The end of May has brought terrible news for Donald Trump, as conventional wisdom would have it. Over Memorial Day weekend, the Libertarian Party nominated two Republican ex-governors, Gary Johnson and William Weld, as its ticket for November, while Bill Kristol assured Twitter that there would be neoconservative-friendly “independent” on the ballot as well. Hillary Clinton led Trump by just 1 point in the RealClearPolitics aggregate of national polling, but as polls catch up to these events, Trump is sunk. Isn’t he?

Not so fast. First, take a look at the electoral map Trump inherits from Mitt Romney. The 2012 Republican nominee did not, of course, win enough states to become president, but where he did win, he won by comfortable margins: those are Trump’s margins now. Of the states Romney won, there were only two that he took by less than 10 percent: North Carolina and Georgia. Trump seems like at least as good a cultural fit for the Republican elements in those states as Romney was. And the states Romney won by “only” ten points—the next closest GOP margins—were Missouri and Indiana, which seem apt to be all the more enthusiastic about this year’s nominee.

Indiana and Missouri were two of the best states for Gary Johnson as the Libertarian nominee in 2012; they respectively gave him 1.91 percent and 1.57 percent of their votes. If Johnson/Weld does fully twice as well in 2016—which, for reasons to be mentioned shortly, is improbable—a 4 percent and 3.1 Libertarian vote in those states would still not stop Trump from winning them. Even a doubling of the Libertarian vote in Georgia, another place where Johnson ran ahead of his national percentage in 2012, would not tip the scales: the Libertarian Party vote would go from less than 1.2 percent of the vote to about 2.4 percent, in a state that Romney won in 2012 by nearly 8 points.

But couldn’t Johnson do much more than 100 percent better in 2016? After all, Trump and Clinton have the highest negative ratings of any major-party nominees since CBS began polling on the question in 1984. This creates an opening, if ever there was one, for another option—if not a Libertarian, perhaps a candidate with Bill Kristol’s “Renegade Party.”

Unfortunately for Kristol and Johnson, that’s not how politics works in 21st-century America: the sky-high negatives for both nominees mean there is in fact less space than usual for a third-party (or fourth-party) challenger, for the simple reason that voting against someone they hate is more important to more voters than voting for someone they like. The most important numbers for Trump aren’t ones that might attest to his popularity but those that demonstrate disapproval of Hillary Clinton. Votes against Clinton, in the abstract, could be votes for Johnson or the Kristol candidate, but in practice voters who are serious about stopping Hillary Clinton and the Democratic Party will all but inevitably vote for the major-party alternative: Trump and the GOP.

Look again at the list of states where Gary Johnson performed best in 2012: voters in places like Alaska and Wyoming had the luxury of casting their ballots for an exotic species of Republican, called a “Libertarian,” because a Republican was sure to win the state anyway. Johnson’s 2012 ticket underperformed its national average in states like Ohio, Florida, and Virginia. No battleground state appears in the ranking of places where the Libertarians did best. (A few Democratic states, where again the outcome was predetermined and Republicans might as well vote for a Republican subspecies, do make the list: this explains why Illinois is more Libertarian than Virginia.)

My small-l libertarian friends bristle at being labeled “conservatives” or “Republicans,” but at the ballot box a difference is hard to discern: the Libertarian Party has never nominated a well-known ex-Democrat to its ticket but has frequently nominated Republicans such as Ron Paul, Bob Barr, Gary Johnson, and William Weld. Wayne Allyn Root, the LP’s 2008 vice presidential nominee, is an eager Trump supporter today. And not only the nominees but also the Libertarian Party’s voters, to judge by the numbers, seem mostly to be “alternative Republicans.”

What might seem like a greater threat to Trump is a Kristol candidate targeted specifically at Virginia, whose D.C. suburbs are perhaps the only place in the country where NeverTrump Republicans could make a critical difference in November. (Other states have plenty of anti-Trump Republicans, but those states are so Republican anyway that the defections don’t matter, just as defections to the Libertarian Party don’t.) Yet it’s not clear that there is a Virginia-marketable neoconservative Republican who wouldn’t risk taking as many votes from neoconservative-friendly Clinton as from Trump. In Virginia, splitting the vote for war, NAFTA, and more immigration between Clinton and a Kristol candidate might work to Trump’s advantage, in much the same way that the divided field in the Republican primaries did.

All this only means that Trump should do as well as Romney did in the electoral college; the alt Republicans and #NeverTrump effort have little chance of costing Trump anti-Clinton votes, which is what most Republican votes are. (One of the flaws in my analysis of the Trump phenomenon early on was that I continued to believe most Republican voters were attached to their party and wanted to nominate someone “electable”; in fact, a plurality of Republican voters hates the establishment in both parties and wants to take a stand against it, even if that means nominating seemingly “unelectable” candidates like Trump or many a Tea Partier.)

