Fear and anger as features in the social media ecosystem
In response to Mattias Desmet's The Psychology of Totalitarianism
As a Substack fan and writer, I am immensely encouraged to see Dr. Mattias Desmet join Substack to continue the thinking he has published in The Psychology of Totalitarianism. I am going to go on record as saying this book will end up being recognized as one of the most important books of the nascent 21st century.
Instead of writing a typical review, I am going to exercise my abundance of chutzpah and try to identify what I think is missing in professor Desmet’s analysis: The nature of social media as a “system;” the two central features of the system - fear and anger; and how this represents the sine qua non (“that without which”) of totalitarian mass formation. When we realize the one thing which must be present to sustain mass formation, and how that thing becomes present in today’s social media environment, we will then understand exactly how to respond.
The social media (eco)system and its use case
Let’s start by defining what we mean by the “social media ecosystem.” It is readily apparent that we have platforms like Facebook, Instagram, TikTok, Twitter, etc. as the core of the social media landscape. Three things are not as readily apparent: 1) All online interactions - browser use, search engine use, email, etc. are also part of the social media ecosystem; 2) Text messaging services provided by cell phone carriers are also part of this ecosystem; and 3) Legacy news media programming must also be considered.
Below I will discuss the “social graph” as the technological embodiment of the social media ecosystem. In broad strokes, this type of data technology - graph data - reduces everything to “nodes” and “edges.” A node is always a noun - a person place or thing. An edge is always a verbal phrase which captures the relationship between two nodes.
Every person who uses any form of Internet technology is a “node” in the social graph (you are surely a person, place, or thing). To oversimplify a bit, every informational “thing” - a post on Twitter, an email, a website and its content, a text message, etc. - is also a node. As nodes interact with other nodes, the social graph captures the nature of those interactions as edges. Data science, then, can design controlled experiments to determine what kinds of content (nodes) will generate what kinds of interactions (edges) from what kinds of people (nodes).
Data science can determine what kinds of content (nodes) will generate what kinds of interactions (edges) from what kinds of people (nodes).
If there are two possible edges between us and the content we encounter as nodes, there are thousands. But only two really matter: X “angers” Y; and: X “scares” Y.
Professor Desmet has identified the nature of “mass formation” for us. We can recognize past examples of Nazi Germany and Stalin’s Soviet Union. We cannot, however, understand this modern instance of mass formation unless we are clear about how modern technology has changed both its why’s and how’s.
Audience Engagement: The output of the social media ecosystem
In the United States those of us old enough to recall life before hundreds of cable channels, email, the web, social media, etc. will probably understand what a “Nielsen Household” is. Nielsen is the company responsible for U.S. television and radio ratings. A Nielsen Household is a household engaged by Nielsen and agreeing to host Nielsen’s monitoring technology used to sample audience engagement with media programming.
If we look at this strictly from the standpoint of technology, we are all Nielsen Households now. The kind of information generated by our use of Internet devices (smart phones, tablets, desktops, etc.) is all but identical to the information generated by the technology fielded by Nielsen to their Nielsen Households. But there is one aspect of audience engagement Nielsen’s technology cannot capture - emotions. I’ll explain more below, but the social media ecosystem now makes it possible to gather detailed information on how media content makes us feel - and at a visceral level.
Media is not Red or Blue - it is Green
To better explain this, I’ll make an unusual claim here: The major news media outlets do not lean one way or the other in terms of political ideology. Fox News is not “conservative” and CNN or MSNBC is not “liberal.” They are not Red or Blue - they are decidedly Green - as in the U.S. Dollar, not the environment. They exist for one reason and one reason only: to identify, attract, maintain, and then monetize the engagement of an audience.
Where they used to rely on relatively limited information statistically extrapolated from the few households hosting Nielsen’s technology, they now rely on the social graph to develop a target audience segment. Fox News is not “conservative,” per se; their audience is a segment of nodes from the social graph who have been “labeled” (more on this below) as conservative. The same holds true for outlets like CNN and MSNBC; they have segmented the social graph to identify users (nodes) labeled as liberal.
So, we know our target audience. Now what?
Nielsen’s technology once captured audience engagement with programming across traditional demographic groups like age and income. But what it did not do is get down to the content in the programming to develop knowledge about exactly what content within the programming was most important to generating audience engagement, and furthermore, what about that content was key.
Thanks for reading Thomas Paine's Blog! Subscribe for free to receive new posts and support my work.
