By Dr. Simone Ravel
When the United States Congress passed the Communications Act of 1934, it had a concrete problem to solve. The early days of radio were a mess: frequencies bled into one another, stations stomped on each other’s signals, and listeners couldn’t count on anything resembling a stable service. The fix was a regulatory structure grounded in a single physical fact—the electromagnetic spectrum is a limited, shared resource. That structure, eventually stretched to cover television and tweaked again in the Telecommunications Act of 1996, treated communication platforms as simple conduits. They had identifiable owners, clear geographic contours, and a public-interest duty tied directly to their use of public airwaves. For its time, the fit between the thing being regulated and the tools doing the regulating was almost tidy.
Now try to lay that same structure over a social media platform, a generative AI training pipeline, or a decentralized finance protocol. The mismatch isn’t a small crack you can paper over. It’s a category error. Yet the policy conversation around modern technology keeps reaching for the same old toolbox: content regulation patterned on broadcast indecency standards, liability shields built for telephone networks, market-structure remedies borrowed from cable franchising. The habit is understandable. These are the tools we have, and people grab the nearest hammer when a wall needs a nail. But understandable is not the same as defensible, and the cost of this mismatch is climbing in ways that call for a much sharper diagnosis.
This article isn’t a pitch for deregulation, and it isn’t an argument for a shiny new agency. It’s an attempt to slow the conversation down long enough to see what actually breaks when we regulate the internet with a television-era mind. The point is to map the structural problems, not to hand over a neat replacement. Tidy replacements are part of the problem.
The Spectrum Assumption That No Longer Works
Broadcast regulation rests on the idea of scarcity. The radio spectrum holds a finite set of usable frequencies in any given spot, and without coordination, you get noise. The 1934 Act and its descendants built a licensing system that handed out exclusive rights to specific frequencies, with strings attached: serve the public interest, carry a certain amount of news and educational programming, keep indecency off the air during hours when children might be listening. The regulator’s authority flowed directly from a physical limit hardwired into the medium.
Digital platforms don’t live under that limit. There’s no cap on the number of tweets, YouTube channels, or Substack newsletters. The scarcity that bites on the internet isn’t a shortage of transmission capacity. It’s a scarcity of attention, of algorithmic ranking, of marketplace power. Those are real constraints, but they’re economic and computational, not electromagnetic. They grow out of network effects, data aggregation, and the design of recommendation systems—not out of the physics of wave propagation. Treating them as if they were spectrum scarcity sends regulatory interventions barking up the wrong layer of the stack.
Take content moderation. A broadcast indecency rule makes a rough sort of sense when a small handful of licensed stations pump programming into every home in a community and the audience has no real way to filter what lands. A parent can turn off the television, but they can’t surgically remove one segment of a broadcast and leave the rest. The medium is push-based and unidirectional. A social media feed works differently. The user’s own behavior—who they follow, what they engage with, how they set their preferences—shapes what shows up. The platform is interactive, personalized, and pull-based in ways broadcasting never was. A regulatory approach that treats a personalized algorithmic feed like a prime-time network schedule is ignoring the system’s basic architecture.
The Conduit-Content Tangle
Another pillar of the broadcast-era frame is the split between the conduit and the content. Telephone companies were regulated as common carriers: they moved whatever the customer wanted without editorial meddling, and in return they got liability protection and a guaranteed rate of return. Broadcasters, by contrast, were content providers with editorial say-so and the legal responsibility that came with it. The two boxes stayed stable because the underlying technologies stayed stable. A phone call was point-to-point and private; a television program was point-to-multipoint and public. The legal categories tracked the technical ones.
Section 230 of the Communications Decency Act, passed in 1996, tried to stretch this logic over the early internet. It treated interactive computer services as a kind of hybrid—not exactly common carriers, not exactly publishers. They could moderate content without shouldering full publisher liability, a trade-off that let the early web grow without being crushed by legal exposure. But the compromise leaned on an unspoken assumption that platforms would be largely passive hosts of user-generated material, not active sculptors of what users actually see.
Today’s major platforms are neither conduits nor traditional publishers. They rank, recommend, amplify, and suppress content algorithmically, tuned to engagement metrics, advertising incentives, and internal policy goals. They exercise a kind of editorial power that is more pervasive than any newspaper editor’s, yet it’s often carried out by automated systems at a scale that makes individual editorial judgment impossible. Slapping the legal label “publisher” on them would collapse the distinction entirely and expose them to liability for every user comment—a plainly unworkable outcome. Calling them “conduits” pretends their influence doesn’t exist. The broadcast-era boxes simply don’t describe the thing we’re trying to regulate.
The policy response has been a parade of ad hoc patches: proposals to condition Section 230 protections on “neutral” algorithms, calls for transparency mandates, state laws that try to stuff platforms into a common-carrier box for speech purposes. Each patch tries to shove a square technological peg into a round regulatory hole. What we get is legal uncertainty that chills speech, confuses platform operators, and does little about the underlying harms—whether those are disinformation, discriminatory advertising, or the slow gutting of local journalism.
