Jonathan Ernst / Reuters

Have we learned nothing? In 2016, very few political writers, myself emphatically included, thought Donald Trump would win the Republican nomination, let alone the presidency. Very few thought Bernie Sanders would win 23 states and 13 million votes in his Democratic-primary battle with Hillary Clinton.

The voters were lousy prognosticators too. Although polls generally suggested that Sanders would fare better against Trump, voters overwhelmingly believed Hillary Clinton had a better chance of winning the general election. And in the closing weeks of the 2016 campaign, they overwhelmingly predicted that Clinton, not Trump, would triumph.

The point is that we, and they, simply don’t know. Electability is extremely hard to predict. And when pundits discuss it, they often rely on unstated and dubious assumptions—which usually lead them to predict that the most centrist candidate with the most establishment support is the person general-election voters will like best.

All this glib talk about electability has a cost. It leads commentators, often implicitly, to give “electable” candidates a pass when their policy views are fuzzy or flat-out wrong. So what should journalists do? It’s simple: Spend less time discussing which candidates can win the presidency and more time discussing what they’d do if they actually won.

To grasp how questionable much of the discourse surrounding “electability” is, consider the two candidates who, according to conventional wisdom, are considered best able to defeat Trump in 2020. The first is Joe Biden. The reason: As a Washington Post headline put it last fall, “Biden Appeals to Working-Class Whites Who Defected to Trump.”

But does he really? The evidence suggests that most of the voters who supported Barack Obama and then Trump are not Democrats who “defected” to Trump. They’re Republicans or Republican-leaning independents who “defected” to Obama and in the years since have grown ever more ensconced in the GOP. So it’s not clear that any Democratic candidate could lure many of them away from Trump next year.

Nor is it obvious that Biden would be best suited to doing so. Yes, his race and gender might prove an advantage. Yes, he might be harder to tar as a socialist radical. But Biden supported NAFTA and the Iraq War, and he’s been a Washington insider for almost a half century. In 2016, Trump voters expressed their deep pessimism about the state of the country by voting for radical disruption. If some have now lost faith in Trump, and thus grown even more disillusioned with politics than they were before, wouldn’t they look for a different species of disrupter, perhaps an anti-establishment populist such as Bernie Sanders or Elizabeth Warren? Why go for the Democratic candidate who, more than almost any other, represents the pre-Trump status quo?

And even if Biden did prove better able to win back working-class whites than his competitors, could he rouse the Democratic Party’s African American and female base? The fact that his advisers are reportedly considering asking Stacey Abrams to be his running mate suggests that they themselves have doubts. But if choosing Abrams boosts Biden among women and people of color, why wouldn’t it hurt him among white men who backed Trump?

Do these questions mean Biden is less electable than other candidates? Not at all. What they mean is that we really don’t know.

It’s the same with Beto O’Rourke. His boosters say he’s like Obama: appealing to moderate whites because of his unifying, upbeat message, but also rousing to progressives, who find him idealistic and inspirational. As one Democratic bundler told Politico, “He’s Barack Obama, but white.” But the white part matters. You can’t assume O’Rourke is more electable than Kamala Harris or Cory Booker without explaining how O’Rourke could match the epic African American turnout numbers that Obama elicited in 2008 and 2012, but that Clinton did not match in 2016.

Moreover, saying that O’Rourke would appeal to “moderate” or “centrist” whites glosses over a critical distinction: It depends on which “moderates” we’re talking about. O’Rourke’s cultural liberalism, pro-business background, and unifying, optimistic rhetoric might serve him well among the upper-middle-class Democrats and independents who admire Michael Bloomberg. But is a candidate who has backed Trade Promotion Authority and praised NAFTA best suited to winning back working-class voters in the industrial midwestern states that gave Trump the presidency? In a recent interview with Thomas B. Edsall of The New York Times, Paul A. Sracic, a political scientist at Youngstown State University, suggested that “O’Rourke’s vague, ‘We all need to come together’ message will not resonate with people who see life as a battle. Working class voters believe in pugilistic politics.” Why is Sracic’s take less plausible than that breathless Democratic bundler’s?

Anyone can play this game. Maybe Sanders is the most electable because his pugilism can win over anti-establishment, anti-corporate Trump voters while eliciting a vast turnout among Millennials? Maybe Warren is the most electable because she’s as passionate as Bernie but more substantive and less radical, and she’ll inspire women as well? Maybe Harris is the most electable because she can replicate Obama’s massive African American numbers while pivoting to the center in a way white candidates can’t? Maybe Booker is the most electable because his message is as positive and unifying as O’Rourke’s, but he’ll do better among African Americans, and his unabashed religiosity will prove a secret weapon with evangelicals?

All these narratives are superficially plausible, and all of them could be nonsense. No one knows. And by embracing some while dismissing others, journalists—sometimes unwittingly—create a double standard for evaluating candidates. It’s fine that O’Rourke is less substantive than Warren, political handicappers imply, because he’s better able to beat Trump, which is what matters most. Let’s not credit Sanders for opposing the Iraq War, which Biden supported, because it’s better to nominate a more hawkish Democrat who can win than a dove who will lose.

The irony is that many political commentators think it’s easier to have an informed opinion about electability than about policy. It’s actually harder. If you want to know which candidate was correct about deregulation or the Iraq War, or whose health-care plan will cover the most people at the lowest cost, you can talk to experts, assemble facts, and come to a reasonable conclusion. Deciding which candidate can best beat Trump is more like looking at someone’s zodiac sign and predicting his or her future. It may be a fun hobby, but it’s a distraction from useful work.

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