Tuesday, December 28, 2021

Antonio García Martínez on Trump

I read all of Chaos Monkeys so you don't have to so I could bring you this, the author's take on the Cambridge Analytica scandal. This is a quick, heavily-editorialized summary of the Afterword of my edition (May 2018).

The book cover of Chaos Monkeys


In the author's opinion, Cambridge Analytica didn't have much impact. Their system of political profiling is basically astrology for political wonks; it has the same pitfalls as advertising models that try to decide if you want to buy a BMW because you're living in Iowa and working at a pet supply store (which is to say, those models aren't proven out).

The Russian interference existed, but likely had little impact given their numbers. They had more organic spread than the ads from other sources, but we're talking some hundred million reshares on a network that hosts trillions of posts per year.

The far likelier contributor to a Trump campaign victory was Trump's team themselves using Facebook's advertising tools to their maximum efficacy. Trump's own ads team has proven more than capable of getting people hype about him (these are, after all, the people who have been successfully selling Trump as a brand for decades). The metrics on how the Clinton campaign used online advertising are far bleaker: less microtargeting, less message-tuning, less use of the tools. I find this surprising given the success the Obama campaign had with online tools, but perhaps I shouldn't; the general story of the Dem party in 2016 was one of hubris, and I probably wouldn't be surprised to learn Clinton decided to kick all of Obama's good tech people to the curb and bring in her own coterie.

In particular, Trump's campaign took advantage of Lookalike Audiences, which is a FB feature that lets you start from a seed of users you want to target and builds out an ad targeting model along the friend network. Cambridge Analytica is doing astrology, but "Republicans are friends with Republicans" is just true, and has become ever more true as America cracks along its political stress-lines. 

So the executive summary:

  1. Trump's ad team worked their asses off to craft a message that resonated with potential Trump voters.
  2. Because Facebook makes it easy to reshare, those voters spread the message hard, for free.
  3. Meanwhile, both the Clinton campaign and Democrat voters (relatively-speaking) slept on getting a message spread (mostly, I think, because they already "knew" Trump was such an obvious no-choice that they didn't need to burn the energy). Everyone in their social networks was already voting for Clinton, so no need to advocate. While everyone in the Trump social networks was not already voting for Trump, so people who'd got hype spread the world harder.
In short, Trump's win was like a small-town store getting curb-stomped by the arrival of Walmart because Walmart brings out the marching band to announce their presence and the mom-and-pop knows everyone in town already knows 'em so they don't need to advertise and, anyway, who would shop at Walmart, they're obviously evil.

I'd like to say "Lessons learned," and I guess we'll find out.

... One more point worth mentioning, given where we are in 2021: our author mentions that when they were building the ad networks that Trump's team leveraged so effectively, basically nobody discussed politics. Those systems were designed to sell shoes, not commanders-in-chief. If you watch Facebook's current behavior after both 2016 and a January 6th insurrection attempt organized mostly on its network and Twitter's, always keep in mind: you're watching the behavior of a huge communications platform that well-and-truly did not see this coming and desperately wants it to be someone else's problem (because selling shoes is profitable but contributing to the dissolution of the world's largest democratically-elected superpower isn't).

I left Google for many reasons. One of the ones that looms over me perpetually is my feeling that when they were also faced with these questions, their response in aggregate as a corporation was "We'd rather not think about it." The remembered paraphrase of a quote from one of the company's two founders echoes in my brain, and I wish I'd had the foresight to write it down verbatim. The gist of the message (during a sex scandal related to, if memory serves, Google+) was that this founder didn't want to be dealing with a political scandal; he wanted to be making cool tech and solving problems. As if those are separable concerns.

Make your decisions accordingly.

No comments:

Post a Comment