An open blueprint for digital power separation: three spaces, one internet, governed so that no single power, commercial or political, can capture the whole. Version 0.5 · July 2026. Offered to be scrutinised, contested, and improved.

Preface: What This Document Is

This is a blueprint, offered in the spirit of one: a serious foundation meant to be built on, tested, and improved. It sets out a way to reorganise our shared information environment so that entertainment, knowledge, and public life each get a home suited to their purpose, and so that none of the three can swallow the others.

The ambition is global. The problem belongs to no single country. It is the condition of the world's shared information environment, and the repair proposed here would have to be taken up widely to work in full. One small country is named late in these pages, but only as a plausible place to try it first.

Two commitments run underneath everything that follows. The first is that no one owns the truth. Any design that depends on a single authority deciding what is true will sooner or later be captured or corrupted, and will deserve to be. The second is that this document is written to make its case as strongly as the case deserves, and also to be argued with. Where the proposal raises hard questions, they are met directly, late in these pages, rather than hidden. A structure that cannot answer its own hardest questions is not worth building. One that can is worth taking seriously.

I. The Diagnosis

We built a single, undivided internet and asked it to do everything. The same feed that carries jokes, advertising, and celebrity gossip also carries journalism, scientific findings, election information, and the public record of what our governments do. All of it is ranked by systems optimised, in the end, for one thing: holding our attention long enough to sell it. And nearly all of it flows through platforms owned and steered by a handful of people who answer to no public anywhere.

None of this is a moral failing of the people who use these systems, nor even, for the most part, of the people who build them. It is a structural outcome. When a single environment must serve incompatible purposes under a single incentive, the purposes that pay (outrage, novelty, conflict) crowd out the ones that do not. Careful explanation loses to confident assertion. A correction never travels as far as the falsehood it corrects. Over time the shared factual ground that public life depends on begins to thin, not because anyone set out to destroy it, but because nothing in the architecture was ever asked to protect it.

The symptoms are familiar by now, and worldwide: rising polarisation, collapsing trust in institutions, publics sorted into incompatible realities. The deeper problem sits beneath the symptoms. We are running the knowledge and the governance of whole societies on infrastructure designed to sell advertising, and then acting surprised that knowledge and governance are suffering.

The ground is shifting beneath even this. Synthetic text, images, and video can now be produced at almost no cost, and a growing share of what people read arrives from no source at all: a machine-generated summary with no byline, no editor, and no one answerable for whether it is true. As the open web fills with content that no human stands behind, the question this document answers grows more urgent, not less. Where is the place where a named, accountable institution still stands behind every word?

This is not a problem to be patched. It is a problem of structure, and problems of structure have structural solutions.

II. The Principle: A Separation of Powers for the Internet

The solution is not new. It is one of the oldest ideas in political thought, applied to a domain its authors could not have imagined.

In 1748 Montesquieu argued that liberty survives only when power is divided: when those who make the laws, those who enforce them, and those who judge them are held in separate hands, each able to check the others. Note what he did not propose. Not three separate states, but three separate powers inside one state, with different footing, different accountability, and different incentives, so that no single one could dominate the whole. The separation is constitutional, not geographic.

Two centuries later the legal scholar Lawrence Lessig added the observation that completes the picture for our age: in the digital world, code is law. What a system makes easy, what it makes hard, what it rewards and what it suppresses governs behaviour as surely as any statute. We do not need to legislate every human choice online if we are willing to think carefully about the structures inside which those choices are made. Change the architecture and behaviour changes with it.

Put the two together and the proposal of this document follows directly. The trouble with our information ecosystem is neither the technology nor the people. It is that we have handed a single architecture, with a single incentive, responsibility for functions that need fundamentally different ones. The remedy is to separate those functions: not into different internets behind walls, but into different spaces inside one internet, each with its own governance, its own funding, and its own rules of the road.

One clarification matters more than any other in this document, so here it is plainly. We are not proposing to split the internet into three walled-off territories that traffic cannot cross. The pipes stay whole. What gets separated is institutional: the home a kind of communication lives in, the incentive that shapes it, and the body answerable for it. A courtroom and a casino can sit on the same street and run on the same electricity. What makes them different is not the wiring but the rules that govern each. So it is here.

III. The Three Spaces

The proposal names three spaces, each with its own purpose, governance, and incentive: the Playground, the Prism, and the Civic Hall.