Fear and Anger: The essential feature driving audience engagement
We can go all the way back to Aristotle and see that fear and anger have long been understood to be tools of persuasion. But what we have not been able to do over this time is measure fear and anger. By measure, I do not mean in isolation. When we report pain to our doctor, he or she will often ask us to rate the pain from one to ten. I am not claiming an objective 1 to 10 scale for either fear or anger. We can, however, objectively measure fear and anger in the context of a user on social media (as a node) and the content with which they engage (also nodes). Allow me to dive into the technology a bit to explain.
As you scroll, content will move into view from the bottom and out from view at the top. The viewable area is called the “viewport.” When you stop scrolling the event listener reports what post(s) are in the viewport. So as a software engineer, I not only know when you stop scrolling, I know what content you stopped on... But I do not necessarily know WHY you stopped on it... unless you tell me.
If you take nothing else from this essay: STOP USING EMOJIS!
Enter the emoji. These are not graphics in the sense we probably think. They are what are called “code-points” (numbers) in a table of values called Unicode. Each letter of the Roman alphabet is also a code-point. The number is sent to a file defining the font to be shown, and the font file then tells the computer how to render the letter. If we are going to use words to express anger or fear - well, if there is one way to say it there are hundreds, and that is just in one language.
But now imagine that no matter what language you are using on social media, the same code-point can be used to express anger or fear (by way of the emoji). This code-point becomes an “edge” on the social graph. We now have a computationally consistent way of joining the person on social media (as a node) to the content (also as a node) with an edge we can simply define as “...is angered by...” or “...is scared of...” We no longer need subjective interpretations of otherwise normal human language.
Labels: the unseen part of the social graph
We have discussed the two core aspects of graph data - nodes and edges. There is a third that, without which, the nodes and edges really don’t tell us much. These are labels associated with nodes and edges. These are also called “meta-data” - or “data describing data” and are how data segments are created.
We can go back to our discussion of Nielsen Households to understand labels. It is unlikely the reader will be surprised that information about their income levels and consumer interests is being captured and used to determine what kinds of advertisements are included in their “feeds.” What we may not be aware of is how a very wide range of original data sources are being extracted, transformed, and loaded into “Big Data” systems to create the social graph. Perhaps most important among these sources are voter registrations.
Here we have to address the political angle to “audience engagement.” Political consultants now use social media data married up to voter registrations - usually via the voter’s email address - to identify a segment of voters to whom they will focus their appeal. The salient question is: “What gets you to the polls?” When you take the answers to that question and correlate them to fear and anger, the social graph reveals them to be the same: The things that make voters afraid and angry are the things which get them to the polls.
So now, if I am a political consultant, my question for data science is: What are the three issues (as seen from the content nodes) which make my identified social media user segment (user nodes) most angry and/or afraid? With the nature of the information now available in the social graph (especially with the use of emojis), getting a highly reliable, predictive answer to that question is trivial. Now I know what my candidate’s stump speeches will stress. If I succeed in keeping my voters angry and afraid, I will succeed in keeping them engaged in my candidate's campaign.
Fear and anger as the sine qua non of mass formation
Professor Desmet shows how mass formation can emerge spontaneously (Nazi Germany) or be deliberately cultivated (Stalin’s Soviet Union).
[Mass formation] can emerge in different ways. It can emerge spontaneously (as happened in Nazi Germany), or it can be intentionally provoked through indoctrination and propaganda (as happened in the Soviet Union). But if it is not constantly supported by indoctrination and propaganda disseminated through mass media, it will usually be short-lived and will not develop into a full-fledged totalitarian state. Whether it initially emerged spontaneously or was provoked intentionally from the beginning, no mass formation, however, can continue to exist for any length of time unless it is constantly fed by indoctrination and propaganda disseminated through mass media.
I believe our attention should be less on propaganda, per se, and more on the underlying emotional state fed by the propaganda. Desmet locates a “free floating anxiety” emerging from a “mechanistic ideology.”
At the level of the population, the mechanist ideology created the conditions that make people vulnerable for mass formation. It disconnected people from their natural and social environment, created experiences of radical absence of meaning and purpose in life, and it led to extremely high levels of so-called “free-floating” anxiety, frustration, and aggression, meaning anxiety, frustration, and aggression that is not connected with a mental representation; anxiety, frustration, and aggression in which people don’t know what they feel anxious, frustrated, and aggressive about. It is in this state that people become vulnerable to mass formation.