Romney fell far short of beating Obama, of course, and since 2012 the country’s demographics have only moved further in a Democratic direction, as more millennials come of voting age and the white proportion of the electorate declines. Surely this dooms Trump, even if Republican divisions do not.

Except it doesn’t, not by itself. Jamelle Bouie suggests that “If Trump could reverse the yearslong decline and bring white turnout back to its 2008 levels—74 percent—then he could win with another couple percentage points among whites [more than what McCain received] … This would give him teetering Democratic states such as New Hampshire, Pennsylvania, and Wisconsin, as well as the three largest swing states: Florida, Ohio, and Virginia.”

(Higher white turnout in 2016 compared to 2012 strikes Bouie as a more plausible winning scenario for Trump than one in which Trump gets a much higher proportion of a 2012-sized white vote: for the latter to work, he would need “an increase of nearly six points over the [GOP’s] white share in 2012, matching Ronald Reagan’s performance in 1984.”)

Even where 21st-century demographics are concerned, Trump may have more of a shot than his dismal polling among young people and racial minorities suggests. Clinton is weak with young voters as well, and the tensions between Clinton’s establishment liberal supporters and the young left have already led to severed alliances and think-tank purges. Trump has an opening—if not to add young voters to his older and whiter base then nevertheless to deprive Clinton of their votes by hammering home her failures. And if Trump could win even in the Republican primaries with forceful opposition to the Iraq War and secretive trade deals like the Trans-Pacific Partnership (TPP)—causes that resonate with Sanders-leaning young voters—he stands to do better still with those positions in the general election. Clinton personifies the old consensus that Obama’s millennial vote was trying to get rid of when it embraced “Hope and Change.”

Trump has shown he’s prepared to campaign much more aggressively on foreign policy than Bernie Sanders has ever dared. And for a preview of how Trump will perform against Clinton in a debate, just recall how he performed against the GOP’s closest counterpart to Clinton: Jeb Bush. Trump will press her hard on Iraq, much harder than Sanders has done. He’ll hit her on Libya, too. Trump also won’t be any kinder than Sanders has been about Clinton’s coziness with the big banks. The young left may be in for a surprise—one that’s unlikely to lead many to vote for Trump but that may drive deeper the generational wedge between them and Clinton.

The 2016 race pits a decades-old center-left establishment against a newly invigorated populist right. That populist right has already defeated the decades-old center-right establishment of the GOP. It has a fighting chance against Clinton, if Trump sticks to his issues and doesn’t attempt to become a more generic, Romney-like Republican on questions of war and industrial policy.

As for immigration and ethno-racial politics, there could be some surprises here, too. Trump’s critics in the media identify him with the racist and anti-Semitic trolls who support him on Twitter; but millions more Americans identify the Democratic Party with the Social Justice Warriors and other activists whose antics and occasionally violent acts have been widely broadcast on national television over the past two years. If Clinton repudiates this Social Justice left, she risks further alienating her young left base; if she fails to repudiate them, she stands to alienate more of Middle America.

Hillary Clinton represents everything that Trump voters, Republicans, Sanders voters, and Middle America have come to hate: the Iraq War, secretive trade deals and job losses, suffocating political correctness, and the risk of “unrest.” The liberal establishment in both parties—free-market liberal in the GOP’s case, left-liberal in the Democrats’—has known all along how much suffering and resentment its policies have generated. But party elites imagined that none of it mattered: what could voters do, pull the lever for Bush instead of Clinton? Clinton instead of Bush? The fix was in, and had been since the first George Bush took office.

Only now, to the insiders’ dismay, voters have an establishment and an anti-establishment choice. In 2008 they selected the relatively less establishment figure, Barack Obama, in the Democratic primaries and general election alike. In 2016, voters are asked to cast their ballots for the Democrat who didn’t represent hope or change eight years ago. Is Clinton any fresher today?

Trump won’t lose any of Romney’s states. Can Clinton really hold all of Obama’s? Probably not: Ohio still has some white working-class Democrats, and Trump’s prospects of winning them seem a lot better than Mitt Romney’s ever were. Trump surprised everyone with his successes in Pennsylvania’s GOP primary; if he can do five points better than Romney in the general election there, the results will be catastrophic for Clinton. Florida remains as much of a battleground as ever: there’s no indication that any trouble with Latino voters will cost him the Sunshine State. This election is as finely balanced and close as 2012 was, and that brings it down to a referendum on the status quo of the last 20 years: should the era of Bush and Clinton continue, or is it time for something new—even if what’s new is named Donald Trump?

Daniel McCarthy is the editor of The American Conservative. //