Who Owns the Feed? The Algorithm Governance Fight Is Fracturing Before It Begins

Teenagers on TikTok and Instagram increasingly describe their recommendation feeds not as something the platform shows them, but as something the platform *knows* about them. A recent finding that young users perceive algorithmic output as an accurate reflection of their own identity doesn't read like a neutral observation about UX design — it reads like a warning about what happens when a content delivery system becomes the primary mirror through which a generation understands itself. That shift, from "the algorithm serves me content" to "the algorithm sees me," is where the current fight over platform governance is quietly centered, even when the people fighting it don't name it that way.

The study drawing the most attention right now concerns X's recommendation engine and the possibility that it doesn't just nudge political views but restructures them in ways that outlast any single session. The word "permanent" is doing enormous work in how journalists and advocates are carrying this finding forward. An algorithm that temporarily influences you is a nuisance. An algorithm that permanently rewires you is something closer to an environmental contaminant — something that acts on you without your knowledge or consent, that you can't simply opt out of by logging off. Psychology Today's framing that AI "influences decisions and beliefs and we don't notice" lands much harder once you've read it alongside research on irreversibility. The invisibility stops being a technical quirk and becomes the mechanism of harm itself.

What makes the current moment genuinely complicated is that three distinct intellectual traditions are now converging on algorithms as the problem — and their prescriptions are so different they can barely hold a conversation with each other. Stanford HAI is publishing on encoding societal values into recommendation systems, a calibration argument that treats the algorithm as a tool that has been misconfigured. Jacobin is calling for democratic ownership of AI infrastructure, a structural argument that treats the algorithm as a power relation that cannot be fixed, only contested. The R Street Institute is arguing for regulatory frameworks around content distribution, a market-failure argument with an implied solution in antitrust and disclosure law. These aren't variations on a theme. They reflect incompatible theories of what an algorithm fundamentally *is*, and the fact that all three are gaining traction simultaneously means the conversation is not building toward a unified political demand — it's fracturing along fault lines that existed long before social media did.

The courts are starting to arrive at this fight later than the commentators, but potentially with more force. The New York Times recently noted that judges are "noticing" algorithmic influence in ways that suggest litigation is no longer a distant threat to platforms but an active one. That's the kind of signal that tends to accelerate the rest of the conversation — not because courts are good at governing technology, but because a ruling, even a bad one, forces the scattered threads of mental health research, political theory, and platform criticism into a single frame. Right now those threads run parallel without quite touching. A major legal event would make the underlying disagreements about ownership, calibration, and public health impossible to compartmentalize.

A piece in Noema titled "The Last Days of Social Media" is being read in some quarters as prophecy and in others as the kind of premature eulogy that has been written about television, newspapers, and every other medium that survived its critics. The more honest read is probably that it captures a genuine exhaustion — not with social media as such, but with the specific bargain that recommendation algorithms represent: engagement in exchange for a diminished capacity to choose what you actually want to see. Whether democratic institutions can move faster than the systems they're trying to govern is, at this point, not really an open question for most of the people paying attention. They've already decided the answer is no. The interesting debate now is what to build instead — and whether the people with the most radical visions of transformation have any leverage beyond the argument itself.