The Social Dynamics of Micro-Blogging Social Media Services

And Why I Think Micro-Blogging is Intellectual Laziness Anyway

Preliminary Note: As usual, the end of the year is hectic, and I’m spending quite some time preparing special content for this newsletter that should be ready for Christmas Eve. Posting should be light until then, and today’s post is more a digression from my usual topics than anything else.

While preparing the special content I just referred to in the preliminary note, I read Arthur Schlesinger Jr.’s 1997 essay in Foreign Affairs about the future of democracy.[1] Schlesinger’s article is incredibly prescient, for it anticipates many difficulties facing 21^st-century liberal democracies, at a time when the general mood was rather unapologetic democratic triumphalism. I don’t intend to comment on it here – I leave that for the near future – but there is one section in particular that I want to highlight. In it, Schlesinger discusses the implications of the “computer revolution” for the practice of democracy. Though he does not mention AI, Schlesinger does note that the potentialities offered by modern computers will likely lead to social “hyperinteractivity.” In this perspective, he contemplates the prospects of what James Madison called “pure democracy,” i.e., “a system in which citizens assemble and administer the government in person.”[2] The American historian does not buy the idea, however, emphasizing the political risks carried by high levels of interactivity:

“Interactivity encourages instant responses, discourages second thoughts, and offers outlets for demagoguery, egomania, insult, and hate. Listen to talk radio! In too interactive a polity, a ‘common passion,’ as Madison thought, could sweep through a people and lead to emotional and ill-judged actions. Remembering the explosion of popular indignation when President Truman fired General Douglas MacArthur, one is grateful that the electronic town hall was not running the country in 1951. The Internet has done little thus far to foster the reasoned exchanges that in Madison’s words ‘refine and enlarge the public views.’”

This paragraph should be read in light of the current discussions about the emergence of alternative micro-blogging platforms to Musk’s X (see, for example, this essay in the New York Times). For reasons that I will briefly explore below, it’s actually hard to challenge the leading platform even if many of its users are not satisfied with it. Many people are hoping that challengers like Bluesky will be able to provide a more peaceful, respectful, and intellectually productive social forum than X has become since Musk took charge. A necessary condition for that is that alternative platforms can reach what can be called “critical mass” while in the meantime filtering for undesired contributors, both humans and bots.

As an economist, I tend to intuitively think in terms of simple models and this issue makes no exception. So here are three alternative toy models that I consider relevant to capture part of the problem at stake.

Model #1 – An Assurance Game

Micro-blogging platforms display what economists call network externalities. That means that what a user receives as benefits from being on such a platform is partially a function of the total number of users. The reason why is relatively obvious: the more users, the more content to be shared and the more mutually beneficial interactions. A way to capture this aspect is to see the choice of a platform as an n-player “assurance game” where a user’s utility increases with the number of total users everything else equals while allowing for other relevant differences between platforms accounting for the fact that, when the total number of users is about the same, some platforms are objectively better than others. In the 2-player version, the assurance game is as follows

Everything else equals it’s better for everybody to be on platform A – this is the Pareto-efficient option in economists’ jargon. However, if the other player is using B, it’s better to use B, though it is socially inefficient. In the n-player setup, it can be shown that (provided that the numbers in the matrix have a cardinal meaning and are interpersonally comparable) a player is better with using platform A only if half or more players in the population are already using platform A. Switching from B to A is an instance of collective action problem because it requires a massive coordination mechanism to move to the Pareto-efficient outcome.[3] This is part of the problem that alternative platforms and their tentative users are facing.

Model #2 – A Sortition Model

The problem can be viewed slightly differently. We may assume that different users may prefer to interact with different persons depending on their characteristics, e.g., political ideas, hobbies, … We may then imagine that individuals will spontaneously sort themselves by going on a specific platform depending on their preferences. Something like that seems to be behind the attractivity of Bluesky. The platform seems to have been chosen by “liberals” (in the U.S. sense) as the place to go to enjoy the benefits of social interactivity without suffering the costs of interacting with far-right trolls. In some circumstances, this kind of “spontaneous sortition” can indeed happen. Thomas Schelling once explored this possibility with a simple “sortition model.”[4] Imagine that at a party, individuals are randomly distributed in two rooms. Suppose that there are two types of guests – men and women – and that each type prefers to be in the room the majority of the people are of the same type as them. In general, some guests may have stronger preferences for homogeneity than others – for instance, I may want to be in a room where less than 75% of the guests are different than me, while your threshold may be lower, say 60%. Depending on the distribution of the thresholds in the population and the initial proportion of types, a spontaneous sortition might occur, meaning that ultimately the types are completely separated – men in one room and women in the other. This is not the only possible outcome but in general, at least a partial sortition is observed.

