Trump and Facebook, Sitting in a Tree

 

I've been reading an insider's book about the peculiar brand of selfishness and narcissism at Facebook -- Mark Zuckerberg and Sheryl Sandberg and on down the ranks of upper management -- written by Sarah Wynn-Williams: Careless People: A Cautionary Tale of Power, Greed, and Lost Idealism (published just last November). According to John Walters, Facebook management tried to suppress it, "which in fact only increased its sales."

No wonder Facebook felt a fit of censorship. Wynn-Williams's title for this takedown of a media monster -- Careless People -- comes from The Great Gatesby, a passage the author quotes as an epigraph: “They were careless people, Tom and Daisy – they smashed up things and creatures and then retreated back into their money or their vast carelessness, or whatever it was that kept them together, and let other people clean up the mess they had made.”

There's no doubt that Facebook had a huge hand in unleashing Donald Jethro Trump in the election of 2016. In fact, according to Wynn-Williams, Facebook staff worked collaboratively with Trump's campaign to mastermind the "single best digital ad campaign" that several experts in mass communications had ever seen. Wynn-Williams gets into some graphic detail:

A Trump operative named Brad Parscale ran the operation together with the embedded Facebook staff, and he basically invented a new way for a political campaign to shitpost its way to the White House, targeting voters with misinformation, inflammatory posts, and fundraising messages....

...Facebook and Parscale's combined team microtargeted users and tweaked ads for maximum engagement, using data tools we designed for commercial advertisers. The way I understand it, Trump's campaign had amassed a database, named Project Alamo, with profiles of over 220 million people in America. It charted all sorts of online and offline behavior, including gun registration, voter registration, credit card and shopping histories, what websites they visit, what car they drive, where they live, and the last time they voted. The campaign used Facebook's "Custom Audiences from Custom Lists" to match people in that database with their Facebook profiles. Then Facebook's "Lookalike Audiences" algorithm found people on Facebook with "common qualities" that "look like" those of known Trump supports. So if Trump supporters like, for example, a certain kind of pickup truck, the tool would find other people who liked pickup trucks but were not yet committed voters to show the ads to.

Then they'd pair their targeting strategy with data from their message testing. People likely to respond to "build a wall" got that sort of message. Moms worried about childcare got ads explaining that Trump wanted "100% Tax Deductible Childcare." Then there was a whole operation to constantly tweak the copy and the images and the color of the buttons that say "donate," since slightly different messages resonate with different audiences. At any given moment, the campaign had tens of thousands of ads in play, millions of different ad variations by the time they were done. These ads were tested using Facebook's Brand Lift surveys, which measure whether users have absorbed the messages in the ads, and tweaked accordingly. Many of these ads contained inflammatory misinformation that drove up engagement and drove down the price of advertising. The more people engage with an ad, the less it costs. Facebook's tools and in-house white glove service created incredibly accurate targeting of both message and audience, which is the holy grail of advertising.

Trump heavily outspent [Hillary] Clinton on Facebook ads. In the weeks before the election, the Trump campaign was regularly one of the top advertisers on Facebook globally. His campaign could afford to do this because the data targeting enables it to raise millions each month in campaign contributions through Facebook. In fact, Facebook was the Trump campaign's largest source of cash.