A crew removes part of a border wall near Tijuana, Mexico.Gregory Bull / AP

President Donald Trump may now be talking more about steel than cement, but his proposed border wall remains the Rosetta Stone for understanding both his conception of the presidency and his political strategy.

Nothing better illustrates Trump’s political calculus than his determination to build the wall, a goal that most Americans consistently oppose in polls, even at the cost of shutting down the federal government, a tactic that surveys show most Americans also consistently reject.

Politically, the showdown over the shutdown demonstrates how much more Trump prioritizes energizing and mobilizing his passionate base, often with messages that appeal to anxiety about demographic and cultural change, over broadening his support toward anything that approaches a majority of the country. It sends the same message about his priorities in executing his office. Trump makes no pretense of governing as president of the entire nation. Instead, he governs as the champion of his slice of America against all the forces in the country his supporters dislike or distrust—an instinct he displayed again this week with his latest threats to cut off disaster-relief funding for California.

For a president to consistently steer his governing agenda and political messaging toward a demonstrable minority of the country is, to put it mildly, a novel strategy. But Trump may feel comfortable playing on the short side of the field because it’s worked for him before. In the 2016 election, a majority of voters said they had an unfavorable view of him personally and did not believe he had the experience or temperament to succeed as president, according to the Election Day exit poll conducted by Edison Research. And yet Trump, of course, won a slim Electoral College majority despite losing the popular vote by nearly 3 million. A critical question looming over 2020 is whether he could squeeze out another victory while facing opposition from a majority of the country.

The struggle over the border wall actually provides a revealing gauge of Trump’s prospects on that front. From the start of his presidential campaign, immigration, more than any other issue, captured the potential benefits of Trump’s strategy of pursuing depth of support over breadth. Even during the 2016 Republican primaries, a majority of voters opposed deporting all undocumented immigrants in all but two of the 20 states where exit polls asked their opinion. Yet the minority of voters that supported deportation backed Trump in such preponderant numbers that they provided a majority of his votes in all but two of the 20 states.

The same pattern was evident in the general election. In the exit poll, just 41 percent of voters supported Trump’s border wall, while a solid majority of 54 percent opposed it. Yet Trump won a much higher share of the wall’s supporters (85 percent) than Hillary Clinton did of the wall’s opponents (76 percent). Roughly one-fourth of the wall’s opponents either voted for Trump (16 percent) or drifted away to a third-party candidate (8 percent).

That disparity reflected a clear trend in the exit poll. On questions about Trump’s personal characteristics—such as whether he had the experience or temperament to succeed as president—he consistently won a higher percentage of those who said “yes” than Clinton won among those who said “no.” That pattern may have reflected doubts about Clinton, a willingness to roll the dice on a political outsider, or a desire for change. But whatever the cause, the pattern was decisive: The pivotal votes that made Trump president came from voters ambivalent at best about him and key elements of his agenda.

After two years of arguing for the wall as president, Trump has shown no ability to expand its popularity. In 10 national polls conducted during his presidency, Quinnipiac University has never found support for the wall higher than 43 percent. With his Oval Office address on Tuesday night, Trump may have further consolidated support among Republicans and conservatives, which could raise that number slightly. But his focus on grisly portrayals of undocumented immigrants is unlikely to dent the preponderant opposition to the wall among all the groups that powered the Democratic takeover of the House last fall: minorities, young people, and college-educated white voters. Depending on the survey, the wall usually faces opposition from at least two-thirds of minorities and young people, and between three-fifths and two-thirds of college-educated whites.

Looking back at 2016, Trump may conclude that lopsided opposition from those groups doesn’t matter so long as he maintains strong support for the wall from his base. But there’s evidence that the voters hostile to the wall, and to many other aspects of Trump’s tenure, are less willing to give him the benefit of the doubt now than they were in 2016.

To test that proposition, I compared results from the 2016 exit poll with findings from Quinnipiac’s latest national survey on the wall in December. That comparison shows that Trump’s position among wall opponents has eroded dramatically.

Overall attitudes about the wall in the December poll remained similar to 2016, with 43 percent supporting and 54 percent opposing. (Other surveys, including Quinnipiac’s, have usually found support slightly lower, at about 40 percent.) Among whites both with and without a four-year college degree, opinions on the wall in the Quinnipiac survey showed only minimal change from the exit poll, with nearly three-fifths of whites with a degree opposing the wall and nearly three-fifths of those without a degree supporting it.

But while overall attitudes about the wall haven’t changed much, attitudes toward Trump have deteriorated significantly among the wall’s critics. Opposition to the wall, just like doubts about Trump’s personal characteristics, was not a deal breaker for a significant share of voters in the presidential election. In the exit poll, 18 percent of the college-educated whites who opposed the wall voted for Trump anyway, according to figures provided by Edison Research. But now, far fewer express support for Trump in general. In the latest Quinnipiac poll, just 3 percent of these voters approved of Trump’s job performance, according to data provided by Quinnipiac. Ninety-two percent disapproved.

Likewise, just over one-fourth of non-college-educated whites who opposed the wall still voted for Trump in 2016. But in the latest Quinnipiac survey, only 9 percent of these whites approved of Trump’s performance, while 83 percent disapproved. In all, fully 88 percent of Americans who oppose the wall say they disapprove of Trump’s performance as president.

Approval ratings correlate closely with the reelection vote for incumbent presidents. So that huge disapproval number suggests that the 2020 Democratic nominee could win a considerably higher share of wall opponents than the 76 percent who voted for Clinton in 2016. By contrast, the share of wall supporters who approve of Trump in Quinnipiac’s latest poll is about the same (83 percent) as his vote among that group in 2016 (85 percent). That means opponents of the wall are now consolidating against Trump’s overall performance at an even higher rate than its supporters are coalescing behind him. That’s a very different landscape than in 2016, and one that springs directly from two years of governing in a volatile, confrontational manner aimed almost entirely at his base, the strategy that he’s escalated again by shuttering the government over the wall.

It’s possible, of course, that with slim victories in key Rust Belt and Sun Belt states, Trump can squeeze out another Electoral College victory even if he loses the popular vote again. His chances will rise if a third-party alternative again divides the voters who are resistant to him.

But these patterns of public reaction suggest that Trump’s relentless effort to cement the loyalty and stoke the outrage of his strongest supporters, compounded by his volatile behavior in office, is building a wall between him and the ambivalent voters who provided him critical support in 2016 (or at least withheld it from Clinton by splintering to third-party candidates). The sharp movement toward Democrats in the midterm elections among independents and college-educated white voters, both groups that broke closely for Trump in 2016, points toward the same conclusion.

Trump’s monomania on the border wall shows that he remains fixated on the priorities and resentments of his core coalition. But even a 30-foot barrier probably wouldn’t protect him in 2020 if he allows the waves of discontent to continue rising among the majority of Americans who don’t consider themselves part of that ardent club.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.