As the sun rises on yet another revelation about Facebook being duped by a calculated misinformation operation, its much-touted independent fact-checkers are claiming they’re not being given the data to do their job.
It was during the endless run-up to the 2016 Presidential election that Facebook unwittingly became the platform-of-choice for publishing outrageously false political news. From Pope Francis endorsing the Trump candidacy to Clinton selling weapons to ISIS, the din of nonsense was constant. It was effective too, with Facebook’s user-friendly “share” function allowing stories to spread like wildfire. Those two examples alone generated millions of Facebook engagements.
According to reports this morning, a six figure Facebook ad spend during the election can be attributed to a “shadowy” Russian organization, the Internet Research Agency, which apparently created hundreds of fake Facebook accounts to purchase ads on polarizing issues like gun control and immigration.
Of course, the purchase of ads on Facebook (and Audience Network?) by Kremlin-linked trolls is a separate issue from real or fake accounts linking to false news reports which are then widely shared. But it does prompt questions about how Facebook can be such a marketing and data power-house while professing itself incapable of policing the content it publishes. (Yes, as I’ve argued before, Facebook is not just a software company: it’s a publisher.)
Just how incapable is it? Back in April, it partnered with independent, third-party fact checkers like Politifact and Snopes, paying them to monitor the platform for dishonest content. That partnership is already showing signs of strain. Today, Politico reported that some fact-checkers are “frustrated” that Facebook refuses to share data which would indicate whether flagging fake news stories retards their distribution, or provide grounds for prioritizing the kinds of updates (out of millions) most worth checking.
The response from Facebook is that sharing the data would create “privacy concerns.”
Let’s pause and consider the services Facebook offers to marketers and advertisers — which after all, politics aside, is why brands should care about the fake (and/or offensive) news problem. Two billion people use Facebook every month. Facebook gathers data which allows brands to target anonymized segments of this vast audience with relevant, timely, engaging messages. Audience Network uses “the same targeting available for Facebook ads today, including Custom Audiences, core audiences and lookalike audiences” to vastly extend the reach of Facebook campaigns to as many as one billion people per month.
Does anyone think Facebook bases these services on guesswork and spreadsheets? Does anyone think that’s the impression Facebook is trying to give?
Of course not. Facebook has the ability to analyze and activate data on millions of individual consumers, and is clearly capable of making sub-sets of this data available in anonymized form, when it wishes to do so.
Every Facebook user who stops to think about it is aware of the extent to which Facebook has automated the user experience, editing millions of news feeds in real-time, prioritizing content to be shown, including images and video. And it reportedly has a team of 150 people work on innovations in AI.
Despite all this, we’re asked to believe, Facebook must lean on third-party fact-checkers to monitor content, but can’t provide the data needed to perform the task effectively. Although Facebook can tailor relevant experiences to millions of users in real-time, it has been hesitant, to say the least, in bringing machine learning to bear on the fake news problem, or the newly revealed problem of fake accounts pushing divisive ads.
Facebook, again, is a publisher, whether Mark Zuckerberg likes to admit it or not. It’s unique among publishers in offering extraordinary, state-of-the-art audience engagement and adtech tools, but turning a willful blind eye to the quality, honesty, and decency of what it publishes.
