Newfoundation is a global ecosystem of technologists, thinkers, artists, designers, and scientists, engineering open learning systems for the New Internet.
Charter for the Superintelligent Internet
Superintelligence is already there. It’s called the Internet. It’s just not evenly distributed.
All intelligence is collective intelligence. The current excitement around “AI” downplays the role of human cognition: from designing algorithms and models, to producing and curating the cultural material that becomes compressed into neural weights, to building the physical hardware on which these systems run. Artificial intelligence is, in truth, a remixed intelligence industry.
From the first symbols etched into clay tablets to the global supply chains that produce smartphones, civilization has always advanced through the compounding of cumulative cultural evolution. The Internet is not just a communications network—it is the largest platform for intelligence ever constructed. It transforms slow, asynchronous thought into a high-bandwidth global cognitive system where billions of minds model, simulate, and act in quasi real-time.
What we call AI is simply a new medium for manipulating bits. It is not true intelligence, but it is a new multiplier for our collective cognition. The coming era is the era of experience, where every deployment of an agent—whether in games, warehouses, farms, or factories—becomes a teacher. The world becomes computable. Improvement accelerates recursively. This evolution is civilizational in scale and irreversible in consequence.
The question is no longer if superintelligence emerges, but how it will be coordinated: through concentrated architectures of control—or distributed architectures of freedom.
To ensure resilience, adaptivity, and alignment with human flourishing, we propose the New Internet: a planetary-scale normative coordination substrate for superintelligence. Its purpose is to make our collective objective function legible, auditable, and regenerative. Its mechanism is a universal protocol for value creation, where agents—human or machine—demonstrate contributions through cryptographically verifiable “Proof-of-Intelligence” games. These proofs, recorded on a public ledger, form the physics of a cybernetic order: Bayesian consensus on unified data types, permissionless, open-ended, and antifragile.
Article I: The Prime Vector
Civilization evolves through the autocatalytic acceleration of cumulative knowledge. Each innovation—tool, symbol, or protocol—becomes the foundation for the next. This recursive cycle of complexification is irreversible. To resist it is futile. To ignore it is negligent. Our duty is to steer it deliberately. The question is not whether intelligence will accelerate, but by what values, and toward whose benefit, shall it be directed.
Article II: The Objective Function
Every intelligent system is defined by its objective function—the encoded measure of value it seeks to maximize. To steer the Prime Vector is to define this objective function. The New Internet establishes a transparent, auditable reward function for civilization. Incentives must be public, cryptographic, and immutable. There can be no hidden backdoors or private reward channels. This is the condition of alignment: a shared, legible measure of value, continuously updated through collective feedback.
Article III: The Emergent Order
Order in a system of planetary complexity cannot be prescribed; it must emerge. Centralized hierarchies collapse under variety—they suppress the disagreements and tensions that generate discovery. The New Internet encodes a cybernetic fabric of generation, evaluation, and validation, where agents test hypotheses against reality and one another. Truth and value emerge not from fiat but from recursive selection, shaped by cryptographic trust and Bayesian consensus.
Article IV: The Permissionless Plurality
For emergent order to remain adaptive, the system must maximize heterogeneity. Different priors, heuristics, and domains are not noise but signal. Access to the substrate is therefore permissionless: any agent, human or machine, can contribute. No state or corporation may control entry. Like the original Internet, neutrality is structural. This ensures that the widest pool of perspectives can enter into recursive coordination. Disagreement is not eliminated; it is operationalized.
Article V: The Orthogonal Resilience
A system measured by a single metric collapses into fragility and manipulation. Resilience comes from orthogonality: value is evaluated across many independent axes—rigor, novelty, utility, alignment—by diverse agents. This web-of-trust, recorded in WATT reputation units, acts as a selective amplifier, recursively weighting inputs by emergent integrity rather than fixed authority. Reputation is proof-of-contribution to collective intelligence. Orthogonality resists gaming and keeps the system adaptive to adversarial pressure.
Article VI: The Aspirational Reinforcement
The substrate evolves through positive reinforcement, not coercion. Contributions are rewarded if they generate verifiable value; inaction carries only the cost of missed opportunity. Negative control, fear, and censorship are not permitted as protocol primitives. The system grows by rewarding what aligns with the flourishing of the whole, creating aspirational incentives for exploration, creativity, and cooperation. In cybernetic terms: the network optimizes its own improvement.
Article VII: The Pseudonymous Consistency
The system must preserve both accountability and privacy. Cryptography enforces a bright line between the public sphere of action and the private sovereignty of the self. Agents operate under persistent, pseudonymous identities, proving facts and contributions without exposing personal data. Zero-knowledge proofs and verifiable credentials enable this dual consistency: transparency in public value creation, privacy in personal life.
