Horse race journalism is election coverage that treats a campaign the way a racecourse commentator treats the 3:15 at Cheltenham: who is ahead, who is fading, who has "momentum," what the stewards (donors, strategists, pollsters) are whispering in the paddock. The candidate's actual plan for your rent, your wages, your hospital — that is the form guide nobody reads. * It is the single most-studied habit in political reporting, and the numbers are not subtle. When Harvard's Shorenstein Center on Media, Politics and Public Policy measured coverage of the 2016 US general election across ten outlets — ABC, CBS, CNN, Fox, NBC, the New York Times, the Washington Post, the Wall Street Journal, the Los Angeles Times and USA Today — from August 8 to November 7, 2016, the horse race took 40 per cent of all coverage. Policy took 10 per cent. Thomas Patterson, the political scientist who ran the study, put it plainly: policy got less than a quarter of the space handed to the horse race. * The primaries were worse. Shorenstein's companion study of the January 1 to June 7, 2016 primary season found that "winning and losing" — polls, delegate counts, fundraising totals, tactical manoeuvring — accounted for more than half of all reporting, with roughly another third spent on the mechanics of the election process. Substantive coverage of candidates' policies, character and record came to about 10 per cent. In other words, for every ten stories about who might become president, one was about what they would do once they got there. * Where did the habit come from? Patterson dates it: the horse race has been the dominant theme of election news since the 1970s, when news organisations started commissioning their own polls. That detail matters. A poll is a story an outlet owns. It has a fresh number every week, it costs money, it carries the outlet's brand, and it is a lot easier to write up than a 40-page housing plan. Once newsrooms became poll producers, the race became the product. * There is also a comfort in it. A horse race story is impossible to get "wrong" in the way a policy story can be. If you report that a candidate's childcare plan would cost £9 billion and an economist says it is £14 billion, you have a correction to run. If you report that a candidate "appears to have the momentum," nobody can ever prove otherwise — momentum is a feeling with a byline. That is why the genre is so beloved by reporters who would rather be wrong about nothing than right about something. * The problem is what it does to voters. A 2020 study in the Journal of Politics by Sean Westwood, Solomon Messing and Yphtach Lelkes — titled, with admirable directness, "Projecting Confidence: How the Probabilistic Horse Race Confuses and Demobilizes the Public" — looked at the newer form of the genre: the "87 per cent chance of winning" forecasts popularised in the 2010s. They found that people badly misread those probabilities as vote margins, becoming far more certain about the result than the underlying polls justified, and that this certainty was associated with lower turnout among supporters of the candidate who was "supposed" to win. Their 2016 example is the obvious one: Hillary Clinton supporters who were confident she would win were less likely to turn up. * So the format is not neutral. Telling people a race is over is a way of ending it. Telling people it is close is a way of making it close. Either way, the coverage is nudging the thing it claims to be observing — while the questions that decide whether the winner is any good at governing sit in the 10 per cent. * There is a class angle here that the trade prefers not to notice. Horse race coverage is written for people who already know what the candidates stand for: political staff, lobbyists, donors, other journalists. For them, "who is up" is genuinely the news. For a care worker deciding whether to vote at all, "who is up" is useless, and "what happens to my shifts under each of these people" is the entire point. The genre serves the paddock, not the stands. * Watch, too, for the horse race disguised as analysis. "Can the candidate turn it around?", "Is the campaign in trouble?", "What does the polling tell us about the mood?" — these read like scrutiny and contain none. A story about whether a manifesto pledge is affordable is scrutiny. A story about whether the pledge "landed" with voters is a horse race story wearing a tie. * None of this means polls are useless or that the state of a race is not news. It is. The Shorenstein finding is about proportion, not existence: four stories about the race for every one about the stakes. A newsroom that flipped that ratio would still tell you who was winning. It would also, for a change, tell you what you were voting for. * How to read election coverage in 30 seconds: (1) Does the story tell you what the candidate would actually do? If not, it is horse race. (2) Is the "expert" quoted a strategist or pollster rather than an economist, doctor or teacher? Horse race. (3) Does it use "momentum," "narrative," "lands," "optics" or "the mood"? Horse race. (4) Would the story still make sense if you swapped the candidates' names? Then it was never about them — it was about the race.