The Playground: where chaos belongs

The first space is the one we already have: the commercial, social, entertainment internet of memes, video, advertising, and spectacle. It is not the villain of this story, and nothing here seeks to diminish or replace it. Play, commerce, and distraction are legitimate human needs. A space optimised to satisfy them is no more shameful than a fairground.

The Playground needs only two things, neither of which is a wall. The first is legibility. Synthetic and AI-generated content should be labelled as such at the level of the file itself, using content-provenance standards that already exist; the Content Credentials effort, backed by a coalition of technology and media organisations, is exactly this kind of cryptographic “where did this come from” tag. A reader should always be able to tell entertainment from journalism, and a generated image from a photographed one. Not because either is forbidden, but because the difference should never be hidden. A label, to be clear, is a claim, and the absence of one is not a crime but the absence of a claim. Most files in the world carry no credential and never will, and metadata can be stripped. The point is not to police every file. It is to let honest sources cryptographically sign what they publish, so that verification is always possible at the source. Unlabelled content remains as legal as an unsigned letter, and carries an unsigned letter's weight. The second is transparency about manipulation: limits on opaque political microtargeting, and disclosure of how content is ranked and paid for, broadly along the lines of the platform-transparency rules now emerging in several jurisdictions. The Playground stays wild. It just stops pretending to be anything other than what it is.

The Prism: where knowledge has a home

The second space is the heart of the proposal, and the part worked out in most detail. It is a knowledge and journalism space: a place where reporting, analysis, and science-grounded information live under standards suited to knowledge rather than to engagement.

Its foundation is a credential. Outlets that bind themselves to a defined standard of accountability earn a verifiable mark. That mark could surface as a dedicated namespace such as .prism, so that an accredited outlet's address (bbc.prism, for instance) announces what it has committed to. Or it could surface as a cryptographically signed emblem carried on any existing address, the way independent quality and safety marks already work on physical goods. The form is an implementation choice. The substance is the public register behind it. None of this is a novel or fragile mechanism. Restricted, vetted top-level domains already exist: .gov and .edu are limited to eligible institutions, and .bank can be held only by verified banks bound to ongoing security obligations, administered by an independent registry rather than a government. A credential that means “this source has met a standard, and can lose the mark if it stops meeting it” is established practice.

What the credential certifies is the most important design decision in the Prism: accountability, not correctness. To earn and keep the mark, an outlet must meet conditions of process, not of conclusion:

• a named, responsible editor who can be held answerable;
• a published corrections policy, and a record of actually issuing corrections;
• disclosed ownership, funding, and editorial methodology;
• an enforceable right of reply for those it reports on;
• adherence to a recognised professional code of conduct.

It does not, and must never, certify that an outlet's reporting is true. It vouches only that the outlet has bound itself to be answerable for what it publishes. That keeps faith with the principle that no one owns the truth, and it is close to what independent national press councils in many countries already do: certify conduct, through a non-state body rather than a government ministry. The conditions are practices, not pedigree. An independent journalist who publishes under their own name, corrects their errors, and discloses their funding can earn the mark on the same terms as a global broadcaster. These are, not incidentally, the same habits of disclosure, correction, and reproducibility that the open-science movement has spent two decades making standard in research.

To keep the credential itself from becoming a point of control, no single authority issues it. Accreditation is granted by several independent bodies, the way web browsers trust a set of certificate authorities rather than one, so that no single gatekeeper can excommunicate a legitimate outlet and the bodies check one another. Every decision to admit or revoke can be appealed, and every such decision is written to a public, tamper-evident log, so that no outlet can be de-listed on the quiet and the act erased from memory. The web already uses this design to keep its own trust system honest: a public, append-only record, created after the authorities that vouch for secure connections were caught misbehaving. Paired with article-level cryptographic signing, it lets a reader verify that a piece really came, unaltered, from the accredited outlet it claims.

