For years we have heard that social media creates echo chambers because people within different social media networks consume divergent information. The most common theory as to why people are becoming more politically divided is that people are ignorant to the other side’s views because of the way algorithms section different people into different digital networks. Although this theory is the most widely accepted, it doesn’t adequately describe how social media platforms create public opinion today. The true form of postmodern divide is not that citizens do not see the same things. Instead, the same things are seen through wholly different algorithmic structures.

I have recently been reminded of this when discussing a political protest with a friend that gained significant coverage. We both did see the protest because we both watched the same marches, speeches, and many of the same protest related videos. Our conversation quickly turned to disbelief. Protest coverage on my feed was framing the movement as disruptive to civil order. The same coverage on my friend’s feed was framing the movement as protecting civil order. Again, we are not talking about different protests. We are discussing divergent viewpoints of the same protest.

This distinction is important. When most people talk about misinformation, they focus on the issue of the distribution of falsehoods or the absence of common facts. However, things have changed. The main issue is elsewhere. While most people focus on the event that attracted the most attention and the most outrage, Algorithms focus on the emotional and moral aspects of the event. The same Protest or Election or Court verdict or Policy Announcement can be different for different users and can be viewed through the lenses of entirely different meanings.

I am calling this new phenomenon Algorithmocracy. These algorithms do much more than recommend or suggest information. Algorithms actually govern public opinion. They don’t tell us what to think. In fact, they do the opposite. They tell us how to think, by surrounding us with certain information, emphasising certain facts, elevating certain reactions, and interpreting certain viewpoints. The real power of these systems is not how they control the information, but how they control how we think.

This model’s opposite of the typical “filter bubble” is built on collision rather than isolation. What social media platforms have learned is that there is nothing more engaging and thus more important to sustain user interest than national disputes and controversies. So, they make sure that every user is exposed to the controversies. Social media also knows that users more readily engage with content that aligns with their emotional state. Years of behavioural data of everything we have done and every subtle shift we have made such as pauses, comments, shares, likes and outages, give recommendation systems the ability to predict the best framing to engage us. This results in one user being presented content about law and order while another user is confronted with content about civil liberties. They both are covering the same subject but see a moral context that is specific to them.

The result of this is not just simple disagreement. Traditionally, when people are confronted with differing viewpoints, the response is usually along the lines of, “Oh, I didn’t know that.” But when two people talk about the same event but end up with opposing conclusions, the response is not fact vs. fact but moral vs. moral. It is assumed that the other person has witnessed what has been witnessed. From this, the response is to question the other person’s judgment, morality, intelligence, etc. This happens not because the facts are diverging but because facts are presented emotionally.

This architecture is largely commercial, but not ideological. Social media companies prioritise engagement, whereas presentations that are calm, balanced and nuanced do not encourage scrolling. Public outrage, certainty and moral urgency do. Therefore, algorithms will segment its user base and continuously show users interpretations of events that elicit the greatest emotional response. The goal, however, is not comprehension of events but the preservation of engagement (a.k.a. attention).

This is a relatively straightforward process. A large, nationally prominent event captures users’ attention. Within many users’ feeds, recommendations for the most popular interpretations will be prominently displayed. Users who do not agree will typically engage by posting comments, sharing the alternative views and/or verbally attacking the participants of the recommendations. The ultimate goal of the system is to sustain and encourage engagement.

This situation is extremely troubling for Liberal Democracy. Vigorous disagreement is the cornerstone of Liberal Democracy. In such a system, it is expected that citizens will disagree on value and policy issues, but will still have broad concordance on the events that are of public importance. Algorithmocracy shows that such a society can have shared events, videos and even prominent moments of public concern and still not share a concordance of public importance.

The consequences reach further than just politics. A minimum amount of shared reality is necessary for public trust, social unity, and the existence of legitimate institutions. Citizens begin to lose trust in each other and the existence of objective public debate as algorithms create a preference for maximising emotionally engaged interpretations in place of socially constructive interpretations. Each argument is simply another representation of the separate camps where supporting evidence is perceived to exist exclusively for each camp’s assertions.

Political theories have addressed the power and scope of politics and the issues of representation for centuries; the focus has historically been on who governs? In the context of the new digital reality, an equally valid question is who governs the representations of public life for the citizens? The power increasingly lies not with public representatives and institutions, editorial gatekeeping, or public service broadcasters, but with commercial enterprises focused on optimising profit over democracy, thinly disguised as social welfare.

Consequently, the greater challenge of the century is not the need for additional information or the addressing of misinformation. The architecture of attention and interpretation is equally fundamental for democracy. If we persist in viewing systems that determine information flows and audience engagement for maximum commercial profit as benign, neutral systems, the fault of citizen disengagement and of systems that strengthen and exacerbate polarisation will continue to be misplaced on citizens. The principal quandary of democratic systems now is not the fight for control of the information systems that dictate the narrative. Rather it is the fight for control over the systems that determine information engagement and audience interpretation.

-The Statesman (India)/ANN