Move slow and break things

On May 24, 2018, Facebook announced that, moving forward, "all election-related and issue ads on Facebook and Instagram in the US must be clearly labeled – including a 'Paid for by' disclosure from the advertiser at the top of the ad." According to the company, the new policy "will help ensure that you can see who is paying for the ad." 

Five months later, an investigation by VICE revealed that these requirements were a sham. VICE was able to "buy fake ads on behalf of all 100 sitting US senators, including ads 'Paid for by' by Mitch McConnell and Chuck Schumer." The ads "were granted to be shared from pages for fake political groups such as 'Cookies for Political Transparency' and 'Ninja Turtles PAC.'"

VICE's report exposed that, while Facebook required ads to be labeled, it was doing very little to verify that these labels were accurate. It took Facebook another 11 months to take action. 

On Wednesday, Facebook announced that "starting mid-September, advertisers will need to provide more information about their organization before we review and approve their disclaimer." Advertisers who don't provide the additional information will have their advertisements paused in mid-October. 

There are several ways to meet Facebook's new requirements. Advertisers in the private sector can provide either a tax ID number or an FEC identification number. Those advertisers will be labeled as a "Confirmed Organization." 

But advertisers who don't want to provide that information have alternatives. They can "submit an organization name by providing a verifiable phone number, business email, mail-deliverable address and a business website with a domain that matches the email." Advertisers can also "provide no organizational information and rely solely on the Page Admin’s legal name on their personal identification document." 

It's unclear whether any of this will bring any more transparency to political ads on Facebook. As CNN notes, "it's not clear whether the new rules would prevent a group or person from simply registering as a company, getting a tax identification number and going forward with their advertisements without ever having to really tell Facebook users who they truly are."

For a company that prides itself on "moving fast," Facebook is taking its time with half-baked fixes to an acute problem. 

"We can't do this alone"

Facebook is a $500 billion company. Political advertising is an extremely lucrative revenue stream. The top eight presidential candidates alone have spent over $25 million on Facebook ads this year. 

But Facebook insists that fixing its political ad system is a shared responsibility. It has created an "ad library," and it's hoping other people will scrutinize these ads for problems. 

"We know we can’t do this alone, and by housing these ads for up to seven years, people, regulators, third parties and watchdog groups can hold these groups more accountable," Facebook's Rob Leathern said in response to the VICE investigation last year. The company echoed this sentiment in Wednesday's announcement: "[W]e can’t tackle these challenges alone."

But Facebook is a highly profitable company and is running political ads to make money. Why should "people, regulators, third parties and watchdog groups" subsidize Facebook's profits by donating their time to fix a problem that Facebook created?

"Seriously flawed"

Whether it's fair or not, there are people -- including this author -- who are spending time monitoring political ads on Facebook for abuse. The most efficient way to do this work is by downloading data in bulk using a tool created by Facebook, known as an API. This tool, however, is badly broken, according to Laura Edelson, who monitors the ads for New York University's Online Political Ads Transparency Project.

Edelson says that Facebook's current system — which provides access via a complex online tool known as an application programming interface, or API — makes it almost impossible to track who's behind political ads, which users are targeted and how effective these paid-for messages are.

...Due to glitches within Facebook's own systems, Edelson says her searches often return incomplete data sets or fail to turn up any results.

Researchers for Mozilla attempted to analyze ads ahead of the UK's elections parliamentary elections earlier this year. It did not go well

However, due to the inconsistent state of the Facebook Ad Library API, our methods to scan and discover ads must be adapted on a daily and sometimes hourly basis — to deal with design limitations, data issues, and numerous software bugs in the Facebook Ad Library API.

Despite our best efforts to help Facebook debug their system, the majority of the issues were not resolved. The API delivered incomplete data on most days from its release through May 16, when Facebook fixed a critical bug. The API was broken again from May 18 through May 26, the last day of the elections.

Popular Information teamed with an experienced computer scientist to access the Facebook API. But despite significant effort, we were not able to consistently generate useful data. Most of the reporting you've seen in Popular Information about Facebook ads is the product of manually reviewing the ads one at a time. 

Humans needed

Only political ads are logged in the ad library and are subject to new verification requirements. Facebook relies on automated systems to identify political ads. But, according to researchers, "Facebook's systems are only able to automatically label just over half of these political ads when groups do not disclose that they are paid-for partisan messages." 

So, a group that is determined to spend money on political ads without disclosing its identity would probably be able to evade detection.

Beyond identity

Facebook doesn't just have a problem with political advertisers concealing their identity. There is also a problem with properly-identified advertisers running ads with prohibited content. None of the steps announced by Facebook on Wednesday address this problem. 

Popular Information reported last week that the Trump campaign was running numerous ads targeting women with prohibited content. The ads were only removed by Facebook after an inquiry from Popular Information. 

This was not an isolated incident. Facebook has been particularly reticent to enforce its prohibition on false and misleading content.

  • Facebook allowed the Trump campaign to run an ad that falsely claimed Joe Biden, Pete Buttigieg, and Andrew Yang supported eliminating private insurance.

  • Facebook allowed the Trump campaign to run a false ad targeting seniors that claimed he was still considering closing the southern border "next week" when he had already publicly announced he would not close the border for at least a year.

  • Facebook allowed the Trump campaign to run an ad scamming its supporters by claiming there was a midnight deadline to enter a contest to win the "1,000,000th red MAGA hat signed by President Trump." The ad was run every day for weeks. 

Again, the core problem is that Facebook is using an automated system to determine if an ad violates its rules. It "relies primarily on automated tools to check ads against these policies," the company told Popular Information. These automated tools are not working. 

Last year, the company promised to hire 3,000 to 4,000 people to manually review ads. It's unclear if these people were hired and, if so, why they aren't reviewing ads. 

Thanks for reading!