrees — a group trending away from Democrats over the long term. And the nonwhite share of the eligible electorate in each of the three has increased at only a quarter to a half of the rate it has surged in California, Texas and Nevada.

There are other diversifying states that work better for Dems. Arizona is basically purple now, having broken for Trump by just three and a half points in 2016 and then elected a Democrat to Jeff Flake’s vacant seat two years later. Yuppies moving to North Carolina will keep that state competitive for liberals too. And of course there’s Florida, although that’s been trending Republican for two straight elections thanks to migration by white retirees. The hard truth for Dems, says Wasserman, is that even if they manage to flip Michigan and Pennsylvania, partially restoring the “blue wall,” Trump can still win a second term by holding Wisconsin and every other state he won in 2016 plus Maine’s 2nd congressional district. 271-269. Enjoy the popular-vote consolation prize, lefties. Again.

Question, though: How easy would it be for Trump to hold Wisconsin if nearby red states like Michigan and Pennsylvania are turning blue? Easier than you think, argues Nate Cohn in another piece foreseeing a sizable Democratic popular vote win and a narrow electoral college win for Trump.

Wisconsin was the tipping-point state in 2016, and it seems to hold that distinction now, at least based on the president’s approval rating among 2018 midterm voters.

Over all, the president’s approval rating was 47.1 percent in Wisconsin, above his 45.5 percent nationwide. This implies that the president’s advantage in the Electoral College, at least by his approval rating, is fairly similar to what it was in 2016…

In [fact], most measures suggest that the president’s rating is higher than 47.1 percent in Wisconsin. If you excluded the Votecast data and added the final Marquette poll, the president’s approval rating would rise to 47.6 percent — or a net 4.2 points higher than his nationwide approval.

At the end of the day, says Cohn, Trump’s weak national job approval doesn’t matter since we don’t hold truly national elections. We hold 50 state elections, and in the state that’s most likely to be decisive in 2020 — Wisconsin — his approval is several points better than it is nationally and within spitting distance of 50 percent. Not only that, the issues Trump is hammering are smartly tailored to appeal to the decisive voters in the Rust Belt, the people who turned those states blue for Barack Obama in 2012 but then turned them red for Trump four years later:

The Obama-Trump voters are basically Republicans on immigration, or at least much closer to Trump’s position than they are to the Democrats’. Relatedly, there’s also evidence that voters who turned out for Obama in 2012 but then stayed home in 2016 are more moderate than the average Democratic voter on health care, another top issue in this campaign, making the progressive push to end private health insurance dangerous for Dems in swing states. Between that and the immigration data from Cohn’s piece, some lefties are beginning to worry:

There have been lots of lame and half-baked takes about the Democrats "blowing it" or "moving too far left", but this is the best empirical case I've seen that the voters most likely to tip the electoral scales are quite conservative on immigration. https://t.co/5B9RrwBF2L

— Chris Hayes (@chrislhayes) July 19, 2019

The Democratic nominee’s going to sound a lot more moderate on immigration during the general election campaign than he or she does right now, I’m guessing. As for Trump, says Cohn, “A strategy rooted in racial polarization could at once energize parts of the president’s base and rebuild support among wavering white working-class voters” who turned out for him in 2016 but skipped the midterms last fall. The clearer this becomes to POTUS, the more eagerly he’ll embrace the “send her back” messaging. He’s already begun to, after all.

All in all, Cohn thinks Wasserman’s being too cautious in estimating how badly Trump could lose the popular vote and still win the presidency. Wasserman thinks the margin could be five million votes; Cohn thinks it could be five points, which would mean something like six and a half million votes. Hoo boy.

But wait. That other guy named Nate wants a say here too:

Basically, I think Clinton 2016 was probably close (maybe not exactly *at*) the local minimum for D's. So a coalition that *either* "turned back the clock" (Biden?) *or* turboboosted demographic change (Harris/Castro?) could be better than Clinton 16, which was stuck in between. pic.twitter.com/yWkWsLCjYt

— Nate Silver (@NateSilver538) July 20, 2019

It’s just too easy to say Wisconsin looks solid-ish for Trump when a variable as crucial as the identity of the Democratic nominee remains unknown, says Silver. If they nominate Biden, maybe he can reassemble the Obama coalition and turn some of those Obama-to-Trump voters back to blue. He wouldn’t need to flip many to flip the entire state of Wisconsin with them. Or, Silver wonders, what if everyone’s underestimating how many younger and minority voters in big cities might show up for a nominee like Kamala Harris? Remember, everyone thought Texas was “safe” for Ted Cruz last fall despite the hype for Beto; Republicans ended up sweating it out on election night as unexpected waves of Democratic voters turned out for O’Rourke, nearly handing him the election. Trump won’t start with nearly as much of an advantage in Michigan, Pennsylvania, Wisconsin, Arizona, Florida, and North Carolina as Cruz did in Texas. Beware overconfidence.