Authored by Antonio Garcia-Martinez (former Facebook product manager), author of Chaos Monkeys: Obscene Fortune and Random Failure in Silicon Valley, originally posted at The Guardian,
For two years I was charged with turning Facebook data into money, by any legal means. If you browse the internet or buy items in physical stores, and then see ads related to those purchases on Facebook, blame me. I helped create the first versions of that, way back in 2012.
The ethics of Facebook’s micro-targeted advertising was thrust into the spotlight this week by a report out of Australia. The article, based on a leaked presentation, said that Facebook was able to identify teenagers at their most vulnerable, including when they feel “insecure”, “worthless”, “defeated” and “stressed”.
Facebook claimed the report was misleading, assuring the public that the company does not “offer tools to target people based on their emotional state”. If the intention of Facebook’s public relations spin is to give the impression that such targeting is not even possible on their platform, I’m here to tell you I believe they’re lying through their teeth.
Just as Mark Zuckerberg was being disingenuous (to put it mildly) when, in the wake of Donald Trump’s unexpected victory, he expressed doubt that Facebook could have flipped the presidential election.
Facebook deploys a political advertising sales team, specialized by political party, and charged with convincing deep-pocketed politicians that they do have the kind of influence needed to alter the outcome of elections.
I was at Facebook in 2012, during the previous presidential race.
The fact that Facebook could easily throw the election by selectively showing a Get Out the Vote reminder in certain counties of a swing state, for example, was a running joke.
Converting Facebook data into money is harder than it sounds, mostly because the vast bulk of your user data is worthless. Turns out your blotto-drunk party pics and flirty co-worker messages have no commercial value whatsoever.
But occasionally, if used very cleverly, with lots of machine-learning iteration and systematic trial-and-error, the canny marketer can find just the right admixture of age, geography, time of day, and music or film tastes that demarcate a demographic winner of an audience. The “clickthrough rate”, to use the advertiser’s parlance, doesn’t lie.
Without seeing the leaked documents, which were reportedly based around a pitch Facebook made to a bank, it is impossible to know precisely what the platform was offering advertisers. There’s nothing in the trade I know of that targets ads at emotions. But Facebook has and does offer “psychometric”-type targeting, where the goal is to define a subset of the marketing audience that an advertiser thinks is particularly susceptible to their message.
And knowing the Facebook sales playbook, I cannot imagine the company would have concocted such a pitch about teenage emotions without the final hook: “and this is how you execute this on the Facebook ads platform”. Why else would they be making the pitch?
The question is not whether this can be done. It is whether Facebook should apply a moral filter to these decisions. Let’s assume Facebook does target ads at depressed teens. My reaction? So what. Sometimes data behaves unethically.
I’ll illustrate with an anecdote from my Facebook days. Someone on the data science team had cooked up a new tool that recommended Facebook Pages users should like. And what did this tool start spitting out? Every ethnic stereotype you can imagine. We killed the tool when it recommended then president Obama if a user had “liked” rapper Jay Z. While that was a statistical fact – people who liked Jay Z were more likely to like Obama – it was one of the statistical truths Facebook couldn’t be seen espousing.
I disagreed. Jay Z is a millionaire music tycoon, so what if we associate him with the president? In our current world, there’s a long list of Truths That Cannot Be Stated Publicly, even though there’s plenty of data suggesting their correctness, and this was one of them.
African Americans living in postal codes with depressed incomes likely do respond disproportionately to ads for usurious “payday” loans.
Hispanics between the ages of 18 and 25 probably do engage with ads singing the charms and advantages of military service.
Why should those examples of targeting be viewed as any less ethical than, say, ads selling $100 Lululemon yoga pants targeting thirtysomething women in affluent postal codes like San Francisco’s Marina district?
The hard reality is that Facebook will never try to limit such use of their data unless the public uproar reaches such a crescendo as to be un-mutable. Which is what happened with Trump and the “fake news” accusation: even the implacable Zuck had to give in and introduce some anti-fake news technology. But they’ll slip that trap as soon as they can. And why shouldn’t they? At least in the case of ads, the data and the clickthrough rates are on their side.