On top of this accredited namespace sits the Prism itself: a free, public front end whose only jobs are discovery, provenance, and the deliberate widening of perspective. It gathers what is published across the namespace into a feed the reader controls. Sliders, not a hidden algorithm, govern what appears: by country and region; by subject, whether political, environmental, or economic; and, crucially, by exposure to opposing views, so the reader can actively resist the echo chamber rather than get sorted into one without noticing. Discussion happens beneath each story, among participants verified as human but free to remain pseudonymous. The conversation is accountable without becoming a register of names. When a reader wants the full piece, the Prism sends them out to the source, to the BBC's or CNN's own site, rather than holding them captive in a walled garden. That choice is deliberate. Aggregators that kept readers for themselves provoked open revolt from publishers, while a referrer that preserves each outlet's direct relationship with its audience gives publishers a reason to take part. The Prism is a reference reader, not a gatekeeper. Because it sits on an open namespace, rival portals can compete to build better readers on the same foundation, and none of them owns the commons. Anyone doubting there is appetite for this need only look at the perspective-comparison services already thriving commercially, Ground News and AllSides among them: private previews of what the Prism makes public infrastructure. The reference reader itself is funded the way libraries, public broadcasters, and Wikipedia are (publicly, philanthropically, or by its readers' voluntary support), at arm's length and under an independent charter, so that no advertiser and no ministry owns its front page.

The mechanism that makes such a reader good rather than merely neutral already exists, in research and in practice: bridging-based ranking, which rewards material that earns agreement across normally opposed groups instead of material that inflames one side against another. It is the opposite of engagement ranking, and the natural ranking principle for a knowledge space. One caution belongs beside it. Consensus is not truth, and a ranking that rewarded only what everyone can agree on would bury the uncomfortable scoop and the genuinely new idea. Bridging is therefore a default lens, not a law of the feed. The sliders keep discovery in the reader's hands, and one of them is there to surface what is contested.

Outlets are paid through a universal front end to the subscriptions readers already hold. The Prism recognises a genuine subscription and unlocks the full piece inside its reader: single sign-on for news, rather than reselling the outlet, with anonymous subsidy so that quality information is not a privilege of those who can pay. The alternative people reach for first, a single bundle of all journalism, is a graveyard; the outlets worth paying for have thriving direct-subscription businesses and no reason to dissolve into a bundle that commoditises them. The routing matters more each year. As AI answer-engines summarise journalism without sending anyone to it, the Prism does the opposite: it exists to route readers, and revenue, to the people who did the reporting. But the deepest point about money is not the model. It is a principle that belongs at the centre of the whole proposal: whatever the Prism pays outlets for is what those outlets will produce. If revenue flows by clicks or time-on-page, the engagement-tabloid incentive has been rebuilt inside the knowledge space, and the Prism becomes the thing it was made to replace. The revenue mechanism is therefore not a detail to settle later but a core ethical choice, arguably the core choice, and it must reward depth, accuracy, and the issuing of corrections, never attention.

On whether this needs a blockchain: a public ledger is the right tool for one part of it, making the registry itself tamper-proof, so that every accreditation and revocation is permanently and visibly recorded. But a ledger can store a decision; it cannot make one. Putting a captured body's bad judgment on-chain only makes it permanent. Who accredits, and by what standard, is an institutional question in technical disguise, answered by the governance above (independent, plural, accountable, appealable), not by cryptography.

The Civic Hall: where public power is answerable

The third space is the digital home of public life: a transparent record of what those in power actually do, and a structured space in which citizens can see it, question it, and deliberate.

Its founding insight corrects something that happened almost unnoticed. Private platforms have become civic infrastructure, the places where public conversation and political life now occur, while keeping the power to erase any person at will, with no due process and no appeal. To be removed from them has become a kind of partial civic erasure. The Civic Hall corrects this by tying civic-digital access to citizenship, as a right that cannot be revoked: something a person holds by being born into or naturalised in a polity, and pays for through taxation, the way one holds the right to a passport or to vote.

This is not a proposal to build a “government Facebook” from nothing. The civic transactions (proving identity, filing taxes, receiving official mail) are in most capable states already handled, and Estonia has built perhaps the most complete version of citizen-owned digital identity in the world. What is missing, almost everywhere, is what sits above the transactions: a transparent, navigable public record of political action, covering votes, promises and their status, lobby meetings, and expenditures, together with a space for structured, good-faith deliberation about it. The closest working example anywhere is Taiwan's experiments in digital democracy, which have used structured deliberation platforms to reach consensus on genuinely contested questions. They are small, and that is exactly why this part of the proposal is the most worth trying.

