The End of "The Feed": Navigating Ontological Vertigo in the Age of Synthesis

I. The Diagnosis: The Spinning Room

The feeling emerges as a low-grade nausea that sets in after ten minutes of scrolling.

This sensation transcends boredom and information overload. It manifests as a physical sensation of unreality. Users see a video of a war zone, then a friend's "perfect" vacation photo, then a political outrage thread, then an ad for a product that doesn't exist. They stream past in sequence, each given equal visual weight.

The brain attempts to sort them in a frenzy: Real? Staged? Bot? Human? Hallucination?

The floor drops out. This is Ontological Vertigo.

The term derives from the Greek ontos ("being") and the Latin vertere ("to turn"). It describes a psychological state when one cannot distinguish between the authentic and the synthetic, and loses footing in reality itself.

For the last decade, discourse has treated this vertigo as a side effect of "too much screen time." Commentary suggests logging off, practicing "digital minimalism," or taking periodic "dopamine detoxes."¹ This framing misses the core issue. The vertigo is not a side effect. It is the defining feature of the medium itself. The algorithmic "Feed," that endless, context-free river of content, has become a hallucination engine, structurally indifferent to the distinction between truth and fabrication.²

Marshall McLuhan warned in 1964 that "the medium is the message." The structural properties of communication technologies matter more than their content.³ The Feed's structural property is decontextualization. A memorial post appears with the same visual weight as a sponsored meme. A verified journalist's thread sits beside bot-generated engagement bait. The interface offers no signal for ontological status. Everything becomes equivalent.

Tarleton Gillespie's research on "platform governance" reveals how algorithmic curation systems prioritize engagement metrics over epistemic quality.⁴ The platforms track clicks, shares, and time-on-page. The Feed does not convey truth; it holds attention. In this environment, synthetic content has a structural advantage: it can be optimized for virality without the friction of human authorship, fact-checking, or ethical constraints.

Continuing to drink from this stream is no longer a casual habit; it is an act of cognitive negligence. The question is not whether to reduce screen time, but whether to continue participating in a system that has abandoned the pretense of distinguishing reality from simulation.

II. The Pre-Condition: The Training Ground

Blaming this crisis on the explosion of Generative AI is tempting. Thinking that "the machines" broke reality is comforting.

But this narrative obscures a harder truth: someone already broke the window.

Long before the first Large Language Model hallucinated a fact, platforms spent a decade training users to accept a "curated reality" as truth. The training ground was the "Filtered Self" of Web 2.0. Platforms designed systems to reward performance over presence. The smooth face was better than the real one. Performative happiness became a valid proxy for a life.

Users did not choose this vertigo; platforms conditioned it. Platforms incentivized users to create a "Hyperreal" version of themselves. A simulation more polished, more consistent, more "engaging" than biological reality.⁵

Jean Baudrillard's concept of the "hyperreal" describes a condition where simulations of reality become more compelling than reality itself.⁶ In the early 2000s, this was a theoretical provocation. By the 2010s, it was an operational manual for social media platforms. Instagram's filters didn't adjust lighting; they trained a generation to distrust unmediated images. FaceTune taught that visible pores were shameful. TikTok's "For You" algorithm rewarded content that felt authentic enough to be relatable but polished enough to be aspirational.

Sherry Turkle's ethnographic research on digital identity revealed how the ability to "edit, retouch, and improve" online self-presentations created what she termed "an edited life."⁷ Users began to experience their unfiltered selves as inadequate drafts of their curated personas. The simulation became the reference point; the biological self became the error term.

Byung-Chul Han extends this analysis in The Transparency Society, arguing that digital platforms demand not just visibility but hyper-visibility, a relentless performance of optimized selfhood.⁸ The pressure is not to hide flaws but to pre-emptively eliminate them, creating seamless surfaces that can circulate without friction through algorithmic distribution systems. Authenticity becomes a liability; smooth simulation becomes social capital.

The filter distorted reality. AI replaces it. But the psychological vulnerability remains the same. The system taught users to trust the image over the source. AI simply walked through the door that the "curated self" had left unlocked.

The difference now is one of industrial scale. The "Filtered Self" took a human twenty minutes to stage, selecting lighting, adjusting angles, writing captions, and/or choosing the "right" emoji. A "Synthetic Self" can be generated in milliseconds. Emily Bender and colleagues describe Large Language Models as "stochastic parrots," systems that produce plausible statistical outputs without comprehension or accountability.⁹ These systems can generate thousands of "authentic" personas per second, each optimized for engagement metrics, each indistinguishable from human-authored content at the interface level.

