Opinion: Scoring Google, Facebook’s fake news efforts and a permanent solution

Photo credit: PDPics.

I made a prediction over a week ago about how Google and Facebook could respond to the ongoing controversy over fake news and outlined steps that they should take. I advocated a three-point strategy: whitelist authoritative news sources (step 1), flag potential fake news and limit their reach (step 2), and use journalists as moderators and give them power to overrule step 2. Let’s check out the scorecard and my solution to end the controversy permanently.

What Google, Facebook have done

Google has done the Googly thing by partnering up with third parties like Politifact and Snopes. Their new feature sees a truthfulness rating given to each Politifact or Snopes search result on hot button news topics. Ratings appear beneath the meta description and range from true to false.

My suggested solutions focused on Google News, so I’m surprised that Google has moved to label questionable news items with the conclusions of fact-checkers. Google decided that this is an enormous problem. Unfortunately, they only took a timid half step forward. By outsourcing the problem of fighting fake news to Politifact and Snopes, Google has replaced one problem with another. They are now associated with outcomes of the fact-checkers’ work while having no control over the conclusions. It is ultimately a weak response.

Facebook, on the other hand, is acting precisely as predicted. So far, Facebook has reassured the public that they will enforce existing ad policies more aggressively, preventing fake news publishers from advertising their tall tales. Stories flagged as fake news by users cannot be promoted via ads either. Flagged stories may also be sent to partner news organizations to be investigated, and if the story is judged as fake, its reach will again be dampened. These steps are not 100 percent sufficient, but they are the right first steps.

Facebook also launched the Facebook Journalism Project and News Integrity Initiative, both of which aim to combat fake news by conducting research into new products, perform public service announcements, and educate the public, increasing public trust. As anyone who has seen overzealous signs regulating public behavior will tell you, this sort of response is often necessary, but does not work well. What is needed is a way to nudge users to the right direction.

The problem

The main problem with how Google and Facebook are approaching fake news is that they perceive it as a rational problem that needs a technical solution. This is a misunderstanding of the nature of fake news.

People seek news that validate their opinions. Studies show that when presented with evidence disproving their belief, people actually harden their beliefs. Against that context, Google’s inclusion of fact-checker ratings only hardens the beliefs of people on both sides of the issue. Ultimately, nothing is solved, and everybody on the opposite side of Politifact and Snopes will still be angry.

Users are prone to be duped by the fake news they agree with and judge news they disagree with as fake.

I think that Facebook did a better job than Google. It still was a surprise why these measures weren’t put in place a long time ago. It seems natural that Facebook would want to moderate the spread of news that elicits an overwhelmingly negative reaction. However, when it comes to the currency of engagement, Facebook opts for the money. Let me share again what former Facebook product manager Bobby Goodlatte said: “Sadly, News Feed optimizes for engagement. As we’ve learned […], bullshit is highly engaging.”

I disagree with Facebook that there is a technological solution to identify fake news. The signals that Facebook uses for fake news, such as a low sharing rate and a high rate of negative comments look the same for fake news and real news that people dislike.

Relying on people to discern the difference between real and fake news so that you can build an algorithm to identify them is putting the cart before the horse. The incentives and signals are all wrong. Users are prone to be duped by the fake news they agree with and judge news they disagree with as fake. We are asking algorithms to learn to tell black from white by observing how angry we get at colors.

The solution

I think the best solution is to make use of behavioral insights. A Yale study found that a high level of scientific curiosity is correlated with a high level of flexibility in beliefs, even for controversial topics.

An unfair consequence of the fake news controversy is that people expect Google and Facebook to “correct the record” for everyone who disagrees with them. A clever solution would place the initiative back with each user and encourage them to discover a perspective on issues that they had not considered before. With this method, the truth of a claim is less important than encouraging users to consider different viewpoints.