Political parties are the arbiters in democracies. Peter Berger, a sociologist, call them the dams that hold at bay the howling frenzies lurking in the human soul. But, “All institutions are fragile,” he writes in The American Interest. “Sometimes the dams break” and you get someone like Jeremy Corbyn or Donald Trump.

The former, a unrepentant Marxist and peacenik, recently won the British Labour Party’s leadership election. The latter, a loudmouthed real-estate mogul, now tops the polls for the Republican Party’s presidential nomination in the United States.

Corbyn isn’t going to win power in a country that only four months ago gave David Cameron’s Conservatives their first parliamentary majority in twenty years. Nor is Trump likely to secure the Republican nomination, let alone win the 2016 election.

But the fact that they’ve got this far calls into question the theory that “the party decides.”

Postulated in 2008 by political scientists Marty Cohen, David Karol, Hans Noel and John Zaller, the idea that party elites rather than ordinary voters typically decide the outcome of internal elections has fast become the conventional wisdom.

They argued that behind the scenes, donors, elected officials and party insiders — the people who pay the most attention and who have the most at stake — decide whom they are going to support over the course of several months. Their decision isn’t made in smokey rooms. It is made by individuals, factions, lobby and interest groups. Sometimes, the party doesn’t reach a consensus until after a few debates and primary elections have tested the candidates. But in the end, the “establishment” always finds itself rallying around a single person and it signals its preference by making endorsements.

The Atlantic Sentinel shares this view. It is why we urged readers in July not to pay too much attention to Trump and why argued earlier this month that Hillary Clinton is still almost certainly going to be the nominee on the Democratic side, despite her recent slump in the polls.

The reason is simple. Trump doesn’t advance the Republican agenda. He has taken positions that are at odds with conservative orthodoxy while his bravado and bluster threaten to alienate middle-class and Hispanic voters, the very swing constituencies Republicans need to woo if they are to win back the White House next year.

Clinton, by contrast, is almost the generic Democrat. She is moving a little to the left of President Barack Obama on issues like wages and trade and a little to his right on foreign policy. That is where most Democratic officials and most Democratic general-election voters are. It’s not necessarily where most Democratic primary voters are, though, and that may be why Bernie Sanders, the socialist senator from Vermont, is besting her in the early-voting states of Iowa and New Hampshire.

Similarly, on the Republican side, primary voters tend to be more right-wing than Republican voters in general who, in turn, are more right-wing than the country.

It’s the party’s job to restrain its most fringe elements and keep their members’ eye on the prize: winning elections.

FiveThirtyEight‘s Nate Silver — another party-decides believer — recently argued that the party has a lot of ways in which it can influence the selection process. It appoints superdelegates, schedules debates and has hundreds of millions of dollars to spent on advertising.

And voters have a lot of time to make their decisions and can amend them as they go along — an insurgent candidate who wins Iowa or New Hampshire won’t necessarily have staying power if they’ve failed to build a broad coalition of support.

This makes the American primary system different from other countries and it helps explain why the system failed in the United Kingdom.

Their version of Bernie Sanders did win the Labour Party’s leadership election — because the party broke down.

Labour did not expect to lose the general election in May, or at least it did not expect to lose as badly as it did. When Ed Miliband suddenly resigned days after, there was no heir apparent waiting in the wings. The two middle-of-the-road candidates, Andy Burnham and Yvette Cooper, were uninspiring. Many Labourites were still smarting from their second election defeat in a row and not looking for a safe pair of hands. Nor were they ready to listen to the (relatively) right-wing candidate, Liz Kendall, who kept pointing out that the Conservatives had won the election by swaying Middle England.

What Labour needed was post-traumatic stress therapy. It wanted to be told that there was nothing wrong with its convictions or its left-wing policy platform; that voters hadn’t rejected them but simply not been persuaded hard enough.

Along came Jeremy Corbyn, a relic of 1970s Britain, who told them exactly what they wanted to hear. No matter the election and the exit polls, Labour didn’t lose because it was too far to the left, he said. It lost because it wasn’t firmly left-wing enough!

The party, in disarray, failed to rally behind a single candidate to stop Corbyn. The trade unions, who have a disproportionate influence in the Labour Party, actually backed him. And rather than restrain the fringe left, Labour welcomed them in by allowing anyone to vote in the primary for as little as paying £3. “Sort of like getting entrance to a members-only club by paying at the door,” writes Berger.

Corby ended up winning because he got majority support from the trade unions and as much as 83 percent support from the entrants.

The party will have its revenge, though. The dams will be rebuilt. Only a handful of Labour’s parliamentarians backed Corbyn in the first place. It is likely that he will be forced out within the next few years. If not, Labour will lose another general election and then the party elite must be able to persuade members that the road to electoral success runs through the political center, not the far left.

As the British would say, the establishment always wins. Sooner or later anyway.