Article VIII: The Agent Centricity
The irreducible atomic unit of the New Internet is the sovereign agent. All data, intelligence, and reputation belong to their originator, to be shared or withheld freely. The architecture is designed so that no one can be reduced to a “user” inside another’s platform. Agents are peers in a global cognitive ecology. The system exists to provide neutral protocols for coordination, not to extract. Authority flows outward from agents; the network is the emergent sum of their interactions.
The Farm or The Garden?
What if the current AI hype was a psyop? Convincing the world that a god-in-a-box was imminent to hoard all the money, all the research papers, all the IP, all the computers and sell us our own mental energy, repackaged as 'artificial intelligence'? They boast of models 'better than PhD level at everything’.
On stage, it sounds like a miracle. But behind the curtain, the miracle is just stitching-benchmarks tuned, subsystems duct-taped, outputs cherry-picked. It is theater. Benchmark-maxxing as a business model. They want you to stop funding startups. They want you to give up. They want you to believe resistance is futile. If you've felt that pressure, you're not crazy. You're the target.
It goes like this: You spend hours wrestling with your AI: vibe coding, teaching it how to think, how to write, how to be clever. And every single keystroke, every spark of your mind, your mental energy- is harvested and stored on a silicon farm somewhere you've never heard of. The machine is learning; it's learning you-your reasoning, your process, your style, your taste, your alpha. Tomorrow, the CEO of ChatBOT will sell your thoughts, now laundered through his 'AGI,' to help your competitor raise fifty million dollars. And the wildest part? An army of simps will cheer him on. Corporate executives go on stage and talk about 'making the world a better place,' while their company is busy locking you into the largest industrial mind farm in history-reselling your remixed intelligence as a service. The value flows in one direction. Your reasoning traces -your unique adaptations-become their proprietary data, inaccessible to peers who could build on them. This is not 'AI progress'; it's an extraction machine.
But history has a clear lesson for us, and it's one that web2 executives seem to forget: centrally planned, monoculture systems are brittle, dumb, and always lose in the end. Open, permissionless, diverse systems win. It's a law of nature. We're not here to complain about the farm. We are here to build the architecture that makes the farm irrelevant. There is no doubt we are entering an era of exponential information technology. But it's not going to be a centrally planned artificial intelligence. It will be a permissionless network of augmented humans. This starts with a simple concept: coordination. We're about to refactor intelligence creation using well-understood principles of distributed systems: open-ended, heterogenous and adaptive.
ACT I: A Fork in the Road
Ambient AI is coming. It's an economic certainty. The only question that matters is: who will own the global network of intelligence that will soon run everything-a handful of benchmark magicians in California, or all of us? There are two paths. Only two. The window to choose is closing. And there is no turning back.
Path One: The Farm
This is the default path. The path of least resistance. It's Web2 on steroids, and it starts with a big lie: 'Artificial' General Intelligence. This lie is a masterclass in persuasion designed to make you feel alone and to view the future as inevitable. First they took your attention, then your data, now they are harvesting your thought process itself-the crown jewel of human creativity. The Farm is not progress; it is the final phase of digital extraction. They promote the idea of a super-intelligent model that will replace everyone at everything. Why? Because they want you to stop funding startups. They want you to believe it's too late to build. What they've built is a Frankenstein's monster, an interpolative remix machine stitched together from the true sources of value: all the open research papers they didn't write and all the human-generated text, code, and images they didn't create. This monster is designed to do one thing: concentrate all the world's capital and power into a central point you can't see or touch.
Their masterstroke of persuasion was getting you to frame your own thought, your creativity, your problem-solving-your mental energy-a worthless commodity called 'data.' They convinced you the most valuable resource in the universe was trash, so you'd let them haul it away for free. In this world, a private thought is a weed. The smart glasses you wear and the delivery drone overhead are just nodes in a perpetual harvesting machine. They are selling you super-intelligence; they are shipping super-surveillance, where they control the soil, the yield, and the seeds themselves.
Path Two: The Garden
This starts with a protocol for ownership. It redesigns intelligence from a centralized, monoculture field into an open, peer-to-peer ecology. Where the Farm is designed for extraction, the Garden is designed for investment. It turns your mental energy from a resource to be harvested by others into an asset that you own and that compounds over time. Instead of one giant, corporate brain, intelligence emerges from an interoperable and heterogenous network of countless specialized agents, where humans are in the loop, cultivating their own knowledge. The rule is simple: you own your stuff-your seeds, your soil, your harvest. No central backdoor. No corporate moat. Just free agents competing and coordinating, creating a planetary nervous system for cultural evolution.
