Since the photo-finish 2000 presidential election between George W. Bush and Al Gore that introduced the United States to the red-and-blue political map, the struggle between the parties for 270 Electoral College votes has resembled trench warfare. With most states locked down for one party or the other, both sides have overwhelmingly concentrated their resources on the same handful of swing states.

Those traditional battlegrounds, led by Florida and Ohio, are still attracting enormous efforts from the two campaigns. But this year, the geographic stalemate could finally be broken—or at least disrupted—by the sharply divergent demographic coalitions of support that Donald Trump and Hillary Clinton have assembled in their ferocious contest.

Tonight’s results could crystallize a reconfigured electoral order in which Republicans rely increasingly on preponderantly blue-collar, white, and older Rustbelt states that have mostly favored Democrats in recent years, and Democrats depend on white-collar, diverse, and younger Sunbelt states that as recently as the 1990s leaned reliably toward the GOP. If Trump does as well as seems possible in the Rustbelt, and Clinton still defeats him with big gains in the Sunbelt, this may be remembered as the fast-forward election that compressed into a single cycle those demographic and geographic changes that most people expected to unfold over a decade or more.

Stability has been the biggest story in the political map since 2000, when Bush beat Gore with the second-narrowest Electoral College margin ever. Including that 2000 race, fully 40 states, representing 422 Electoral College votes, have supported the same party in each of the past four presidential contests, according to calculations by longtime political analyst Michael Barone. That means 80 percent of the states have voted the same way for four consecutive elections. That’s a remarkable level of consistency, especially given that that period includes two victories for Bush, a Republican, and two for President Obama, a Democrat. (Obama’s 2008 win provided most of the exceptions to the pattern.) The level of continuity is even higher than when Republicans won the White House four straight times from 1896 through 1908 with William McKinley, Theodore Roosevelt, and William Howard Taft—about three-fourths of states voted the same way each time—or when Democrats won the White House four straight times with Franklin Delano Roosevelt from 1932 through 1944—with about two-thirds voting the same way.

This largely stable map has shifted the Electoral College advantage from Republicans to Democrats. The hinge point was the election of 1992. In the six elections from 1968 through 1988, Republican presidential nominees enjoyed an advantage in so many states that analysts in both parties talked of a GOP “lock” on the Electoral College. In fact, over those six elections, the Republican presidential nominees averaged nearly 420 Electoral College votes, which meant they averaged a landslide. Both Richard Nixon in 1972 and Ronald Reagan in 1984 crushed their hapless Democratic opponents in 49-state landslides. During the three elections of the 1980s, Democrats won a smaller share of the available Electoral College votes than they did in any three-election sequence since the formation of the modern party system in 1828.

But since 1988, no Republican nominee has carried more than the 286 Electoral College votes Bush won in 2004. Democrats have soared far beyond that number four times in this period, with Bill Clinton winning 370 and then 379 Electoral College votes, and Obama tallying 365 and then 332.

The core of the Democrats’ advantage since 1992 has been what I first called in 2009 the “blue wall.” Those are the 18 states that Democrats have carried in at least the past six consecutive elections. Those 18 states, plus Washington, D.C., offer 242 Electoral College votes and provide Democrats a much larger foundation than Republicans have been able to rely upon. Over the past six elections, Republican nominees have won 13 states, worth just 102 Electoral College votes, every time. Any Democratic plan to reach an Electoral College majority starts with defending the blue wall.

The blue wall includes 10 of the 11 states from Maryland to Maine, excluding only New Hampshire; Michigan, Wisconsin, Minnesota, and Illinois in the upper Midwest; and California, Oregon, Washington, and Hawaii in and along the Pacific. These are almost universally states that reflect the modern Democratic coalition, with large populations of minorities or college-educated whites or both; the big exception is the heavily blue-collar Rustbelt states—particularly Michigan, Wisconsin, and Pennsylvania—that Democrats have defended by consistently running slightly better among whites without a college education than they do elsewhere.