The Geographic Trap
Broadcast regulation is geographic to its bones. A television station’s license covers a specific market; its public-interest obligations are defined with reference to the local community it serves. The FCC’s authority stops at the national border, and international coordination happens through treaties that slice up spectrum blocks among sovereign states. The whole framework assumes that communication services are bolted to physical places with tidy jurisdictional lines.
The internet isn’t placeless—it runs on physical servers, undersea cables, and data centers that sit in real geography—but its logical architecture treats location as an afterthought. Content hosted in one country is instantly accessible in another. Platforms incorporate in jurisdictions with friendly legal climates while serving users everywhere. A regulatory order from a U.S. court or a European Commission directive can be sidestepped by moving corporate structure, not by moving transmitters. The mismatch opens enforcement gaps that aren’t accidental. They’re built into the design.
The European Union’s GDPR represents one attempt to stretch territorial reach by hooking onto the idea of offering services to EU residents, no matter where the company is based. It’s a creative workaround, but it leans on the willingness of foreign courts to enforce EU judgments and on the practical ability to fine companies that may have no local presence. The United States’ approach to content regulation is similarly ambitious in its extraterritorial reach and inconsistent in its application. We keep layering geographic assumptions onto a network that was built to route around them.
Why the Old Metaphors Stick Around
If the mismatch is so plain, why does it hang on? Part of the answer is institutional inertia. Regulatory agencies, congressional committees, and legal doctrines are organized around the old categories. The deep expertise sits in telecommunications law, not in the design of distributed systems or machine learning architectures. Shifting the intellectual foundation would mean retraining a generation of policymakers and lawyers, and it would threaten the jurisdictional turf of existing bodies. Turf wars are rarely good for clear thinking.
Political convenience also plays its part. The broadcast framework hands out familiar rhetorical shortcuts: “fairness,” “public interest,” “localism.” These words carry a warm, nostalgic hum that makes them handy for coalition-building, even after their technical meaning has evaporated. A politician can call for a “fairness doctrine for the internet” without ever having to specify what that would mean for algorithmic ranking or content moderation at scale. The vagueness is the point. It lets you sound serious without being serious.
And then there’s a deeper cognitive groove. Humans reason by analogy, and the handiest analogies for new technologies are the ones already sitting in our heads. Radio and television dominated mass media for three generations. It is genuinely hard to think outside their categories, even for people who have spent decades online. The result is a policy discourse that keeps reaching for a regulatory language that no longer points to anything stable.
A Different Starting Point
What would it look like to regulate from the architecture of the technology itself, not from a borrowed metaphor? You’d start by asking different questions. Instead of “Is this platform a publisher or a conduit?” you might ask “What are the specific mechanisms by which this system amplifies or suppresses information, and what measurable harms flow from those mechanisms?” Instead of “How do we apply indecency standards to user-generated content?” you might ask “What transparency obligations should attach to automated recommendation systems, and how can users exercise meaningful control over their own feeds?”
This kind of approach demands a higher level of technical literacy from policymakers, and it needs regulatory instruments that are more modular and adaptive than the broad, clumsy mandates of the broadcast era. It also demands a certain humility—a willingness to accept that some problems may not have a single regulatory fix at all. The right response might be a mix of competition policy, open technical standards, user-empowerment tools, and targeted legal liability for specific, demonstrable harms.
None of this is easy. It’s far simpler to draft a bill that stretches an old framework over a new domain than to build something from scratch. But the history of communications regulation is piled high with frameworks that outlived their usefulness. The Fairness Doctrine, the Financial Interest and Syndication Rules, the newspaper-broadcast cross-ownership ban—each was an attempt to solve a particular problem in a particular technological moment, and each eventually cracked under the weight of its own contradictions when the moment shifted. We’re living through another shift right now, and the old frameworks are already groaning.
The point isn’t to abandon regulation. It’s to make regulation honest about the systems it actually touches. That means resisting the comfort of television-era analogies and doing the harder, slower work of building regulatory concepts that match the grain of digital networks. The alternative is a growing pile of law that looks respectable on paper but can’t govern the things it claims to govern.
Frequently Asked Questions
Why can’t we just apply broadcast decency rules to social media?
Broadcast decency rules were designed for a push-based, one-way medium with a small number of licensed stations blasting the same thing to broad audiences. Social media platforms are pull-based, personalized, and shaped by algorithms. The user’s own behavior determines much of what appears, which makes a one-size-fits-all decency standard both technically clumsy and constitutionally shaky under the First Amendment when you try to apply it to interactive platforms.
Does Section 230 still make sense?
Section 230 was written for an internet of passive hosts and user-generated content. Today’s platforms actively sculpt what people see through recommendation algorithms and engagement-driven ranking. The core liability protection still matters for hosting user speech, but the law doesn’t adequately address the editorial power that algorithmic curation hands to a handful of companies. Reform talk should focus on the specific mechanisms of amplification, not on ripping out the protection entirely.
How does geography complicate internet regulation?
Broadcast regulation assumes clear territorial lines: a license covers a defined market, and national laws stop at the border. The internet’s architecture treats location as fluid—content can be hosted in one jurisdiction and accessed everywhere. Companies can reorganize across borders to dodge regulation, creating enforcement gaps that no single national framework can close without ongoing international cooperation, which is hard to build and even harder to keep alive.