While I agree with his larger views on this “mechanistic ideology,” the ideology itself is powerless without some manner of persuasion behind it. I liken it to the difference between power and voltage. You can generate electrical power, but in order to transmit that power over long distances you need to generate voltage. Ideology is like power; persuasion is the “voltage” needed to transmit the ideology.
Let’s restate Desmet’s quote above, with some key substitutions in brackets.
At the level of the population, [social media nurtures] the conditions that make people vulnerable for mass formation. It [keeps people angry at and afraid of their political other], create[s] experiences of radical absence of meaning and purpose in life, and [has] led to extremely high levels of so-called “free-floating” anxiety, frustration, and aggression, meaning anxiety, frustration, and aggression that is not connected with a mental representation; anxiety, frustration, and aggression in which people don’t know what they feel anxious, frustrated, and aggressive about. It is in this state that people become vulnerable to mass formation.
Let’s look again at the first quote above from Desmet’s work. At the end he says: “[N]o mass formation, however, can continue to exist for any length of time unless it is constantly fed by indoctrination and propaganda disseminated through mass media.”
Indoctrination and propaganda are secondary to fear and anger. If you are part of a political segment (captured from analyzing the labels on the social graph), data science can also identify the segment on the social graph which is your “Political Other.” Ideologues can certainly propagandize you, but to what end? That you agree with an openly presented political ideology? Or that you learn to see a segment of your neighbors as the Political Other.
This is where I think Desmet’s analogy to hypnosis is most on point. Once you have been hypnotized by fear and anger directed at your Political Other, you naturally look for an answer or solution. There is no need to overtly propagandize one who has already been hypnotized by fear and anger; they will propagandize themselves quite well on their own.
Four key lessons
Let me leave with four lessons and four recommendations.
1) Some, but not all, political actors are actively exploiting the free floating anxiety created by social media to feed their otherwise amoral totalitarian messianic impulses. (I believe Klaus Schwab, the rest of the Davos crowd, George Soros, Bill Gates, and others like them are examples of this. I also believe there are some among hard core right-wing Evangelicals Christians who have deceived themselves into believing there are political ends that will somehow “speed up” what they believe to be events of the “end times.” This line of thought is often called Dominion Theology.
While I cannot speculate as to any of their motives, they all strike me as suffering from the same amoral totalitarian messianic delusion we can see in examples from Nazi Germany. (Compare Hannah Arendt’s Eichmann in Jerusalem with Soros’ own writings and his justification for helping the Nazis confiscate Jewish property.)
2) The information needed to scientifically determine how to keep a population angry and afraid arises from our own use of social media and tech platforms. We simply must learn that none of this is possible without us (unwittingly) providing the needed data points.
3) The larger business model of Big Tech is simple: The monetization of the discovery of knowledge about us. And the most valuable of that knowledge has to do with what makes us angry and afraid. Information technology takes data, brings it into context with other data to create information, and then uses machine learning to discover knowledge. To remind us again of #2 above - none of the data needed to create this information exists unless we provide it.
4) Media no longer follows the rule: “If it bleeds, it leads.” Today, if it pisses you off or otherwise makes you fearful, it will consistently be covered on the news you consume. And you will likely not realize you are literally becoming addicted to alterations in brain chemistry triggered by that fear and anger. It is why you keep coming back to the programming.
So, what should we do?
1) If nothing else, follow Professor Desmet’s work closely and look for repeats of the patterns he highlights.
2) Ask critical questions about every “crisis” reported in the news. Is it the latest way to keep you afraid? When something is presented as a crisis, and dissenting viewpoints are delegitimized (e.g. “climate deniers” are equated with “Holocaust deniers;” a “pandemic of the unvaccinated,” dissent is considered “dangerous dis-information,” etc.) you should take careful note how these two things happen together - an appeal to fear on the one hand and an effort to stoke anger toward dissenting voices on the other. It is not likely a coincidence.
3) Following from #2 above, nurture emotional self-awareness as you interact with content online. And be far more guarded with how you communicate those emotions. Above all else - stop using emojis. They provide the critical edge data point by which data science can then actually predict how we will respond to content.
4) Be sure to be skeptical especially of your preferred news outlets. They monetize your engagement with their content like any other media organization. The key test should be whether they seem to always need to ratchet up the fear and outrage. Desmet shows that propaganda has to be ever more absurd in order to hold the formation of the masses. This is true of fear and anger. A media outlet that depends on it for audience engagement will have to always move to something ever more outrageous and fearsome than the last thing.