Blog post

Why Facebook’s Labelling of Fake News might Make Matters Worse

By Magnus Revang | December 19, 2016 | 1 Comment

After receiving massive criticism during and after the US Presidential election Facebook has taken steps to fight the publishing and spreading of “Fake News”. First impression is that this is a Good Thing. But, before you rejoice, the changes might have the exact opposite effect – making people even less informed and even less able to distinguish fake from real. It might in-fact make matters worse.

So, firstly, what is Facebook doing? A change that appeared a couple of days ago is that some news stories is labelled “disputed by 3rd party fact-checkers”. Facebook is working with Poynter’s International Fact-Checking Network (IFCN). This is coupled with a change where fake news publishers don’t get a share of ad revenue from spread of their content.

To play devil’s advocate. If you were a fake news profiteer, what would you do to keep your business running? I see several likely developments:

  • Diversify and Flood. Human fact checking will be no match for legions of low paid workers rewriting the language of the story and posting on an exponentially larger volume of websites. This mirrors perfectly how one beat search engines in the good old days.
  • Bait and Switch. The web is dynamic, and webpages might change at any time. This can be used to bait and switch. One is publishing truthful articles, then switching them after a fact-check is done, the other is far more sinister: publishing fake news and when fact-checked change it and claim the fact-checked to be wrong – even threatening with court action.
  • Subterfuge. Nothing like a challenge to up the game. English is a highly ambiguous language and masters of it can say one thing, factually being correct, but still evoke misinterpretation. Fake news might make way for Fake-ish News, and focus might be turned to places where facts are more easily disputed.
  • Fact-check the fact-checkers. I predict we’ll see articles attacking the fact-checkers themselves, possibly mistakes they’ve done (I’m sure they are not perfect). If factually correct, the fact-checkers have no choice but to let them through which will erode trust in them.

But, these likely developments is not the only challenge. The main challenges is speed, scalability, economics and psychology.

It’s an undisputed fact that fact-checking takes time (see what I did there). The time from publishing to an article getting fact-checked could take days, if it’s ever done. In this time, an article will not be disputed.

Fact-checking costs money, and requires competent people to do. It’s not scalable (although Facebook may have some AI tech up their sleeve for the future).

Which brings us to the psychology. When articles gets tagged with “disputed”, what happens to the articles that are not tagged as “disputed”? We will treat them as more credible, that’s what. Because of speed, scalability and economics – there will always be a sizable percentage of “fake news” that is never fact-checked, and due to other “fake-news” being labelled we will trust those articles a lot more. I predict a lot more comments in the nature of “sure this is right?” with the answer “it’s not labelled as untrue by Facebook, so it must be, right.”

As long as publishing fake news is cheaper and more profitable than fact-checking them, the fact-checkers will loose – the only thing we are achieving is that we are making the “fake-news” publishers more sophisticated and the public more gullible for non-labelled news.

Disclaimer: This is not an endorsement of “Fake News”, it’s simply a likely scenario of consequences arising from the change that Facebook is making.

The Gartner Blog Network provides an opportunity for Gartner analysts to test ideas and move research forward. Because the content posted by Gartner analysts on this site does not undergo our standard editorial review, all comments or opinions expressed hereunder are those of the individual contributors and do not represent the views of Gartner, Inc. or its management.

Leave a Comment

1 Comment

  • Doug Laney says:

    Great piece Magnus. The early days of spam checking deployed a combination of crowdsourcing and deterministic approaches. It worked pretty well. But I can see the additional challenges you outline being more difficult.

    Cheers, Doug
    VP and Distinguished Analyst, Data & Analytics Strategies, Gartner