The defining feature of the Civic Hall is not an algorithm but the deliberate absence of one. The public record runs chronologically and transparently, like a court docket, not as a ranked feed built to maximise engagement, because a feed optimised for attention is the last thing civic information should ever be. Its safety comes from how it is built. The citizen owns their data; every instance of the state accessing it is logged and auditable by the citizen; and authentication, proving you are entitled to access, is kept strictly separate from any behavioural tracking. The model to follow is Estonia's. The model to avoid is any identity system built without those guarantees, where the same convenience becomes a leash. Access is a right. It must never become a record kept against you.

IV. The Four Guarantees

Part of what makes this idea a repair rather than a new instrument of control is what it refuses to do. These four guarantees are not disclaimers added for reassurance. They are load-bearing, and a version of the idea that breaks any of them is a different and worse idea. They are also the promise itself: the reason the structure can be trusted with the things it is meant to protect.

• It routes and blocks nothing. No one is assigned to a space or prevented from reaching any part of the open web. The spaces are additive, never exclusive: you gain homes for knowledge and public life without losing anything you have now.
• It gates nothing behind a test. There is no compulsory course, no certification of the reader, no examination to pass before taking part. Access is never a privilege to be granted, because gates on access are exactly what can be captured.
• It bans no one's speech. It does not bar politicians, or anyone, from the commercial and social spaces. It asks only the narrow and ancient thing democracies already ask: that the official acts of public power happen in the open, on the record. Not that private speech be forbidden anywhere.
• It appoints no authority on truth. No body in this design certifies that any claim is correct. The structure helps people find the truth for themselves. It never hands anyone the power to declare it, because that power is the single point of capture the whole design exists to prevent.

A reader who finds anywhere in these pages a mechanism that breaks one of these four, even quietly, should treat it as a flaw to be fixed. They are the shape of the promise, and the proposal stands or falls on keeping them.

V. Answering the Hard Questions

A serious proposal earns serious questions, and the strongest objections to this one are worth meeting head-on. Not as reasons for doubt, but because answering them is how an idea shows it has been thought through.

The keystone is the governance of accreditation. Everything in the Prism rests on the bodies that decide who earns the mark, and the safeguard against their capture is structural. No single body decides; the standard is process rather than conclusion; every decision is public and appealable; and the record is tamper-evident. That is no absolute guarantee (nothing is), but it is what a would-be captor would have to defeat in full public view.

Because the credential certifies accountability and not correctness, an outlet some readers find odious could earn it by genuinely meeting the conditions: a named editor, real corrections, disclosed funding, a right of reply. This is deliberate, and it is a strength. A mark that admitted only the outlets one faction approved of would be worth nothing. Certifying conduct rather than agreement is what makes it legitimate rather than partisan.

The mirror objection will come just as quickly: that accreditation is a club for legacy media, built to lock independent voices out. The design answers it three times over. The conditions are practices anyone can adopt, so a one-person newsroom that stands behind its name, corrects its errors, and discloses its funding qualifies on the same terms as a century-old broadcaster. Several accreditors operate side by side, so no single establishment taste controls the door. And because the open web is untouched, no one needs the mark to publish. Only to carry it.

Running this costs money, and pretending otherwise would be its own dishonesty. Accreditors need staff and infrastructure; outlets carry compliance work. The design keeps both deliberately light. The criteria are practices serious outlets largely follow already. Fees scale with size, waived entirely at the small end so the mark never becomes a moat, and registries fund themselves the way the .bank registry and national press councils do: through their members and arm's-length public or philanthropic support. None of it waits on a world agreement. The namespace is opt-in infrastructure, adopted outlet by outlet and country by country, useful at every scale along the way.

A universal, un-revocable civic identity is powerful, and power can be abused. That is why the Civic Hall's safeguards are its core rather than its trim. Citizen-owned data, auditable access, and the separation of authentication from tracking make misuse difficult and visible. They cannot make a worse future government impossible. They can make sure it leaves fingerprints, and that is why they belong in the deepest, hardest-to-change layer of the design.

One rule runs through every space: whatever is paid for, ranked, or measured will be optimised toward. The Prism's revenue model is the sharpest case, but the principle is general. The incentives must be designed as carefully as the architecture, because in the end they are the architecture.