Model #3 – A Prey-Predator Model

The previous model arguably simplifies too much. At least two of its assumptions are problematic. First, information may be incomplete, meaning that types are not perfectly known. We may think initially that you’re interacting with someone like you, only to discover later that this person actually has quite different opinions from yours. In some cases, that may happen because the other individual is actively hiding relevant information from you. Second, some users may actually enjoy interacting with persons of a different type. What I’ve in mind here is the case of “trolls” whose sole purpose is to mess with other users’ interactions for the sake of creating confusion, anger, and resentment, and ultimately to destroy the whole point of micro-blogging platforms. Incomplete information obviously makes it more likely to happen, for it becomes difficult for users to take ex-ante measures to avoid this happening.

This can be modeled in terms of “prey-predator” dynamics. Here, you have two types of agents, regular users and trolls. Regular users want to benefit from the network externalities by going on the platform with the most users but want to avoid trolls. If there are too many trolls, they will leave the platform for another one and just stop using this kind of service. Trolls also want to go where regular users are and are not bothered by other trolls. Once again, we don’t have to assume that thresholds are the same for every regular user (some may leave the platform with a lower of trolls than others). The resulting dynamic is easy to conceive. At first, regular users will arrive in number (if they have been able to solve the above collective action problem). Trolls (who don’t face any collective action problem) will progressively arrive on the platform. Once their number reaches a threshold, some regular users will start to leave, while in the meantime the number of trolls continues to grow. The number of regular users will decline faster and faster, until a point where trolls also start to leave the platform. Platforms obviously have an interest in limiting the number of trolls and what I’ve read about Bluesky suggests that regular users themselves are trying to actively monitor them. It remains to be seen if informal or formal mechanisms can be effective in maintaining the sortition over the long run.

Now, to return to Schlesinger’s quote, I think that there’s something fundamentally wrong with micro-blogging platforms, besides the problem with trolls, eventual misinformation, and other adverse effects that are largely documented everywhere nowadays. In a nutshell, the practice of micro-blogging considerably impoverishes the way individuals learn, interact, and think. I started to blog (in French) back in 2008, while I was still a PhD student. I’m proud to say that, in France, I was among the second wave of econ-bloggers – the first being actually a single and very influential blog. The emergence of Twitter has had a very negative effect on academic or semi-academic blogs. At first sight, platforms such as Twitter could have helped blogs to gain visibility. However, very quickly, micro-blogging started to be used not to spread information and refer to content located elsewhere but to produce “content.” I’ve always seen this as a disaster. The short format encourages randomly throwing assertions without making any effort to argue for them. To have a serious conversation, you need arguments and a public willing to listen. Micro-blogging created the illusion that this conversation could consist of short messages in which informed judgments fight mere opinions for attention. It favored superficial engagement with ideas and, therefore, tribalism (since, without argument, the only way to judge an idea is whether you agree with it). You may respond that this is a caricature and that some intellectuals, academics, or journalists have made a more reasoned use of the tool. So-called “tweetstorms” were regularly used to produce a real argument for instance. But I’ve never understood the point of tweetstorms. If you have something interesting to say, then make the effort to write it in a coherent text that is easily accessible and readable, rather than a list of bullet points. Micro-blogging just favors intellectual laziness.

I quit Twitter in 2000, in the midst of the pandemic, after 7 years on the platform. It was well before Musk took control. The context of the pandemic made it clear to me what I had already sensed a long time before. Even very smart persons that I respect were making silly assertions based on nothing else than their badly informed opinions. Micro-blogging is easy, it demands no effort and no time. As it requires no intellectual investment and its opportunity cost is essentially zero, it produces a lot of garbage that cannot be easily discriminated from what is valuable. This is a problem inherent to the very practice of micro-blogging and has nothing to do with the particular platform (and yes, this applies to Notes, except for the fact that most users also are writing a newsletter). The perfectionist in me would like to have this practice restricted. Fortunately, my liberal side is stronger and, anyway, I’m skeptical about the existence of collectively harmful practices. But I would like everybody to consider the following rule: if one has something important to say, one should make the effort to carefully formulate it with arguments that may require more than 300 signs to be developed.

---

[1] Arthur Schlesinger, “Has Democracy a Future?,” Foreign Affairs 76, no. 5 (1997): 2–12.

[2] Ibid., p. 6.

[3] Thomas Schelling discusses many examples of this type of collective action problem. My personal preferred one is how long it took for professional ice hockey players to agree on wearing helmets. In this case, the coordination mechanism was a regulation by the National Hockey League. See Thomas C. Schelling, Micromotives and Macrobehavior (Norton, 1978).

[4] Ibid.