1 November 2017

What can we trust, in an age of fake news?


I stopped driving recently. Because we live in the suburbs, where choosing not to drive is seen as bizarre, I’ve developed plausible explanations for the decision: that I prefer to read on the train, for example. That’s true, but not the whole truth.

The reason I’ve stopped driving is that I no longer trust other drivers. To drive is to trust in the judgement — the rationality and benign intent — of strangers. I’ve seen too much counter-evidence about their judgement on my daily commute to maintain that trust.

It’s a cold conclusion to come to. But I’m not alone in my untrustworthiness. You don’t have to be (philosopher) Baroness Onora O’Neill to have noticed that general levels of trust, in each other and in our institutions, have fallen in our lifetime. In particular, trust in the media is at an all-time low (this Ipsos MORI survey puts journalists on a par with estate agents and, good God, politicians); you can’t (ironically) open a newspaper without reading a story about “fake news”.

Just like me and the car, when you don’t trust the media, you stop engaging with it. Young people, we’re told, no longer consume news from the MSM but instead take all their information from Facebook.

Despite being a news refusenik myself, I think this is a retrograde step. If we stop trusting edited content — “MSM” — then we’ll lose something valuable, regardless of the editorial slant of the publishing organisation.

Believe this jobbing hack: my pieces that have been through an editorial process – I don’t mean censorship, I mean a proper edit to make it more readable – are always more worth your while to read, than raw “blog” output.

If we lose our trust in edited content – through whatever proprietorial prism such editorialising occurs – then, in time, we’ll be left with nothing but non-overlapping stream-of-consciousness echo chambers. “Fake news” may simply be news that doesn’t agree with one’s a priori opinion.

There are, however, a few changes that could make MSM outputs more palatable. My suggestion, at a packed debate at the Institute of Ideas’ Battle of Ideas last Sunday (“Do you trust the media?”) concerned evidence. Specifically, in the MSM’s handling of lobbyists who begin their interviews with “The evidence says…”

Memorable cases are to do with global warming, where lobbyists for the pro-AGW side (ie that it’s man-made) start nearly every interview with “The science says…”, as though evidence — data — had proven their AGW theory, implying that anyone who disagreed with the lobbyists’ position was anti-science; irrational; a fool.

Not only those concerned with AGW do this: listen carefully to interviews with lobbyists about obesity, or alcohol, or mental health — almost every issue where a lobby group is seeking to change your behaviour with the lever of government policy — and you will most likely hear sentences which begin: “The science is in …” or “The evidence says…”.

The irony is, I’m often on the lobbyist’s side of the argument; certainly I’d vote for anyone with a serious intent of reducing car ownership, whether or not that helped slow down global warming. But my point — and how it relates to mis-trust — is that assertions that “data”/ “evidence”/ “science” has proven some theory incontestably true are nearly always false. Data alone can rarely tell you what to believe about a theory (beyond the easily refutable sub-species, such as “All swans are white”: “Oh look! a black swan!”), whether that theory be geophysical or political or related to public health.

As the eminent British statistician Dennis Lindley put it, in 1993: “All evidence does is to change opinions: it does not create them.” An army of fact-checkers can’t tell you what’s true.

A failure to appreciate this (fairly simple; but I’d lie if I pretended it were uncontroversial) epistemic truth ties much of our political news in knots. Remember the pharmacology expert and policy adviser David Nutt, who lost his job because he was deemed to be at odds with government policy, despite his claims about “the science”?

Here’s a thought experiment we use at work, when we’re introducing R&D staff to this form of critical thinking.

Imagine I’ve a coin in my pocket and my theory is that the coin is fair: it’s equally likely to fall heads as it is tails. I tell you I’ve tossed the coin five times, and it came up heads on four occasions. Is the coin fair?

Now imagine I take the coin from my pocket, and it’s not the bright and shiny freshly minted pound coin that you’d assumed; it’s some ancient Roman thing I dug up from our garden, more than three thousand years old with only faint markings on the tails side, but an embossed representation of a very ugly emperor on the other, with a huge Roman nose, stuck on with putty.

So I ask you again: do you think, on the basis of four heads in five tosses, that my coin is fair?

At this point people say “You’re cheating: you didn’t tell us all the background information you had about the coin.” But that’s my point: had more prior information about the coin than you did, so of course we might arrive at different conclusions on the basis of the same evidence.

The point is “something other” than evidence (four heads in five tosses) affects your opinion about the fairness or otherwise of the coin in question. That “something other” is your prior belief; or the position from which you currently view the validity of the policy or theory in question.

So when the lobbyist says “the evidence from our new survey is in favour of our policy; the government must fund our objectives”, I’d love the journalist to ask:

  • Did you believe your theory was true before the survey was held? (Of course they did. They already had huge prior faith in their theory. A little more data merely makes them more confident.)
  • What would your evidence do to the belief of someone who was honestly sceptical about your policy objective?
  • How many other theories are as strongly supported by the evidence, as the one you’re promoting today? (Most lobbyist data is vague; it would support almost any reasonable political theory, not merely the lobbyist’s favourite.)

To ask questions such as these isn’t to pass moral judgement. I’d argue that to ask them is the sine qua non of any media organisation which seeks the trust of its consumers (and, in the case of the BBC, its funders.) And if journalists don’t ask these questions, what is their function? To magnify the press releases of lobbyists?

In the meantime, if you know anyone who’d be interested in a Mini Cooper S, let me know. The car’s in a very good condition and if you bought it, the evidence is that you’d save money in the long run. You can trust me on that.

Graeme Archer is a statistician and writer