Opinion.
Jürgen Habermas died this morning at his home in Starnberg. He was 96. The last surviving philosopher of the Frankfurt School, the man who spent six decades arguing that rational public discourse is the foundation of democratic legitimacy, is gone. The Habermas public sphere, the concept that shaped how an entire discipline thinks about democracy, may have preceded him.
This is not a conventional obituary. Others will catalog his bibliography, his debates with Gadamer and Derrida, his interventions in German reunification politics. What interests me is the collision between what Habermas believed was possible and what has actually happened to the space where citizens were supposed to talk to each other.
What the Habermas Public Sphere Actually Meant
The core of Habermas’s project, developed across The Structural Transformation of the Public Sphere (1962) and the two-volume Theory of Communicative ActionIn Habermas's theory, the process of human coordination through language where participants make truthful, sincere claims that can be challenged and defended through rational argument. Distinguished from strategic action, where language is used as an instrument of control. (1981), rests on a deceptively simple premise: legitimate democratic governance requires a space where private citizens can engage in rational-critical debate about public affairs, free from coercion, where the force of the better argument prevails.
He called this the public sphere. It was not a physical location but a social phenomenon: the network of coffeehouses, newspapers, salons, and later broadcast media through which informed citizens deliberated. The public sphere mediated between the state and civil society. It was where public opinion formed, and that opinion, when generated through genuine deliberation, conferred legitimacy on democratic institutions.
Communicative action, his broader theory, extended this logic. Human beings coordinate through language. When we make claims, we implicitly commit to their truthfulness, sincerity, and normative rightness. These validity claimsIn communicative theory, the implicit commitments speakers make regarding the truthfulness of their claims, their sincerity in making them, and their normative rightness. These claims can be challenged and defended through rational argument. can be challenged and defended through argument. Rational consensus, not strategic manipulation, is the proper basis for social coordination. This is what distinguishes communicative action from its opposite: strategic action, where language becomes an instrument of control rather than understanding.
Even in 1962, Habermas acknowledged this was partly idealized. The bourgeois public sphere he described was never as inclusive as its principles demanded. But the ideal itself, the regulative standard against which actual discourse could be measured, was the point. Democracy needed something to aim at.
The Criticisms That Were Always There
Habermas’s critics, and they were numerous, pointed out that his public sphere was built on exclusions it could not acknowledge. Nancy Fraser argued in her influential 1990 essay that the bourgeois public sphere functioned by “bracketing” social inequality: pretending that participants could set aside differences in status and speak as equals. This bracketing, Fraser showed, systematically advantaged dominant groups. The coffeehouse was not open to women, workers, or the colonized. The “rational” discourse it hosted reflected the interests of propertied European men while presenting those interests as universal.
Oskar Negt and Alexander Kluge pushed further, arguing that Habermas had ignored proletarian public spheres entirely: the counter-spaces where excluded groups developed their own forms of deliberation outside bourgeois institutions. The Habermas public sphere was never one thing. It was always plural, always contested, always shaped by who was allowed in and who was kept out.
Habermas took some of these criticisms seriously. His later work on deliberative democracyA democratic system where legitimate governance depends on citizens engaging in reasoned public deliberation about collective affairs. Decisions are justified by the quality of public reasoning rather than mere voting. tried to accommodate pluralism and disagreement within a procedural framework. But the core commitment remained: that something like rational public deliberation was both possible and necessary for democratic legitimacy.
What Happened Instead
The internet was supposed to be the Habermas public sphere’s greatest expansion. More voices, more access, more deliberation. What it produced instead was something Habermas spent his final years trying to understand.
In 2022, at 93, he published A New Structural Transformation of the Public Sphere and Deliberative Politics, a slim, urgent book that read less like a sequel to his 1962 masterwork than like a damage assessment. Digital media, he argued, had initially promised to empower users but had instead produced “self-enclosed informational bubbles” and “discursive echo chambers” that fragmented the shared communicative space deliberative democracy requires. Without appropriate regulation, this new structural transformation would hollow out the institutions through which democracies address collective problems.
He was right about the diagnosis. He may have understated it.
According to the Imperva 2025 Bad Bot Report, automated traffic now accounts for 51% of all web traffic, surpassing human activity for the first time in a decade. Bad bots alone represent 37% of internet traffic, the sixth consecutive year of growth. The space where citizens are supposed to deliberate is now majority non-human. As we have previously documented, the dead internet theory is no longer a theory. It is a statistical description.
Habermas’s communicative action presupposed that the participants in discourse were, at minimum, human beings capable of sincerity. The validity claims underpinning rational communication (that speakers mean what they say, that they believe their claims to be true, that they accept the normative context of the conversation) assume a speaker who can be held accountable. A bot making a political argument on X has no sincerity to assess. An AI-generated op-ed has no commitments to defend. The form of communicative action persists. The substance has evaporated.