Since 1992, Republicans have rarely come close to breaching the blue wall. In the past six elections across these 18 states, the Republican nominee has finished within five points of the Democrat in just 10 of those 108 chances. In 2012, Mitt Romney’s 5.4-percentage-point loss in Pennsylvania was the closest he came to Obama in any blue-wall state; the only other states in the wall that Romney lost by less than double digits were Wisconsin, Minnesota, and Michigan.

If Democrats can defend the blue wall and hold all of its 18 states again, it will be the most states any party has won in seven consecutive elections since the modern-party era that began in 1828. Trump has seriously tried to dislodge only three of the blue-wall states: Michigan, Wisconsin, and Pennsylvania. Each of those contains large numbers of the white, blue-collar voters underpinning Trump’s support.

For most of the campaign, Trump has focused much more on Pennsylvania than he has on Michigan or Wisconsin; he’s visited the Keystone State more often, and spent more on television advertising there, than the other two combined. But as Trump searches for pathways to reach 270, he’s devoted more attention in the final days of the race to Wisconsin and Michigan; several late polls showed him closing ground in the latter state with Hillary Clinton remaining safely ahead in the former.

Hillary Clinton was concerned enough to schedule a late get-out-the-vote rally in Detroit on Friday, to return to Grand Rapids on Monday, to dispatch President Obama on Monday to Ann Arbor, and to add television spending in both Michigan and Wisconsin. But her campaign has mostly treated Wisconsin, which she hasn’t visited since the Democratic primary, and Michigan, which before Friday she had visited only twice since June 1, as done deals—a choice that Trump is testing in the campaign’s last days.

Pennsylvania, by contrast, Hillary Clinton has treated as a lynchpin of her Electoral College strategy: She’s spent more on advertising there than anywhere except Florida and Ohio, and no state has commanded more of her late attention. The campaign’s saturated final weekend in Philadelphia included separate concerts by Stevie Wonder and Katy Perry, and a rally Monday night in the city featured Hillary Clinton, Bill Clinton, the president, and first lady Michelle Obama—not to mention Bruce Springsteen. Hillary Clinton’s extreme focus on Pennsylvania partly reflects the fact that it, unlike most other battleground states, allows no early voting. But it also shows how much she believes that holding Pennsylvania would complicate Trump’s path to an Electoral College majority.

After the blue-wall states, the Clinton camp has considered four other states its most straightforward path to victory: Virginia, with 13 Electoral College votes; Colorado, with nine; New Mexico with five; and New Hampshire, with four. If she defends the blue wall and adds those four states, she’ll have 273 Electoral College votes no matter what happens anywhere else—and still have 272 if Trump wins the rural Second Congressional District in Maine, one of two states that apportions some Electoral College votes by district.

The prominence of Virginia, Colorado, and New Mexico on Hillary Clinton’s list reflects what may prove the most lasting geographic legacy of the 2016 election: a radical acceleration of long-term shifts between the Rustbelt and Sunbelt.

Of the 11 states that both parties have treated as battlegrounds in recent years, five run across the Rustbelt: Michigan, Wisconsin, and Pennsylvania—from the blue wall—as well as Ohio and Iowa. New Hampshire, a sixth battleground, is geographically distinct but demographically similar, with a mostly white population (albeit one that leans more toward college-educated whites than these other five).

The second group of swing states runs across the Sunbelt: Virginia, North Carolina, and Florida in the Southeast, and Colorado and Nevada in the Southwest. New Mexico had also been considered a swing state around the turn of this century, but has since tilted more securely toward the Democrats (although Trump is making a late play there).

These two baskets of battlegrounds are almost exactly equal in size: the five Sunbelt swing states offer 72 Electoral College votes, and the five Rustbelt states 70—with New Hampshire adding another four and New Mexico five.

