We repeated our 2016 test to figure out whether Facebook was adequately policing itself. It wasn’t.
by Ariana Tobin
Last week, we bought more than a dozen housing ads excluding categories of people explicitly protected by the Fair Housing Act of 1968.
Were these actual ads? No. And as someone who’s spent the past month on a New York City apartment hunt, I’m pretty confident that no one would mistake our “real estate company” for an actual brokerage.
But here’s the question: could they have been real? Yes — and our ability to limit the audience by race, religion, and gender — among other legally protected attributes — points to the same problem my colleagues Terry Parris Jr. and Julia Angwin reported out a year ago, exciting much outrage from people who care about fixing discriminatory housing practices.
Here’s what they tested last year:
Their work put Facebook on the defensive. It put out a statement that “discriminatory advertising has no place on Facebook.” In February, the company announced that it had launched a system to catch problematic housing ads and “strengthen enforcement while increasing opportunity on Facebook.”
Meanwhile, Facebook has amped up its efforts on real estate this year. You can see housing ads all over Marketplace, a slick Craigslist-like section of the site. Facebook recently partnered with real brokers at Zumper and Apartment List, and it has announced it will be rolling out new features over the coming months. With that in mind — and fresh off our other Facebook ad portal investigation into “Jew haters” — we wanted to know whether it had actually fixed the problem.
So we more or less repeated the exercise:
Our ads skated right through the approval process. Again. Approved in under two minutes. It wasn’t just this ad. We also managed to buy ads excluding users based on religion, family status, national origin, sex, race, ability, and more — every group that’s supposed to be protected under major housing laws.
Facebook apologized profusely for what it called “a technical failure,” and promised — again — to strengthen its policies and hire more reviewers.
I want to answer a question that has been posed by quite a few people on social media in response: Why would ProPublica take out discriminatory ads?
Sure, Facebook, who's surprised? But what the heck is up with ProPublica?!
— Mary Christopher (@MaryChr01644671) November 21, 2017
It’s a fair question. Let’s be clear: we don’t take the decision to buy a fake ad for non-existent housing lightly, and it is against our newsroom’s policy to impersonate for the sake of newsgathering. But there was no other way for us to test the fairness of Facebook’s ads. And our disturbing findings suggest that, even if Facebook has the best intentions and just overlooked a technical glitch, somebody on the outside has to watch its advertising platform more closely than it seems to do itself.
We had a lot of conversations internally about the best and most ethical ways to do this. Here’s how our editor-in-chief Stephen Engelberg puts it:
“Social networking platforms like Facebook are enormously powerful in shaping the modern world. In this instance, we viewed the public interest in testing its promises about discriminatory advertising as sufficiently important that it outweighed the possibly detrimental effect. Remember, we canceled these ads as soon as they were accepted so that the likelihood they would be seen by anyone was remote. In our judgment, placing this ad was far less deceptive than a reporter posing as someone else, a practice we continue to bar under virtually all imaginable circumstances.”
Meanwhile… if you see any eyebrow-raising ads or advertising categories on your Facebook feed this shopping season, let us know. We’re gathering political ads, and we want to stay on top of ads for housing, employment and credit as well. Clearly, this is a story that requires persistence.
Originally appeared on ProPublica.org