Could micro-targeting of politically inflammatory Facebook ads happen in Seattle, too? It already has.
Here's how it went down: Last year, during this city's municipal elections, a lawyer named Scott Lindsay was challenging two-term incumbent Pete Holmes for the position of Seattle City Attorney. To beat a popular incumbent like Holmes, it helps to make an argument that the incumbent has failed significantly, and so Lindsay laid out a number of arguments making that case.
None was more controversial, however, than Lindsay's argument about a Seattle property crime plague that had allegedly occurred on Holmes's watch.
Rolled out just as ballots were being mailed to voters, this Lindsay argument was easily summed up in a single sentence, like the one that made up the headline of Lindsay's October 19, 2017 press release on the matter: "New FBI data shows Seattle has highest property crime rate in the nation."
Having laid down this alarming claim, Lindsay went on to ague that he had a plan to succeed where Holmes had allegedly failed. Lindsay would, he promised, use smarter, progressive law enforcement tactics to break the cycles that lead to property crime in the first place.
The very next day, October 20, The Seattle Times's FYI Guy took aim at Lindsay's central claim in a story headlined, "No, Seattle doesn’t have the nation’s highest rate of property crime." (The Stranger initially published a blog post based on Lindsay's misleading claim. We took down the post upon further scrutiny and replaced it with a corrected version.)
Lindsay, who went on to lose the race for city attorney, told me by phone recently that he stands by his claims about crime in Seattle. Within the subset of "major cities" that Lindsay had chosen to compare Seattle to, our city's property crime rate was indeed the highest.
But as Gene Balk, the Times FYI Guy, had quickly pointed out in his article, the group of "major cities" Lindsay chose led to a distorted sense of reality. Balk even got Lindsay to admit "that he could have better defined his parameters," so that people would know Lindsay was only comparing Seattle to 20 major cities. (Rather than 50, as Balk said would be more appropriate. "I choose to look at the top 50 to capture a broader range of cities," Balk wrote. "Who wouldn’t consider Washington, D.C., or Boston a major U.S. city? Neither rank in the top 20.")
In response to Balk's October 20 story, Lindsay updated a campaign blog post he'd written outlining his "highest property crime rate in the nation" argument, so that the blog post came to contain more nuance. That post and Lindsay's entire campaign web site have since disappeared. But the revised post, which is archived here, made clear that Lindsay was only comparing Seattle to 20 other "major cities."
It all sounds like a tale of local watchdog journalism successfully getting a candidate to change his messaging so that it's more accurate and better informs voters.
And it was—if one's only sources of information were The Seattle Times and Lindsay's campaign blog, and if one read both completely and religiously, and checked each for updates.
But with two-thirds of Americans getting their news from social media, it's important to know that on Facebook, in Lindsay's paid advertising, things played out differently.
According to Facebook's recent disclosure of hundreds of ads aimed at Seattle's 2017 municipal elections—delivered in response to Stranger reporting—exactly one week after the Seattle Times story, Lindsay's campaign went right ahead with a 13-day Facebook ad campaign that micro-targeted Seattle neighborhoods with images of a very full-looking local crime map.
Along with the local maps covered in red "crime" dots was this un-nuanced message: "Seattle has the highest property crime rate per capita of any major city in the United States."
According to Facebook's disclosure, Lindsay put between a quarter and one half of all his Facebook advertising dollars behind the four ads that made up this campaign.
Here are Lindsay's four "Property Crime Report" ads, along with the text that, according to Facebook, accompanied each of them:
Lindsay argued that the ads weren't misleading because they sent people to his blog post, which offered more context for his "highest property crime rate per capita of any major city in the United States" claim.
"The point of the ad is that you click on the ad and then it takes you directly to the blog," Lindsay said. "The challenge here is that Facebook has a strict restriction on the amount of text you can use in the pictures. So there was only so much explaining we could do."
It's not clear how many people actually clicked on Lindsay's ads to get to the more nuanced picture of what he was claiming.
But Facebook's disclosure says the ads collectively received between 110,000 and 250,000 impressions. According to the definition of "impressions" that Facebook provided along with its disclosure, that only means that Lindsay's ads were "displayed on the screens of target audience members" between 110,000 and 250,000 times.
For context, when the ballot counting was all over, only about 142,000 total votes had been cast in the Lindsay-Holmes race (with the better-known Holmes winning overwhelmingly).
“People were already concerned about property crime," Lindsay said, explaining this ad campaign. "I wasn’t trying to scare people… I was just trying to tie them in and say, 'Let’s get smarter about this.'”
This is certainly not shadiness on the level of the Cambridge Analytica scandal. Who knows, it may not even be the most controversial thing a candidate did with social media advertising during Seattle's 2017 elections. (I'm still in the thick of looking at all that has, and hasn't, been disclosed.)
But what Lindsay's 2017 "Property Crime Report" campaign on Facebook makes clear is not just the potential for local candidates to use micro-targeting in an attempt to sway Seattle voters, but the fact that it's already happening.
Lindsay himself told me that, when it comes to micro-targeting on social media, "very clearly abuse is possible here."
