A pundit is the person on the news who is not the anchor, not the reporter and not the politician, but who is nonetheless in the chair explaining what it all means. Sometimes they are called an "analyst," sometimes a "commentator," sometimes a "strategist," sometimes a "former official." The one thing they are almost never called on air is what they mostly are: a guest whose job is to have an opinion on demand, at 8:47pm, in ninety seconds, about something that happened at 8:31pm. * The word is older and grander than the job. It comes from the Sanskrit pandita, a learned man, via the Hindi pandit — and it entered English in the 1600s to describe court scholars in colonial India who advised British judges on Hindu law. Somewhere between then and cable news, "pundit" travelled from "someone who has studied the thing for a lifetime" to "someone who is available and has a good lighting setup." * How good are they? The best answer anyone has produced comes from the psychologist Philip Tetlock, whose 2005 book Expert Political Judgment reported on a study that ran for roughly two decades. Tetlock recruited 284 people whose professional lives involved commenting or advising on political and economic trends — academics, journalists, government and think-tank experts — and collected about 28,000 forecasts from them about real events. Then he waited to see what happened. * The headline result, in Tetlock's own famous phrasing, was that the average expert did about as well as a dart-throwing chimpanzee. Expertise did not make people much better than chance at predicting political outcomes. What did matter was style of thinking. Tetlock split his experts into "hedgehogs" — people with one big idea that explains everything, delivered with confidence — and "foxes," who hold several ideas at once, hedge, and update. The foxes forecast better. The hedgehogs forecast worse. And the hedgehogs, being confident and quotable, were the ones who got booked on television. * That last point is the whole trick. Broadcast punditry does not select for accuracy, because accuracy is never checked. It selects for availability, fluency, certainty and the ability to fill a segment without dead air. A guest who says "honestly, it could go either way, and here are three reasons I am not sure" is a good forecaster and a terrible booking. A guest who says "this is the beginning of the end for the prime minister" is a bad forecaster and a great one. The incentives are exactly backwards, and nobody keeps score. * It gets worse when the pundit has a client. On April 20, 2008, the New York Times published David Barstow's investigation "Behind TV Analysts, Pentagon's Hidden Hand," based on some 8,000 pages of Pentagon emails, transcripts and records the paper won through a two-year Freedom of Information Act lawsuit. It showed that, from 2002 onwards, the Pentagon's public affairs office under Victoria Clarke had recruited more than 75 retired military officers who appeared on Fox, CNN, NBC, ABC, CBS and MSNBC as independent "military analysts." * Internal Pentagon documents called them "message force multipliers" and "surrogates" — people who could be counted on to deliver the administration's themes on Iraq to millions of viewers "in the form of their own opinions." They were given private briefings, talking points and Pentagon-paid trips to Iraq and Guantánamo. Many also held positions with defence contractors that benefited from the policies they were defending on air, which viewers were not told. The Pentagon suspended the programme within days of the story. Barstow won the 2009 Pulitzer Prize for investigative reporting. * The epilogue tells you how seriously the system took it. In January 2009 the Pentagon's inspector general issued a report clearing the programme, then withdrew it after an internal review found it "did not meet accepted quality standards." A second inspector general report, released in 2011, again concluded the programme had complied with Defense Department rules. The networks, for their part, mostly declined to comment. Nobody on your screen was ever labelled "retired general, also on the board of a company that sells the thing he is recommending." * Why does it matter for you rather than for media ethics seminars? Because the pundit is the layer of the news that tells you how to feel about the facts, and it is the least accountable layer. A reporter's factual claim can be corrected. A pundit's prediction simply evaporates. A reporter's conflict of interest is a firing offence. A pundit's is, at most, a chyron nobody reads. When the strategist says "voters won't wear this," they are not reporting on voters. They are, quite often, one of the people trying to make sure voters won't. * There is a class dimension too. Look at who is in the chair. The people asked to explain what a rent cap will do are rarely tenants. The people asked what a strike means are rarely the ones on strike. The pundit pool is drawn from the same postcode as the politicians and the producers, which is why so much of it sounds like a dinner party analysing the staff. * How to watch a pundit in 30 seconds: (1) What is their job title when they are not on television? "Strategist," "consultant" and "adviser" mean they have clients — ask who. (2) Are they predicting or explaining? Predictions are free and nobody will check. (3) Do they ever say "I don't know"? Tetlock's data says the ones who do are the ones worth listening to. (4) Would the segment collapse without them? If a reporter with the documents could have said the same thing, the pundit was there to fill time, not to inform you.