As the new year dawned, the top editor at The Associated Press – one of the most influential journalists in the country – was making a resolution.

“We should all resolve to spend less time, or perhaps no time at all, on horse race polls that project forward to the 2020 presidential election,” Sally Buzbee said on CNN’s “Reliable Sources.”

The Atlantic’s James Fallows responded to her pledge by tweeting an Amen: “No poll right now is worth anything. Who ‘should’ run? Anyone who wants to try. The best way to tell who’s ‘likable’/’electable’ is to listen to them and see how primaries turn out.”

Buzbee herself said she doubted news organizations, including The AP, would heed her words.

As the presidential election season kicked off in earnest this month, it was obvious the media would do what it always has done: focus on personalities and electability, and blow gaffes way out of proportion. (I researched this in early 2016 while New York Times public editor; in a fairly typical two-week period, three out of every four pieces of political journalism was horse race coverage.)

You can see it happening again, already. News organizations were agog at Elizabeth Warren’s video in which she began to lay out her substantial ideas but then awkwardly declared a brief pause – “I’m gonna get me a beer.”


CBS News, among many others, found this last part newsworthy, tweeting: “Sen. Elizabeth Warren drank a beer on Instagram Live – and it received mixed reactions.” Boston Herald columnist Jaclyn Cashman (amusingly) teed off on it.

Nearly every news organization published articles on Warren’s personality and gender, comparing her to Hillary Clinton. Then commentators wrote meta-pieces on all this. But none of that had much to do with acquainting voters with what Warren stands for.

The press critic and New York University professor Jay Rosen may have a better idea. He says the mission of pre-election coverage should not be to determine who’s going to win, but to respond intelligently to what voters want and need to know. It’s so simple it seems absurd that the reality is so far from it.Rosen writes on his PressThink blog that one of the problems with election coverage as it stands is that no one has any idea what it means to succeed at it.

“Predicting the winner? Is that success? Even if journalists could do that (and they can’t) it would not be much of a public service, would it? A very weird thing about horse race or ‘game’ coverage is it doesn’t answer to any identifiable need of the voter. Should I vote for the candidate with the best strategy for capturing my vote?”

What Rosen calls the “citizens agenda,” which he says was once tried to good effect by the Charlotte Observer partnering with the Poynter Institute, asks different questions: “What do you want the candidates to be discussing as they compete for votes?”

From the best answers to that, everything else flows.


Rosen knows this all sounds dorky and not much fun. But given we’re in an extreme situation these days – with a president dismissive of democratic norms and commanding the news cycle with a wave of his hand – such ideas take on new urgency.

Meanwhile, the clock is already ticking toward November of next year. Ticking toward what may be the most consequential election in American history, other than the last one – before which news organizations for the most part failed to serve the public interest.

Obsessed with polls, they still called the election wrong. They helped crown the Democratic Party’s favored nominee, Hillary Clinton, literally years before the party convention. They allowed Donald Trump to dominate coverage as he wished. They focused on personality flaws and minor mistakes while failing to effectively drive home obvious character disasters.

The prospect of repetition is grim but, from the early signs, all too likely.


Only subscribers are eligible to post comments. Please subscribe or login first for digital access. Here’s why.

Use the form below to reset your password. When you've submitted your account email, we will send an email with a reset code.