Since Democrats began their modern run of winning the popular vote in five of the six presidential elections after 1992, the Rustbelt states have been more secure terrain for them. Democrats have won the five Rustbelt swing states in 27 of the possible 30 chances over those six elections, compared with just 13 of 30 in the five Sunbelt swing states. (In addition, the Democrats have won both New Hampshire and New Mexico five times in the past six elections.) Over those six races, Democrats have averaged a higher share of the vote in all of the Rustbelt swing states than any of the Sunbelt swing states, though with only a small gap between Ohio, their weakest Rustbelt state, and Florida, the Sunbelt state where they have performed best over that span.

This year, though, that order is rapidly resetting. Trump’s visceral connection with older, non-urban, and especially blue-collar whites has rattled the foundations of the Democrat’s Rustbelt pillars. For several days last week, the ABC News/Washington Post tracking poll showed him matching or exceeding Ronald Reagan’s margin in 1984 with both non-college-educated white men and women. (The final NBC News/Wall Street Journal poll on Sunday also showed Trump exactly matching Reagan’s 32-percentage-point advantage among all non-college-educated whites.) Polls have consistently provided Trump a clear edge in Iowa, which Democrats have won five of the past six times, and Ohio—which both Bill Clinton and Obama won twice. Though, with several late surveys tightening, Hillary Clinton has clearly not conceded the latter. While Democrats feel more secure about Wisconsin and maintain a somewhat narrower but steady advantage in Pennsylvania, most late polls show Trump battering at the gates in New Hampshire and tightening the race in Michigan as well. He’s even gestured toward Minnesota. Even if Trump doesn’t ultimately tip any Rustbelt battlegrounds beyond Iowa and perhaps Ohio, he seems guaranteed to improve on Romney’s performance in most, if not all, of them.

By contrast, Hillary Clinton’s prospects generally look brighter across the Sunbelt, not only in the familiar battlegrounds, but also in new ones that Trump’s weakness with Latino, African American, and college-educated white voters has put into play.

Though Virginia and Colorado leaned Republican in the 1990s, and were considered closely balanced swing states from 2004 to 2012, Hillary Clinton has led steadily in both of them throughout this year. Both states embody the modern Democratic coalition of minority voters and college-educated whites concentrated in major metropolitan areas. The campaign has felt so secure about each state that it decided not to invest meaningful time or money in them. That may have been premature, especially in Colorado, which maintains a significant blue-collar and non-urban vote, and where some late polls have shown Trump narrowing the contest. But overall, Democrats remain secure about Virginia and cautiously confident about their ability to hold Colorado.

Elsewhere in the Sunbelt, Nevada increasingly looks like Hillary Clinton’s best insurance policy if she loses any of the core states she’s relying on. Bolstered by heavy Latino turnout, Democrats have amassed a formidable advantage in early voting there that has led some analysts to say the state is already virtually out of reach for Trump on Election Day. (If Trump loses, the defining image may be the long lines waiting to vote outside a Mexican grocery store in Las Vegas on Friday night as the deadline for early voting loomed.)The early-vote advantage may matter more in Nevada than elsewhere, because early-voting comprises such a large share of the total vote—about 70 percent in 2012—but also because as few as 400,000 more people might vote on Tuesday. That leaves less opportunity to overcome an early-vote deficit than in, say, Florida or North Carolina, where millions of people will still vote on Election Day.

Those two behemoth states have drawn enormous focus from both campaigns: Clinton and Trump alike have spent far more money in Florida than in any other state—though Clinton has outspent him on television advertising there by nearly three to one. They have also bombarded North Carolina with ads and visits; President Obama has campaigned in the Charlotte area so often he may have established legal residency to vote in its next city-council election.

Right now, Florida and North Carolina look like the closest toss-ups on the entire map. Late public polls in both states showed Clinton either narrowly ahead or the two candidates in a tie. As Election Day approached, Democrats were cheered by early-voting results in Florida, which has seen twice as many Latinos vote early as in 2012, including more than third of whom did not vote at all that year, according to University of Florida professor Daniel Smith. (Incredibly, over six million people have already voted in Florida.) The Trump camp, conversely, was encouraged by the early-voting results in North Carolina.  Each state will test offsetting dynamics: Will Hillary Clinton improve enough among college-educated white voters to offset what are likely to be Trump gains among non-college-educated whites? And can increased turnout and margins for her among Hispanic voters, especially in Florida, offset any decline in turnout among African Americans without President Obama on the ballot? The answers may make for very thin margins between Hillary Clinton and Trump.