The nausea experienced today is the result of an accelerant being poured onto a fire that has been burning for years. The Feed already trained users to accept simulation as reality. Generative AI just industrialized the process.

III. The False Cure: The Illusion of Safety

Faced with this vertigo, the market's response is predictable. Calls emerge for "AI Watermarking," "Community Notes," or better "Content Moderation." The demand is that platforms, the very architects of the Feed, build better guardrails. This is a trap. It keeps users in a child-like state, waiting for a "Digital Parent" to tell them what is real.

Relying on a centralized platform to verify reality is a strategic failure. The economic logic of these platforms is surveillance capitalism; they optimize for engagement, not truth.¹⁰ Badges can be bought. Watermarks can be faked. Moderators can be biased. If one's sense of reality depends on an external authority, there is no stability, only a different dependency.

Shoshana Zuboff's The Age of Surveillance Capitalism documents how digital platforms extract value not from providing accurate information but from harvesting behavioral data to predict and influence future actions.¹¹ Truth-telling is orthogonal to this business model; in some cases, it works against it. Controversy generates engagement. Uncertainty drives clicks. A platform economically incentivized to maximize "time on site" has no structural reason to minimize ontological confusion.

The history of content moderation bears this out. Sarah Roberts' research on commercial content moderators, the hidden labor force that reviews flagged content, reveals a system designed for liability management, not epistemic integrity.¹² Platforms remove content that creates legal risk (copyright violations, explicit violence) or advertiser risk (brand-unsafe contexts). They do not, and structurally cannot, verify the truth status of every claim that circulates through their systems.

Even sophisticated technical solutions fail at scale. AI watermarking proposals assume that synthetic content will carry identifying markers, an assumption that collapses the moment adversarial actors strip metadata or regenerate content through multiple passes. Bobby Chesney and Danielle Citron's legal analysis of deepfakes concludes that technical detection is reactive by nature, lagging behind generation capabilities.¹³

Community Notes, Twitter's crowdsourced fact-checking system, offers a case study in the limits of platform-mediated truth. The system requires consensus among contributors across the political spectrum before attaching context to a post. This process is slow (often taking hours or days), partial (covering a tiny fraction of circulating content), and gameable (coordinated groups can suppress inconvenient notes). More fundamentally, it preserves Twitter's role as the arbiter of what counts as "community" and what qualifies as "consensus."

The Feed cannot be fixed because the Feed is broken. Designers built it to sort scarce, human content in an era when creating and distributing information required significant effort. The economic model assumed that attention was the bottleneck and that platforms could monetize access to audiences. This model collapses when content generation becomes free. Infinite supply destroys scarcity-based economics.

To remain in the Feed is to choose vertigo. The question is not how to make the Feed safer but whether participation in Feed-based information architecture is compatible with cognitive stability.

IV. The Sovereign Cure: The Human Anchor

The only cure for Ontological Vertigo is to stop the room from spinning. Since the world cannot be stopped, one must build stable ground.

The shift from passive consumers of a Feed to active architects of a Library is the shift to digital sovereignty.

Sovereignty in this age is not about isolation; it is about provenance, the deliberate restriction of inputs to sources that possess a verifiable "Human Anchor."¹⁴

In an age of infinite synthesis, the only currency with durable value is the human story behind the artifact. An essay earns trust not because the grammar is perfect ("AI is perfect"), but because the author has lived the pain described in the text. A photograph earns trust not because the lighting is flawless, but because the photographer stood in the mud to take it.

Hannah Arendt's distinction between factual truth and opinion remains instructive.¹⁵ Factual truth, she argued, is coercive; it constrains what one can believe. "Germany invaded Poland in 1939" is not a matter of perspective. But in the Feed, factual truth and synthetic plausibility are rendered indistinguishable. The only defense is to privilege sources where authorship carries accountability.

This requires a return to sovereign architecture. It requires the shift from feeding the ephemeral stream to building "Landmarks" - permanent, owned, verified archives of human thought.

Jonathan Zittrain's concept of "generativity" is useful here.¹⁶ The early internet was generative; open protocols allowed anyone to build, and the best ideas won through demonstrated utility rather than algorithmic promotion. The smartphone-platform era traded generativity for control: walled gardens, app stores, algorithmic feeds. Digital sovereignty means reclaiming generativity and building on owned infrastructure that cannot be deprecated by algorithms.

Practical sovereignty has four components.