The ads he paid Facebook to run all bore the same controversial message, no matter the neighborhood, so if the ads were misleading they were, in the end, equal opportunity misleading. But as Lindsay acknowledges, some future candidate, using the same methods and Facebook capabilities, could decide to aim one inflammatory or misleading message at one particular Seattle zip code where it wants to run up—or suppress—voter turnout in a close election.
Given this, and given the stakes for democracy and an informed public, it would be useful to have a mechanism for daylighting so-called "dark ads" that are aimed, via digital platforms, at only a small percentage of voters.
Fortunately, Seattle has a unique law that provides such a mechanism.
It requires Facebook and other digital platforms to disclose "the exact nature and extent of the advertising services rendered" to the Lindsay campaign—or to any other person or entity who buys election ads aimed at influencing our municipal races.
Unfortunately, Facebook's latest attempt at complying with this law only revealed limited information, from which one could then guess—but only guess—at whether a particular ad might have been micro-targeted.
Facebook revealed only that Lindsay's ads, and pretty much all the other Seattle ads it released information on, had reached people in "Washington." That's no doubt true, since these ads were all bought to influence Seattle's municipal elections, but it's not very helpful when it comes to trying to figure out whether, say, Lindsay's ads, which on their face seemed likely to have been micro-targeted at particular Seattle neighborhoods, were, in fact, micro-targeted.
In addition, when it comes to the amount spent on Seattle ads, Facebook has only disclosed a dollar range for each ad. For example, it reported that Lindsay spent $500 - $999 on each "Property Crime Report" ad. That's not very "exact," and leaves an open question as to whether Lindsay spent half or just one quarter of his total Facebook ad spend on the "Property Crime Report" campaign.
Because Facebook also failed to disclose information on each ad's intended audience—as Seattle Ethics and Elections Commission director Wayne Barnett said it should—I only know that Lindsay's ads were micro-targeted because Lindsay was willing to tell me.
The "Property Crime Report" campaign was, indeed, geographically targeted, Lindsay said. (And not just broadly targeted at "Washington," as the Facebook disclosure suggests.)
Each Lindsay ad was targeted at the particular part of Seattle it names. Then, within the targeted neighborhoods, his ads were further targeted at people determined to be "likely voters" in those neighborhoods.
Lindsay told me categorically: "The ads were not targeted by race, gender, or age."
This, too, was information that I couldn't get from Facebook's most recent disclosure. Facebook provided only "age gender reach" data for the ads it released, but the "reach" data doesn't make clear which ages or genders were actually targeted by the ads—or whether other demographic attributes (say, race, marital status, or parental status) were targeted, too.
I asked Lindsay whether he'd ask the consulting firm that handled the vast majority of his Facebook ads, Sermo Digital, to open up its digital ad data for me—similar to what the Holmes campaign did when I was exploring "The Case of the City Attorney's Google Ads."
Lindsay responded: "I've tried to be helpful here but I don't plan to ask my former campaign team for further work on this."
Seattle Ethics and Elections Commission director Wayne Barnett is still reviewing whether Facebook's most recent disclosure is robust enough to meet the requirements of Seattle law. I'll report back on what he says.
But it would seem that, in the case of Lindsay's targeted ad campaign, Facebook has not yet shared "the exact nature and extent of the advertising services rendered," as required by law.
Facebook, according to Lindsay, sold his campaign ads that were micro-targeted at various Seattle neighborhoods and "likely voters" within them. But Facebook's recent disclosure didn't make that clear.
"If our law says that advertisers must disclose, then they should disclose," Lindsay told me. "I think it’s good for there to be openness about what all the advertisements are.”
He added: "I would absolutely ask any vendor that I used to comply with the Seattle law."
Lindsay also noted, however, that he hasn't studied the Seattle law in question and, like others, is looking to Barnett for guidance on what it means in the digital era.
He also pointed out that social media is a favorite medium for insurgent and challenger campaigns. Trump enthusiastically embraced social media, of course, and locally outsider and long-shot candidates like Mike McGinn and Nikkita Oliver embraced it, too (from the opposite end of the political spectrum). “If there was real-time transparency," Lindsay mused, "it would make it harder for challengers to campaign.”
But it would also make it harder for challengers, incumbents, and everyone else to mislead small groups of voters through "dark" ads targeted at them via Facebook and other platforms.
When it comes to online political ad transparency, here in Seattle much rides on what Barnett does next.
Back in February, Barnett wasn't impressed with Facebook's first effort at disclosure. "We gave Facebook ample time to comply with the law, but their two-page spreadsheet doesn’t come close to meeting their public obligation," Barnett said at the time. "I’ll be discussing our next steps this week with the City Attorney’s office."
He's been happier with Facebook's second attempt, but has also said: "We will likely have questions for Facebook after we've completed our review."
In the meantime, perhaps you want to dive into Facebook's latest disclosure yourself? Well, here it is!
To download a .zip file with all the local ads Facebook has disclosed related to Seattle's 2017 municipal elections, click here.
For a .PDF of Facebook's cover letter accompanying the ads, click here.
For a key to terms used in Facebook's spreadsheet, click here.