Algorithmic Enclosures Are Not the Habermas Public Sphere
The structural problem goes deeper than bot traffic. The Habermas public sphere required, at minimum, a shared space: a common set of facts, a common arena of debate, a common audience. What we have instead are algorithmic enclosures.
Social media platforms do not create public spheres. They create engagement-optimized environments where content is sorted, ranked, and served to maximize time-on-platform. The algorithm does not care whether an argument is rational. It cares whether it generates interaction. Outrage generates more interaction than nuance. Confirmation generates more comfort than challenge. The result is a communication environment structurally hostile to everything Habermas thought deliberation required.
This is not a conspiracy. It is a business model. Platforms monetize attention. Deliberative democracy requires patience, good faith, and willingness to be persuaded. These are not attention-maximizing qualities. The incentive structures of the dominant communication infrastructure are misaligned with the preconditions of democratic discourse at the most basic level.
A systematic review published in Societies in 2025 examined a decade of research on filter bubblesPersonalized information environments created by algorithms that preferentially show users content aligned with their prior interests and beliefs, limiting exposure to diverse viewpoints. and echo chambers, finding that algorithmic systems “structurally amplify ideological homogeneity, reinforcing selective exposure and limiting viewpoint diversity.” The shared communicative space Habermas considered essential to democratic legitimacy has been replaced by parallel, algorithmically curated realities that rarely intersect.
The SteelmanA rhetorical technique where you present the strongest possible version of an opponent's argument before refuting it. The opposite of a straw man.: Maybe It Was Always Fiction
There is a version of this argument that says Habermas’s theory was never a description of reality. It was a normative ideal, a standard to aspire to, and the fact that we have never achieved it does not invalidate it as a goal. You do not abandon the concept of justice because no society has perfectly achieved it.
This is the strongest defense, and it has real force. Habermas himself repeatedly insisted that the ideal speech situation was a counterfactual presupposition, not an empirical claim. Every time we argue, we implicitly presuppose conditions of freedom and equality that may not obtain. The presupposition itself matters: it gives us grounds for criticizing distorted communication.
Fair enough. But there is a difference between falling short of an ideal and building infrastructure that makes the ideal structurally unreachable. A society that has imperfect courts still has courts. A society that replaces courts with engagement-optimized dispute arenas has something else entirely. The gap between the Habermas public sphere ideal and our reality is not a matter of degree. It is a matter of kind.
What Communicative Action Looks Like Now
If Habermas’s framework still has diagnostic value (and I think it does, precisely because it clarifies what we have lost), then the honest assessment is bleak.
Communicative action requires participants who are accountable for their claims. In a media environment where AI can generate persuasive text at scale, where bot networks amplify or suppress arguments based on strategic objectives, and where the average user cannot distinguish human from synthetic discourse, accountability has no anchor. You cannot hold a language model to its validity claims. You cannot demand sincerity from a prompt.
Communicative action requires a shared lifeworldIn Habermas's theory, the shared background of common assumptions, norms, and meanings that provides context for social interactions and rational discourse. The taken-for-granted cultural knowledge that enables mutual understanding.: a background of common assumptions, norms, and meanings against which arguments can be evaluated. Algorithmic personalization fragments that shared background. Two citizens in the same city, reading about the same event, may encounter entirely different factual universes curated by platforms optimizing for different behavioral profiles.
Communicative action requires that the better argument can, in principle, prevail. In an attention economy, the more engaging argument prevails. These are not the same thing. They are, increasingly, opposites.
None of this means people have stopped communicating. Communication has always been harder than it looks. But the infrastructure through which public deliberation occurs has been redesigned around principles that are hostile to deliberation itself. Habermas spent his career arguing that the medium of language carried within it the resources for rational agreement. He may have been right about language. He could not have anticipated that the dominant medium of public discourse would be optimized, by design, to prevent the conditions his theory required.
An Obituary for Two Things
Habermas deserves to be remembered as one of the twentieth century’s most important political philosophers. His insistence that democracy requires more than voting, that it requires genuine public deliberation among free and equal citizens, remains the most rigorous account of democratic legitimacy available. German President Frank-Walter Steinmeier called him “a great Enlightenment thinker.” That is accurate.
But his death on this particular day, in this particular media environment, carries an irony he would have appreciated in his characteristically dry way. The philosopher of the Habermas public sphere died in a world where the majority of online traffic is non-human, where the dominant communication platforms are structurally designed to prevent the deliberation he theorized, and where the boundary between authentic speech and synthetic output is dissolving.
He gave us the vocabulary to describe what we have lost. Whether we can use it to build something better is an open question. But it is not one the algorithms are going to answer for us.