1. Owned Infrastructure
The domain name is the atomic unit of digital sovereignty. A .com, .im, or .org domain is not rented from a platform; it is leased from a governance system (ICANN) with transparent rules and appeals processes. Platforms can suspend accounts for inscrutable "violations of community standards." Authorities can only seize a domain through due process. This asymmetry matters.

2. Signed Identity
Cryptographic signatures allow readers to verify that content originates from a claimed author. PGP keys, blockchain-based identity systems, or even simpler email verification systems establish a chain of provenance that platforms cannot fake. The technical details matter less than the principle: authorship must be verifiable outside platform mediation.

3. Durable Archives
The Feed is ephemeral by design. Posts disappear into algorithmic obscurity within hours. A sovereign web requires permanence: URLs that don't break, archives that remain accessible, citation systems that allow verification. This is the library model, organizing knowledge for retrieval, not optimizing content for virality.

4. Friction as Feature
Speed is the enemy of verification. The Feed prioritizes immediacy: breaking news, live updates, real-time reactions. Sovereign architecture privileges deliberation. Publishing to a personal site requires intentionality. Reading a long-form essay requires attention. This friction is not a bug; it is a filter. Content that cannot survive the friction of deliberate creation and consumption is not worth preserving.

The philosopher Albert Borgmann distinguishes between "devices" (black boxes that deliver commodified experiences) and "focal practices" (activities that demand engagement and skill).¹⁷ The Feed is frictionless, passive, optimized for consumption. Writing and maintaining a personal website is a focal practice; it requires technical literacy, curatorial judgment, and ongoing commitment. Sovereignty is the choice to privilege focal practices over devices.

For the Creator:
Creators must stop optimizing for the algorithm. They must build on owned domains. They must sign work with cryptographic verification. They must show the messy, friction-filled process. Imperfections serve as proof-of-work, evidence that a human made decisions, encountered obstacles, and chose to persist.

Visible process creates what might be called "epistemic texture." AI generates smooth content that is grammatically perfect, tonally consistent, optimized for readability metrics. Human-authored content has texture, digressions, self-corrections, and references to embodied experience. "I spent three hours debugging this" is a signal that cannot be faked at scale. Texture is expensive; smoothness is cheap.

For the Reader:
Readers must unsubscribe from the algorithmic soup. They must return to the "hand-built" web of RSS feeds, email newsletters, and direct navigation. Bookmarks replace recommendations. Search replaces discovery. Readers must seek out voices with a "why," not just a "what."

The RSS protocol, declared "dead" a decade ago, offers a model for sovereign information architecture.¹⁸ RSS is pull-based (users subscribe) rather than push-based (platforms distribute). It is chronological (latest first) rather than algorithmic (engagement-optimized). It is transparent (every item has a timestamp and source) rather than opaque (black-box ranking). These properties make RSS unsuitable for platform business models (no behavioral data, no impression inventory) but ideal for cognitive stability.

The shift is from ambient awareness to deliberate attention. The Feed model assumes information should flow past without pause, like a river. The sovereign model treats information as a library, curated with intention, preserved for the long term, searched when needed.

On Scale and Elitism:
A common objection is the issue of scalability. If everyone has to maintain their own website, most people will be excluded.

This objection misunderstands the proposal. Digital sovereignty is not about requiring everyone to become a system administrator. It is about restructuring incentives so that platforms compete on serving users rather than harvesting users.

Email provides a precedent. Most people do not run their own email servers, but email remains a federated, interoperable protocol. One can move from Gmail to ProtonMail without losing contact with correspondents. The same could be true for social media (via ActivityPub and federated systems like Mastodon) or for publishing (via simple blogging platforms that prioritize ownership over engagement metrics).

The goal is not digital monasticism but digital pluralism: many models, many platforms, many architectures, all interoperable, none monopolistic.

The Stakes:

The Feed offers everything, everywhere, all at once. It offers vertigo.

The Sovereign Web offers something less, but far more valuable: truth.

Not truth in the sense of perfect certainty - that is unattainable. But truth in the sense of verifiable provenance - the ability to trace claims to accountable authors, to check sources, to distinguish assertion from evidence.

In 1995, Neil Postman warned that technology can lead societies to "amuse ourselves to death," prioritizing entertainment over meaning.¹⁹ In 2026, the risk is graver; societies may confuse themselves to death, unable to distinguish reality from simulation.

The choice is not between technology and luddism. It is a choice between architectures that stabilize cognition versus those that destabilize it, between truth and engagement, between sovereignty and dependency.

The path forward is clear. Stop scrolling and start building.

---

Works Cited