Roger McNamee, an early Facebook investor and mentor to Mark Zuckerberg, in Washington Monthly:
Whenever you log into Facebook, there are millions of posts the platform could show you. The key to its business model is the use of algorithms, driven by individual user data, to show you stuff you’re more likely to react to. Wikipedia defines an algorithm as “a set of rules that precisely defines a sequence of operations.” Algorithms appear value neutral, but the platforms’ algorithms are actually designed with a specific value in mind: maximum share of attention, which optimizes profits. They do this by sucking up and analyzing your data, using it to predict what will cause you to react most strongly, and then giving you more of that.
Algorithms that maximize attention give an advantage to negative messages. People tend to react more to inputs that land low on the brainstem. Fear and anger produce a lot more engagement and sharing than joy. The result is that the algorithms favor sensational content over substance. Of course, this has always been true for media; hence the old news adage “If it bleeds, it leads.” But for mass media, this was constrained by one-size-fits-all content and by the limitations of delivery platforms. Not so for internet platforms on smartphones. They have created billions of individual channels, each of which can be pushed further into negativity and extremism without the risk of alienating other audience members. To the contrary: the platforms help people self-segregate into like-minded filter bubbles, reducing the risk of exposure to challenging ideas.
It took Brexit for me to begin to see the danger of this dynamic. I’m no expert on British politics, but it seemed likely that Facebook might have had a big impact on the vote because one side’s message was perfect for the algorithms and the other’s wasn’t. The “Leave” campaign made an absurd promise—there would be savings from leaving the European Union that would fund a big improvement in the National Health System—while also exploiting xenophobia by casting Brexit as the best way to protect English culture and jobs from immigrants. It was too-good-to-be-true nonsense mixed with fearmongering.
Meanwhile, the Remain campaign was making an appeal to reason. Leave’s crude, emotional message would have been turbocharged by sharing far more than Remain’s. I did not see it at the time, but the users most likely to respond to Leave’s messages were probably less wealthy and therefore cheaper for the advertiser to target: the price of Facebook (and Google) ads is determined by auction, and the cost of targeting more upscale consumers gets bid up higher by actual businesses trying to sell them things. As a consequence, Facebook was a much cheaper and more effective platform for Leave in terms of cost per user reached. And filter bubbles would ensure that people on the Leave side would rarely have their questionable beliefs challenged. Facebook’s model may have had the power to reshape an entire continent.