Then there is the cold-start problem. A knowledge space with no readers gives outlets no reason to seek the mark, and a mark no outlets hold gives readers no reason to come. The answer is the oldest one: provision and quality, not compulsion. Build something good enough, and open enough at the point of discovery, that people choose it, the way a great public library or broadcaster draws people without forcing anyone through its doors. Working with human nature is part of that build. Reading verified sources can be made visible, even a quiet mark of identity and status. But the moment the Prism imports the Playground's streaks and dopamine loops to hold attention, it has rebuilt the machine it exists to replace. The discipline is reward without addiction.

And beneath all of these sits the hardest question of all: what if people simply do not care? A head of state can already reach millions from the Playground, and for many, that voice will outweigh any accredited source on any day, in any conflict. The honest answer is that no architecture can compel belief, and this one does not try. What it changes is what is checkable. The official record exists whether or not a given citizen consults it; the provenance of a claim can be verified by anyone who wants to; and the excuse that the truth was unavailable disappears. A shared floor does not need unanimity to be worth building. The courts, the researchers, the journalists, and every citizen who does care need somewhere solid to stand. Reliability, given time, is how trust has always been rebuilt. There is no faster way, and no architecture that skips it.

It is worth adding that in much of the democratic world the coercive version of this idea is not merely unwise but already unlawful. Strong protections for free expression forbid compulsion and the restraint of speech. The law, in other words, already insists on the open, non-coercive version proposed here, which is the only version worth wanting.

VI. Where to Begin

An idea this large still has to be tried somewhere first, on a scale where it can be built honestly and its effects actually measured. That argues for starting in a single small country with capable public institutions and civic infrastructure already in place, somewhere the experiment is tractable rather than continental. Several democracies fit that description. Norway is one plausible candidate: small enough to measure, with widespread digital identity and a strong public-service-media tradition to build on. This is no claim that any such country is free of the pull of money and politics (none is), only that a smaller, well-equipped system is an easier place to test the idea fairly before asking the world to adopt it. The purpose of a first trial is not to prove the idea is anyone's. It is to let results, rather than this manifesto, make the case to everyone else.

VII. The Intellectual Lineage

This proposal is not invented from nothing, and it is stronger for saying so. Its political foundation is Montesquieu's separation of powers, and behind him the older worry, running back to Plato, about what becomes of a society when truth loses its footing. Its reading of the public sphere (the shared space in which a society reasons about itself) and of that sphere's fragmentation is the territory of Jürgen Habermas and of Cass Sunstein's work on echo chambers. Its core technical insight, that architecture governs behaviour, is Lawrence Lessig's. Its sense of the present crisis is shared by writers such as Jonathan Haidt.

More than that, its central proposal has serious company already building in the same direction. The idea of designing healthier digital public spaces on the model of parks and libraries rather than malls is the work of Eli Pariser, Talia Stroud, and the New_ Public project. The closest working instance of structured, depolarising civic deliberation is Taiwan's. The most complete model of citizen-owned civic infrastructure is Estonia's. And the mechanisms this blueprint relies on (content provenance, transparency logs, bridging-based ranking) are not hypothetical. They exist and are in use.

The contribution here is not the invention of any single piece. It is the integration, the argument that these scattered efforts are facets of one design: a separation of powers for the information ecosystem. And it is the identification of where the open ground lies: the civic-deliberation space that almost no one has yet built.

VIII. An Invitation

This is an open blueprint, and the word open is meant literally. It is published not as a finished plan to be admired or adopted, but as a serious foundation to be argued with, corrected, and improved by anyone willing to take it seriously. Its hardest questions are put in the open on purpose, so they can be worked on rather than discovered too late. Its strongest claims are offered to be tested. Responses, objections, and improvements are invited at gonzodocs.com/3, where this document lives and will be revised in the open.

The aim beneath all the detail is a single structural one. We are trying to build an information ecosystem in which no single power, commercial, political, or technological, human or otherwise, can capture the whole. An ecosystem where entertainment can be wild without governing us, where knowledge has a home built for knowledge, and where public power is conducted in the open and answerable to the people it serves. Not a system that decides for us what is true, but a structure within which we can find out for ourselves.

That is an old ambition, the same one Montesquieu had, and Plato before him: that the arrangement of power, carefully designed, can protect the things a society most depends on. This document tries to carry that ambition into the one domain where it has not yet been built, and to do so everywhere it is needed, which is everywhere. Whether it succeeds is now a question for everyone willing to help answer it.

Audun Amundsen · Norway, July 2026