Review: The Hype Machine

 I read The Hype Machine hoping to find some answers about social media and how it was used to destroy democracy. I did learn a few things, some of which you probably already knew, about the lies spreading and being more easily clicked on and read and believed than the truth:

Abnormal trading volume rose by 37 percent over the three days following the publication of real news articles and 50 percent more following the publication of fake news articles relative to real news articles. In other words, investors reacted to fake news even more strongly than to real news. The reactions were more pronounced for smaller firms and for firms with a greater percentage of retail (as opposed to institutional) investors. Fake articles were clicked on and read more often than real articles, and trading volume increased with the number of clicks and times an article was read. (Kindle loc 685)

I learned the technical term for measuring the effectiveness of a social ad, "lift", which basically measures the influence and ad has for changing your actions. This is more subtle than you might think, since for instance, an ad that's shown to you that confirms your biases and reminds you to buy something you would have bought anyway doesn't contribute to lift.

What the book does not provided are easy answers. At no point does the author Aral come to a conclusion as to whether the Russians succeeded in throwing the 2016 election to Trump. He does not that the amount of lift a campaign of that size, if targeted properly at the right voters in the right places, would decide the election:

“the proportion of misinformation was twice that of the content from experts and the candidates themselves.” When they calculated whether a state had more or less Russian fake news, they found that 12 of 16 swing states were above the average. They concluded that Russian fake news was “surprisingly concentrated in swing states, even considering the amount of political conversation occurring in the state.” Although more than 135 million votes were cast in the 2016 U.S. presidential election, six swing states (New Hampshire, Minnesota, Michigan, Florida, Wisconsin, and Pennsylvania) were decided by margins of less than 2 percent, and 77,744 votes in three swing states (Wisconsin, Michigan, and Pennsylvania) effectively decided the election. (kindle loc 778)

He notes that a society that can converge onto the truth is wise, and clearly we do not live in such a society:

They called societies that did so “wise.” They found that a society’s network has one simple and complete characterization of wisdom—one requirement for being able to arrive at the truth. And that is that it does not have, in modern vernacular, “influencers.” As Golub and Jackson wrote, “Disproportionate popularity is the sole obstacle to wisdom….Having agents who are prominent, causes learning to fail, since their influence on the limiting beliefs is excessive.” Societies without wisdom lack “balance,” meaning some groups have disproportionately more influence than others and may not pay sufficient attention to the rest of the world. Sound familiar? This is the world we live in, where Barack Obama and Donald Trump have 110 million and 67 million Twitter followers, respectively. Where Kanye West has 30 million followers and follows 300 people. (Kindle loc 4496)

Basically, societies where a few influencers lead and a bunch of followers follow without thinking for themselves will lead to a polarized society where truth loses value and arguments are meaningless because nobody changes their minds. Unfortunately, like many researchers and media people, Aral commits the mistake of "bothsidesism" when clearly one side has gone mad while the other still has some grip on reality. 

Another interesting phenomenon is how protest movements in recent years appear to gain viral momentum only to disperse with no impact on policy or lasting change:

Zeynep Tufekci’s book Twitter and Tear Gas provides a comprehensive deep dive into how social media has changed social movements. Her thesis, backed by stacks of research, suggests that protest movements enabled by technology rise rapidly but often sputter at their zenith. At the precise moment they make headway and capture the world’s attention, they experience what Tufekci calls “tactical freeze”—an inability to adjust tactics, negotiate demands, and push for tangible policy changes. What causes tactical freeze? The rapid mobilization that the Hype Machine enables is typically accompanied by leaderless, ad hoc decision making and a shallow organization that develops without much early planning. (Kindle Loc 4769)

This is in contrast to the social movements that brought lasting change in the past, which erupted into protests only after years of organizing, training, and discussing how to bring about that change. So social media has corrupted even basic functions such as the ability to organize and protest and bring about change.

he very technology that enables protests can be co-opted by the governments they oppose. In 2019, for example, the Chinese government used disinformation on social media to disrupt the Hong Kong protests and to change the domestic and international perception of the protesters by exaggerating the harm caused to bystanders. The Chinese government is suspected of hiring as many as 2 million people to insert propaganda into social media; the political scientist and statistician Gary King and his colleagues at Harvard estimated that they fabricate and post about 448 million social media comments a year toward this end. In Russia, Putin’s allies simply took over VK and squashed protesters’ online presence. As the stories of modern protests make clear, the Hype Machine enables social mobilization but in a fragile way. (Kindle loc 4779)

Is there any hope? Aral suggests that breaking up Facebook, for instance, isn't as useful as forcing Facebook to allow competitors to reach into its network and integrate with it. This would generate true competition. In this case I think Aral commits the usual media mistake of ignoring the fact that by forcing Facebook to spinoff Instagram and WhatsApp and then implementing the forced integration API at arms length (and allowing competitors to access that same integration) would immediately also provide competition where there was none.

All in all, I think the book is worth your time --- but be careful of Aral's conclusions. I don't know if his intentions are pure (he certainly spends many words in his book promoting himself and his relationship with Facebook which enabled him to perform experiments on Facebook users at scale and collect data), but he seems altogether too protective of Facebook's monopoly for me to take anything he says at face value.