Facebook’s political ad monitoring is already a problem for the company.
Facebook (FB) wants you to know that it’s not going to fall for that Russian-fake-news trick again. As it gears up for elections in the U.S. and elsewhere, the social networks has been deploying multiple defenses against thedisinformation campaigns that overran the company throughout 2016. You’ll recall CEO Mark Zuckerberginitially brushed off those same issues as a small, inconsequential issue.
“In retrospect, we were too slow,” chief product officer Chris Cox admitted a conference Facebook hosted in Washington Tuesday. He and other company executives vowed that Facebook would do better, thanks to countermeasures already being deployed.
Many of these steps—in particular, a crackdown onfake accounts—look like good moves. But Facebook’s attempt to force transparency on politically-oriented advertising is already tripping up non-political advertisers, from apartment listings to news stories.
Not even Facebook seems happy about the results so far. But the company can’t blame anybody but itself for having let things get to this point.
In September, Zuckerberg said that the company would require any buyer of ads for or against specific candidates to disclose their identity, and then keep all those ads visible to anyone that visits their pages. A month later, Facebook said it would also require election ad buyers toconfirm their identity offline.
And in April, the company said it would extend these verification and transparency rules to buyers of adsaddressing political issues, not just candidates. It would also maintain an online archive of all political ads to inventory their targeting and performance.
Running ads thatFacebook deems political entails averification process that starts with submitting a scan of your driver’s license or U.S. passport, as well as providing your mailing address and the last four digits of your Social Security number.
You then have to wait for Facebook to mail you a letter—yes, on paper—with a confirmation code that you enter at Facebook’s site to confirm your U.S. location. Without all that, Facebook will take down the theoretically-offending ad.
These rules went into effectMay 24, and a look at their results as shown in Facebook’s political-ads archive—facebook.com/politicalcontentads—already shows problematic results.
A query for“apartments,” for example, yielded dozens of obviously political ads but also a Brooklyn apartment building’s shout-out to Pride Month, a New York TV station’s report about the city housing authority and listings for a complex in Indianapolis.
An example of a Facebook ad that the company erroneously labeled as political.
A search for“airline” showed that Facebook had taken down an ad for a business-news story about the Asian commercial-aviation market and a promoted post by a former flight attendant alleging sexual assault by a coworker.
“The way Facebook’s classifying issue ads is quite broad,” said Miranda Bogen, an analyst with the policy groupUpturn. She cited checks by that Washington-based non-profit (in May, itpublished a study critiquing Facebook’s initial ad-transparency steps) that found “purely factual information” tagged as political.
“That’s obviously a problem for advocates who are trying to provide resources to different groups,” she said, citing immigration issues in particular. “The same with news organizations; if they’re even writing about a topic that is among the issues Facebook has listed as political, it’s listed as a political ad.”
It appears that ads can escape this purgatory—in second searches of the ad archive for the same term, some false-positive results no longer appeared—but Facebook doesn’t seem to document an appeals process.
The company has, however, pledged to treat ads for news stories differently… somehow. Wednesday, ads manager Rob Leathern tweeted that it had to flag them “to prevent workarounds for bad actors” but would soon “launch a separate section in our archive for news ads about politics.”
It’s not like Washington is helping
The idea that an automated screening process could go disturbingly wrong shouldn’t be a stretch. See, for example, when YouTube decided a video of a giant American flag at a football gamewasn’t eligible for advertising income.
But while it’s tempting to blame Facebook for once again placing too much faith in its algorithms, you can’t accuse the company of doing so out of self-interest. By making its own ads an unreliable medium, it’s hurting its own bottom line and encouraging businesses to find more stable platforms.
And meanwhile, it’s not like anybody else is doing much. Last year’sHonest Ads Act, which would require most online ads addressing specific candidates to meet the same disclosure standards as on radio and TV, has yet to advance past committee in either theHouse or theSenate.
Facebook may be making a mess of this. But Congress isn’t trying to make a change at all.
