Only 19 days before this year's presidential election, it is worth looking at the state of the race in Michigan, the site of Donald Trump's most surprising victory in 2016.

On paper it is tempting to say that things appear to be more or less exactly where they were at this point four years ago. If anything, Hillary Clinton's seemingly insurmountable 11-point lead in polling averages on Oct. 15, 2016, seems far more discouraging than the seven-point one currently held by Joe Biden.

One anomaly is the apparent closeness of the Senate race. If recent polling is to be believed, the businessman John James is within as much as a single point of the Democratic incumbent, Sen. Gary Peters. According to a survey jointly conducted by Siena College and The New York Times, while Michigan voters support Biden by an eight-point margin, they are more or less evenly split, 43-42, on James and Peters. (The Real Clear Politics polling average shows Peters up 5.)

This is, not to put too fine a point on it, impossible. Michigan, like many other states, has a long history of split-ticket voting, but it tends to work in the opposite direction, with many voters who break for Republicans in presidential contests continuing to support moderate Democrats in House and Senate races. (Another even starker example of this phenomenon came in 2004, a year in which voters overwhelmingly approved a Republican-backed amendment to the state constitution prohibiting same-sex marriage as John Kerry defeated President Bush.)

It is just about possible to imagine thousands of Trump voters, especially in the world of agriculture, with which both of Michigan's Democratic senators have cordial relations, supporting Peters. But the idea that James, a political outsider running against a reasonably popular incumbent in a state that has not sent a Republican to the Senate in two decades, could significantly outperform his own party's sitting president beggars belief. Trump could very well lose the state he astonished the country by winning in 2016. But he will not do so lagging behind a political unknown in a Senate race that virtually no one expected to be competitive.

Instead one of two things must be possible: Either James and Trump have the same low ceiling in the race despite voters being less enthusiastic about the opponent of the former, which some of the polling suggests, or many voters in this state continue to feel reluctant declaring their support for Trump publicly or in conversations with strangers. The reasons for the latter are as obvious as they are impossible to work around, despite the repeated assurances of pollsters that they have adjusted their methodology to factor in shy Trumpists. After the failures of 2016, it is not clear why anyone should defer to them here.

How much of this really matters, though? One of the least discussed aspects of the outcome in Michigan four years ago was the role played by Jill Stein, the Green Party candidate, in Trump's victory. Stein received some 51,000 votes here in 2016, a share much larger than Trump's eventual margin. Where are the disaffected progressive voters this year? And should we expect anything like a repeat of the 150,000 votes received by the Libertarian Party in 2016? Some 15 percent of self-identified independent or third party respondents to the Siena College poll were either undecided or unwilling to say which presidential candidate they support. The worst possible outcome for Trump in 2020 is that they coalesce around Biden.

The Trump campaign has conceded that it does not need to win all of the states the president flipped in 2016. If Florida holds, which seems not unlikely, along with Arizona (a less certain bet), he can afford to lose Michigan, Wisconsin, or Pennsylvania — any of two of these, in fact, but not all three.

If James's recent surge in polls reflects something more closely resembling the actual state of the race, Trump's people need not despair. But they cannot be complacent either.