The Hype Machine by Sinan Aral is a book filled with insights into the inner workings of our modern social networks and the social media platforms behind them. Here are a few key points:
1. Clusters of Homophily
The connections we make on social media can be visualized as a social graph, like the one you see above; the dots represent users and the lines represent ties between each user. The friends you make on social media are usually those who are similar to you in some way; perhaps you live in the same community, go to the same school, share similar interests, professions, or hobbies. Other friends are made through the weak ties you hold with another person due to your relationship with a mutual friend. In both of these scenarios, the People You May Know (PYMK) algorithms on social media will connect you with these people, virtually. This is why the social graph shows large clusters of people (depicted by color) -they are all similar in some way, so they connect to one another via social media. In the book, Aral uses the word ‘homophily’ to describe a group containing similar kinds of people. Therefore, the cluster(s) a particular person is a part of typically represent an overlying theme that connects each one of the users within that cluster. For example, a cluster might include all the social media users who go to the same high school or college, or are members of the same political party.
2. Innovation Brokers
An innovation broker is a person who introduces new ideas, concepts, and information to a cluster. These people are often attributed to being successful in their work, or are popular amongst those in their social group. The innovation broker, in the context of a social graph, is simply a person who has ties to another person located in a different cluster. This means that when novel information spreads across the users in one cluster, an innovation broker will capture and share that information with another user within their own cluster, effectively facilitating the transfer of information from one group of people to another.
3. The Power of Network Effects
Network Effects are an intriguing concept in the context of digital social networks. In simple terms, network effects are like a magnet that brings more users to a platform. When more users join a platform, the network effects of that platform are magnified. When you made the choice to join Instagram, for example, the reason you made that decision was in part due to strong [local] network effects -that is, a lot of people in your social group already had Instagram, so it was more beneficial for you to create an account on their platform. This is why there are only a few big social networks — in order to gain a sizable market share, any other competitor would have to have a product that significantly rivals the quality of existing products to such a degree that it could overcome that social media platform’s strong network effects.
4. Polarization of Populations
The design of modern-day social media platforms is one that acts as an “echo chamber” for our personal and political beliefs. The recommendation algorithms of the social media platforms we use are optimized to present information that resonates with our internal beliefs. Thus, as algorithms are given access to our personal data and begin to infer our preferences, they start recommending content that appeals to our tastes. This is why, before the rise of social media (see chart above), the average Republican and Democrat did not lean too far right or too far left, respectively. Looking at the same chart, the year 2017 reveals a different story; each political party is more far-right or far-left leaning than the 1994 survey. While this information doesn’t represent a causal link between the rise of social media and political polarization, many indicators point to social media as one of the key reasons for this increased polarization.
5. Micro Targeting
The advent of social media has given rise to a new type of currency: attention. On social media, businesses, institutions, and other entities are competing for your attention to sell a product, promote an idea, inform citizens about public health and safety measures, or gain political endorsement, among other things. To optimize the click-through rate (a common metric for measuring interaction with a digital ad) of their messaging, social media algorithms will employ micro targeting tactics to curate advertisements based on web searches you make. By modeling the characteristics of a user, targeting algorithms can decide what content to show a user in order to maximize the CTR on a particular ad.
6. A Tool For Misinformation
Social media networks such as Facebook, Twitter, and Instagram have opened the door to a new age of information sharing: information can now reach a mass audience just minutes after it is published. While novel information and live reporting on world affairs have always been available through television networks since the mid-1900s, the difference with information on social media is that anyone can propagate misinformation, or fake news, regardless of how many followers they have (and thus, the accountability of the source is no longer an issue). In the last decade, bad actors have used a combination of internet bots and “troll” accounts to spread misinformation to sway populations to act in their interest, profit from advertisement revenue on fake news websites, and interfere with elections. In one study conducted by researchers at MIT, the spread of fake news was measured to be over six times faster than the spread of facts. In the latter chapters of The Hype Machine, Aral addresses potential solutions to help users determine fact from fiction, such as labeling the credibility of a post, or limiting the spread of fake news by restricting the number of times a user can reshare information. But until such solutions are widely implemented, we must continue to use social media with the knowledge that the information we consume might be untrustworthy.