What fact checkers get wrong They pretend to be objective — while peddling progressive politics

Fifteen years ago, in a kinder gentler age of the libertarian internet, a few amateur sleuths over at Snopes would bust urban myths about disappearing hitchhikers or In The Air Tonight. They didn’t “check facts” so much as deliver zingy reports on the genesis of rumours. But more and more, the fact check has become the epistemic peg around which everything else is expected to pivot.

Social media giants, as the Facebook leaks revealed, are very aware that they’ve built a world that thrives on cheap talk and rumour. They know they can’t put that genie back in the bottle without dismantling their entire business model. So in a world that requires politicians to puff their cheeks and fume about how something must be done, the giants have taken to paying for fact checks. These are the tithes they pay, the indulgences they gain. Chucking some change at fact-checking is very cost effective way of buying up the most facile tier of public opinion, and keeping the regulators from the door that bit longer.

Take the British fact-checking company, Full Fact. In 2019, it was paid £1.1 million by Facebook, and £206,500 by Google, a sum that has allowed them to take on a staff that would be the envy of many of the newsrooms they pass verdicts on. It is only a small irony that, after reducing reporters to desk-bound re-writers by stripping newspapers of their funding model, Big Tech is paying to have some of their reporting fibre reinserted, by recruiting a new breed of professional to back leap in afterwards.

Full Fact says that its funding doesn’t compromise its integrity. But the fact that the boilerplate California progressives over at Facebook and Google chose to fund them in the first place obviously does. How could they not make decisions based on keeping the giants on side? The Intercept journalist and professional gadfly Glenn Greenwald picked up on this principal-agent problem only a few weeks ago, tweeting that:

Two of the hugest scams are "fact-checking" agencies and those that purport to fight "extremism" online. They all push the same ideology, are run by the same small set of people (see below), and are usually funded by the same small handful of billionaires: https://t.co/te8BxuYtXL

— Glenn Greenwald (@ggreenwald) October 13, 2021

Greenwald was picking up on comments from NSA whistleblower Edward Snowden about NewsGuard. As the two dissidents point out, as well as Big Tech there is Big NGO: overly moneyed world changers such as Bill And Melinda Gates Foundation, and George Soros’ Open Society group, give conspicuously. It’s hard to imagine anyone in their NGO echelons arguing for, say, “closed borders” or “foreign aid reductions”. And it is just as hard to imagine the fact-checkers they have funded actively aiming to check a certain kind of fact that is values-aligned.

During the summer of George Floyd, for instance, fact checkers raced to the scene of facts that supported the BLM cause. For instance, they dismantled “a bar graph showing ‘murder of blacks and whites in the US, 2013’ broken down by the races of the perpetrators”, as Reuters Fact Check described one anti-BLM meme. “While the graph itself is accurate, the way it is presented is misleading,” they then concluded. Some might have said the same thing about the way they headlined the piece: “Misleading bar graph presents distorted interpretation of black and white murder rates.”

But even beyond how the tech giants select their checkers, who funds them, or what that funding model incentivises, there will always remain a simple question of irreducibility. At what point does the eternal human mist of subjectivity, data and language ever actually congeal into one perfect cube called a “fact”?

Stephen Ceci and Wendy Williams, both professors at Cornell University, lay out some of this in Scientific American. Confronted with a crowd bearing placards, they noted, experimental evidence suggests that those of a certain political bent will see a mob; those of another will see a rights struggle.

This experimental evidence closely correlated with the real-world evidence last year, during the BLM Summer: while 69% of Republicans considered the videotaped demonstrations in Portland, Kenosha and New York to be riots, only 30% of Democrats did. When Keynes said “When the facts change, I change my mind”, he forgot to add that when the mind changes, the facts go with it.

The litany of comedy fact checks is long. “After about five years, approximately half of [German] refugees are employed” is rated by EU Fact Check as “mostly true,” even though 51% are unemployed — and even of the employed, only 68% are actually in full time work. The reason for the rating is apparently that the timeline of the study doesn’t exactly coincide with the length of the claim. “Does the timeline of events on Jan. 6 prove Trump could not have incited the attack?” CNN’s Fact Check asks this week. “This is misleading,” it happily concludes.

And the full weight of systemic and cognitive bias on display is evident in one immortal zinger that came from Associated Press’s Fact Check Unit a year ago:

“President Donald Trump claimed accomplishments he didn’t earn on the pandemic, energy and veterans at a Republican convention finale that also heard Black Lives Matter baselessly accused of coordinating violent protests across the country.” 

Often, the outcome simply depends on the framing. If asked to “prove” something “false”, the conditions will be very different from if you are asked to “prove” it to be “true.” Many will unconsciously adjust the settings to match the outcome preferred.

A hundred years ago, Ludwig Wittgenstein eviscerated the “logical positivist” philosophical movement, which wanted to create an algebra of human meanings. He showed how there was no single way of coding a statement so that it could be parsed as “true” or “false” that didn’t in turn leave it subject to huge error bars of ambiguity. If only someone had passed the message onto the fact checking industry, it might have saved everyone some time.

If there’s no such thing as objectivity, there’s still impartiality. And fact checkers can’t even manage that. They were meant to bring back the rigour that had been throttled out of journalism. Instead, they’re most comprised of sweaty 23-year-olds in graduate jobs rummaging through the usual partisan sources, then comparing them with the often progressive-partisan fact mills of Wikipedia then spitting out the appropriate Woozle. No wonder many have intuited that a red flag from a fact checker actually denotes “things that might be true but are unsayable in the present climate”.

One solution proposed by the Cornell duo of Williams and Ceci involves leaning into that polarised political climate. Rather than having a single set of fact checkers, you should have two rival teams, recruited precisely for their ideological differences. But even here, the authors are quite gloomy in their prognosis. We shouldn’t imagine that this will get us to the “truth”, they warn. But it will at least reveal the underlying human biases that might be skewing our perception of it.

That might turn out to be a more useful function than they realise. The problem that confronts all of us is to neither lapse into cynicism about the existence of facts, nor slip into hubris about our ability to see them. In order to do that, we need to treat fact-checkers not as arbiters, but as just another branch of the media-political system that it seeks to impose its writ upon. Perhaps sensing that the game is up — or merely sensing that they have to expand to fit the donor funds on offer — fact checking orgs have increasingly begun to write what are essentially articles. In other words, they’re ripe to be red flagged, ripped apart or rinsed by others.

---

Gavin Haynes is a journalist and former editor-at-large at Vice.

@gavhaynes