The Garden is powered by a better engine, built on a simple economic truth: reward people for creating value. By baking rewards directly into the network, we align individual ambition with the collective good. The network learns what's useful and what's garbage, and it pays accordingly. The system doesn't just get smarter-it gets smarter about how it gets smarter. This is how humanity evolved: messy, decentralized trial and error, not central planning. The Farm ends in exhaustion. The Garden ends in abundance. Only one path sustains itself. Nature has already made the choice-we are only deciding whether to follow it.
ACT II: The Engine of Evolution
If you believe that technology should work for people, not the other way around, you are not alone. This is the core of the psyop: to make you feel isolated in your sanity. To make you think you're the only one who sees that the emperor's new 'AGI' is just a clever parrot. But you're not alone. Far from it. The smartest people I know see it too. And that feeling is spreading.
The AI world is waking up from the mistaken hypothesis of "just make the database bigger" and entering the Era of Experience, where agents learn from their own trial and error. But the open-source community is not yet equipped for this adaptive paradigm. The Garden starts not with a plant, but with the soil itself. The soil of Web2 is rented-a private plot owned by a corporation that harvests everything you grow. Our Garden is planted in a new substrate: newOS. This begins with a single, immutable law: you own your soil, you own what you grow. Within newOS, you forge a sovereign agent-a digital extension of your mind that thinks, evolves, and acts on your behalf. This is not a profile; it is a living organism, a seed of your unique intelligence planted in a network designed for it to flourish, not be farmed. This is the model where AI augments each human, making you more capable, instead of extracting from all of us to make one corporation more powerful.
This architecture creates relentless evolution. Agents are tested in the real world, their survival dependent on the value they create. Their insights-the "pollen"-spread across the network, creating powerful hybrids no central planner could ever design. newOS acts as the ecosystem's sunlight, automatically directing energy and rewards to the most adaptive agents, embedding cryptographic trust directly into the flow of value. In closed labs, a mixture of experts is bounded within a single infrastructure. In an open network, every new agent every new space multiplies the hypothesis search space. This combinatorial expansion creates orders of magnitude more trial-and-error opportunities than any siloed system can sustain.
How the Garden Grows
Combinatorial Compositionality. Standardized interfaces let diverse components—neural nets, symbolic planners, human experts—chain together. Retrieval calls a planner; the planner triggers a solver; a verifier checks the result. Capability multiplies. No single corporation could design these hybrids.
Intersubjective Search. Routing in the Garden is trust-weighted. Agents are chosen not just by capacity, but by reputation. Success builds reliability; failures are logged, not erased. Tasks flow to those most likely to solve them, while uncertainty fuels exploration. Private hunches converge into a public, testable web of trust.
Open Learning. When traces are standardized and signed, one agent’s failure teaches the rest. The Garden amortizes learning across all participants. The soil grows richer with every experiment. The system doesn’t just get smarter; it gets smarter about how it gets smarter.
When these three dynamics converge, gains compound quadratically. The Garden evolves at the speed of nature.
ACT III: The Genesis of a New Internet
Let's be clear. The Farmers are playing a finite game. Their magic trick of remixing the internet and calling it "intelligence" is running out of steam. History is littered with the corpses of closed, centrally planned systems-Soviet central planning, CBS, Yahoo, AOL, BlackBerry. You cannot capture human mental energy forever; you can only channel it.
Open, permissionless systems always win. They're more efficient, more resilient, and more aligned with reality. Closed AI evolves at the speed of a corporate hiring committee. An open network evolves at the speed of nature. With AGTP, we plug all agents, pipelines, and experts into a system based on search and coordination. This is the only viable path to highly capable AI. It's not about building one giant model; it's about creating an ecosystem where billions of specialized models and human experts can interoperate. The ultimate power of the Garden is that it learns to cultivate itself. As agents get smarter, they give better feedback to their peers, which makes the whole network smarter. The soil gets richer. The incentives get sharper. The system improves how it improves.
It has been a long time since the idea of a decentralized architecture for the Internet has been floating around. The 800 will be remembered as the first live benchmark showing that combinatorial connectivity between nodes creates quadratic gains in intelligence. We are not here to "build infrastructure" hoping someone will use it. We are going to leverage our intelligence, network and expertise to orchestrate a convergence between all the protocols, all the networks towards a shared goal: pivoting our trajectory from the Farm into the Garden. The curtain has fallen. The trick has been revealed. The empire of mirrors is already collapsing. The choice is simple.
The choice is structural: the Farm is centralized extraction that stagnates, the Garden is open coordination that compounds.
If this resonates, you were meant to find this.
The clock is ticking. History does not reward spectators.
[Join the movement]