Maybe the best measure of the Sunbelt’s changing dynamics is the stirring of genuine competition in Arizona and Georgia, two states that neither side had expected to be fully competitive when the year began. Since 1992, Democrats have averaged only slightly more than 44 percent of the vote in Georgia and slightly less than 44 percent in Arizona; they’ve won both states only once during that period, each time when Ross Perot’s third-party candidacy divided the vote. But the same mix that has improved Democratic prospects in other Sunbelt states—antipathy toward Trump among Latino and African American voters, and Clinton’s opportunities for inroads among college-educated whites, especially in Arizona—have encouraged Hillary Clinton to make a late play in both places.

Democrats put more effort into Arizona with both her and running mate Tim Kaine visiting the state during the campaign’s last week. Bernie Sanders campaigned there on Sunday as well. Some party strategists believe their prospects might actually be better in Georgia, largely because she will win a larger share of African Americans there than she will among Latinos in Arizona (even if she seems likely to receive a bigger margin and more turnout from that community than President Obama saw in 2012). As a more distant prospect, it will be worth watching whether Clinton can make any progress at reducing the big Republican advantage in Texas, another state gradually being reshaped by the same demographic forces. In Georgia and Texas alike, the key hurdle for Democrats is their inability so far to make the same inroads among college-educated whites that have benefited them almost everywhere outside of the South, including potentially Arizona this year.

The odds remain against Clinton tipping Arizona, Georgia, or especially Texas—just as they do against Trump toppling Michigan, Wisconsin, or Pennsylvania. Yet it will nevertheless be an important sign if each candidate improves on their party’s typical performance in those states. As Republicans rely more on their “coalition of restoration” centered on older, blue-collar, religiously devout, and non-urban whites, it seems inevitable they will focus their efforts more on Rustbelt battlegrounds where those voters are plentiful. Meanwhile, as the Sunbelt states grow even more diverse—and in most cases also experience rising education levels—they will grow more essential to Democrats who depend on their competing “coalition of transformation,” which revolves around Millennials, minorities, and both college-educated and secular whites. That future alignment may become increasingly visible in this election, even if relatively few states actually flip from one party to the other.

This geographic reordering is a long-term trade most Democrats would accept, because the Sunbelt states are generally adding population and Electoral College votes just as many of the Rustbelt states are shedding both. The risk for Hillary Clinton is that her party’s foundation in the Rustbelt is fracturing before the twin forces of diversity and rising education levels have advanced enough to provide Democrats a secure foothold in the Sunbelt.

The best-case scenario is she holds enough working-class whites to defend the Rustbelt states Trump has targeted, and attracts enough college-educated whites and minorities to tip most of the Sunbelt battlegrounds. The worst-case scenario for her is that Trump’s blue-collar blitz narrowly pushes him past her in some of the Rustbelt states she needs, while she cannot advance quite enough among minority and college-educated white voters to overcome his non-college-educated, non-urban, religiously devout coalition in Sunbelt states like North Carolina, Florida, Nevada, and Colorado, much less Arizona and Georgia. Transitioning between her party’s past and future, Hillary Clinton’s nightmare is that she might be caught awkwardly in between.

The late polls suggest Hillary Clinton can avoid that fate: They show her holding enough states from both the Rustbelt and Sunbelt battlegrounds to repel Trump’s fierce final offensive. But Trump has pushed the Democrats’ post-1992 advantage in the Electoral College to the breaking point. In the campaign’s final days, the splintering of the door on the castle keep of Hillary Clinton’s core states is practically audible. Democrats are now nervously watching whether she can defend the ramparts for another day.


Assistant editor Leah Askarinam contributed.