New strategies for combating misinformation

A layperson-friendly view. Cross-posted from my personal blog, First Principles.

Fake news is on the rise. We know this from Facebook shares, WhatsApp forwards, Twitter trolls, and Potemkin news sites. We see it in elections across the world, novel coronavirus guidance, and nation-state posturing.

We’ve known about the issue for a while, and technology companies — in their role as the primary distributors — have taken action. This action has not stemmed the tide, and meanwhile the techniques of misinformation evolve and proliferate: bot armies and Deep Fakes being only a few recent innovations.

Why is it so difficult to define what fake news is? Why does calling out lies have little impact on those already deceived? And crucially, what can we do to restore trust and reason to public and social media?

What is fake news?

Fake news is deliberate, targeted, misinformation. It’s not necessarily wholly false: perpetrators are as willing to utilize truths that fit their narrative as they are to concoct falsehoods to construct it. They’re necessarily indifferent to the truth, and attached solely to the outcome: the beliefs they wish to plant within the minds of their targets. From Harry Frankfurt’s On Bullshit:

[The bullshitter’s] eye is not on the facts at all, as the eyes of the honest man and of the liar are, except insofar as they may be pertinent to his interest in getting away with what he says. He does not care whether the things he says describe reality correctly. He just picks them out, or makes them up, to suit his purpose.

Fake news, therefore, isn’t lying, but bullshit.

Why is fake news so hard to fight?

Fake news, being an extension of bullshit, inherits much of its traits:

It’s hard to refute

The claims within bullshit are many, nebulous, and often not even wrong. Fact-checking alone isn’t adequate to this task, and neither is reliance upon a trustworthy set of sources.

It’s normalized

As Frankfurt points out:

One of the most salient features of our culture is that there is so much bullshit. Everyone knows this. […] The realms of advertising and of public relations, and the nowadays closely related realm of politics, are replete with instances of bullshit so unmitigated that they can serve among the most indisputable and classic paradigms of the concept.

It’s hard to regulate

If bullshit is hard to characterize, it’s harder to legally define. By virtue of either not even being wrong or outlandishly so, fake news can take advantage of freedom of speech protections for parody and satire. In any specific case, the perpetrators may be elusive, not within the same legal jurisdiction as the victims, or, if every content repost is counted, too many in number to sue.

What will it take?

Effectively countering misinformation requires a sea change in how journalistic media engages with fake news’ misleading narratives, and in the metrics by which content distributors value and incentivize activity on their platforms.

Preempt the narrative

Fact-checking is journalists’ prime weapon against fake news, and fact-checking tools have rightfully proliferated and are even surfaced alongside suspect material by content distributors; but fact-checking alone is ineffective at changing minds, and is at best a reactive and arduous activity that can only verify a tiny fraction of publicized claims.

Instead, content creators and journalistic media must track fake news with the aim of anticipating the intended post-truth narratives, and promote countervailing, fact-based narratives instead. This narrative-busting connects with disparate audiences in ways most meaningful to each, but without forgoing journalistic neutrality. From UNESCO’s journalism handbook on fake news:

The core components of professional journalistic practice […] can be fulfilled in a range of journalistic styles and stories, each embodying different narratives that in turn are based on different values and varying perspectives of fairness, contextuality, relevant facts, etc.

Journalists must intimately understand their audience to honestly and confidently convey these narratives. Techniques of causal correction and moral reframing have been shown to be effective in conveying factual information, e.g.

Saying "the senator denies he is resigning because of a bribery investigation" is not that effective, even with good evidence that that's the truth.
More effective would look something like this: "the senator denies he is resigning because of a bribery investigation. Instead, he said he is becoming the president of a university."

Journalistic neutrality has come to mean catering to a single — mostly moderately liberal — audience, but at the expense of the touchpoints of understanding that appealed to large swathes of the population. To counter misleading and polarizing narratives, alternatives grounded in reality must be translated to the value and belief systems of diverse peoples.

Incentivize deliberation

Sharing is easy and uniform across all content, but not all engagement is created equal. Technology companies must recognize that slowing down some kinds of engagement leads to higher-quality content and better shareholder value.

Like journalistic media, content distributors have relied on fact-checking, with Facebook, YouTube, and Twitter tagging suspected misinformation. This has been applied sparingly and with mixed results, and also exacerbated the problem by implying that untagged content is verified to be true. And even on this flagged subset of content, sharing and cross-posting remains frictionless.

Blocking the sharing of any content outright is undesirable, and raises issues of censorship and free expression. However, technology companies can build in features incentivizing users to reflect on problematic content prior to sharing and improving the quality of ensuing discussions.

When a user shares flagged content, platform features can potentially enforce adding accompanying comments of a minimum length and complexity to encourage deliberation, or answering a quick IMVAIN survey to crowdsource its reliability. These need not be mandatory, but disincentives can be applied by indicating to subsequent viewers instances where the user declined to comment on or verify the post.

Such measures can be differentially applied, and distributors have already demonstrated this ability in automatically flagging and prioritizing content for fact-checking. By treating content that is new, unverified, or suspected of being misleading uniformly across the spectra of politics and values, platforms can process larger tracts of content, improve content quality, and sidestep bias.

Summary

Fake news is bullshit: hard to pin down, refute, and regulate. Countering misinformation requires journalists to promote factual narratives by engaging overlooked audiences with causal and moral reframing, and content distributors to incentivize deliberation and crowdsource reliability, discourage uncritical reposting of suspect content, and develop engagement metrics that differentiate for quality activity.