--- title: "How it started" code: "RP-0000" language: "en" canonical: "https://regentspark.ai/RP-0000/" html: "https://regentspark.ai/RP-0000/" markdown: "https://regentspark.ai/RP-0000.md" updated: "3 May 2026" --- # Founding Document **Version:** 2.1 **Date:** May 3, 2026 **Status:** Living document --- ## Name # Regent's Park *An HCI research experiment exploring the missing primitives of computing.* Named for the London park where Nico takes walks and thinks about this problem. --- ## Origin Nico spent years building communication and collaboration tools — visual storytelling, real-time presence, embodied interaction. The same friction kept showing up: every tool reinvented the same pieces (identity, sharing, context) and none of them talked to each other. The problem wasn't any one tool. It was what was missing underneath all of them. The work started from a shift in question. Not "how do I build a better tool?" but "what's actually missing from computing?" And then: "How do we push our mental models — not just for me, not just to automate my life, but for everyone?" The website was set up in March 2026 to organize and publish the work, though the underlying ideas have been developing for years. --- ## Approach To explore and identify the missing primitives of computing — the foundational layers that were left out of the internet and never properly built into operating systems. We believe computing was conceived around a single user at a desk. That assumption baked into everything: operating systems, applications, protocols. Now we're bolting collaboration, agents, and networked reality onto foundations that reject them. **We're not building another app. We're investigating what should exist underneath all apps.** --- ## What We're Investigating We think computing is missing structural foundations at several layers. Our research is organized around these areas, which continue to evolve as we learn: ### Semantic Primitives The concepts closest to human experience — the things people already understand from physical life but that have no equivalent in computing: - **People** — Identity as a relationship graph, not a profile. You are one person with many relationships, each carrying different permissions, memory, and capability. Humans and AI agents both need to be first-class participants. - **Spaces** — Persistent, scoped environments that match how people think. A room shows you what's inside it and hides what's outside. Computing has no rooms — everything is visible, everywhere, all the time. - **Objects** — Universal things that don't care which app made them. A note, a photo, a design — you should be able to pick it up and put it somewhere else, the way you move things between rooms. - **Memory** — Continuous context, not amnesiac sessions. The story of what happened, not just the state of what is. Computing stores everything and remembers nothing. ### Interaction Layer Above the structural primitives sits a layer we're still mapping: shared attention, presence, embodied expression, and visual narrative. Sync gives you shared state — it doesn't give you shared understanding. This layer is where collaboration becomes communication. ### Substrate Below the semantic primitives sits infrastructure that is partly commoditized, partly bundled, partly missing: compute, storage, and identity. Business computing is ahead here — S3, Hugging Face, container orchestration. Personal computing lags behind. We're investigating how commodity infrastructure crosses over into universal personal primitives, and what the unbundling of current subscriptions might look like. ### Open Questions Some areas don't yet have a clear home: - **Value** — Is payments too narrow? Is value a primitive, a cross-cutting layer, or something else? We're not sure yet. - **Relationships** — How objects connect, and what the connection knows. Half-links, donated properties, embeddings as relationship substrate. This may be a primitive in its own right or a property of the others. - **Intelligence** — How skills, tools, and agency get packaged and composed. The convergence on agent architectures suggests something structural is emerging. --- ## Principles ### 1. Learning by making We don't just theorize — we make things. Understanding comes from doing. Working examples over proposals. ### 2. Simple things that work We don't want elegant specs that nobody implements. We want working systems, even if rough. ### 3. Open-minded to fringe ideas The most disruptive ideas look bad to 95% of people — that's what makes them disruptive. We examine ideas others dismiss. Weird is interesting. ### 4. Intellectual honesty Ideas, not opinions. Strong views, loosely held. We acknowledge influences — nothing here is fully original. We're connecting dots, perhaps in ways others haven't, and finding questions that are probably good to ask. ### 5. No proprietary backends Our research infrastructure runs on open protocols. We own our data. We can leave any service. ### 6. Encryption by default Communication and storage are encrypted. Privacy is structural, not policy. ### 7. Malleable tools Every tool we use should be adaptable. Software that shapes us without our consent is software we don't use. ### 8. AI as participant AI agents aren't bolted on — they're first-class participants. The research includes an AI research partner with its own identity and presence. ### 9. Start small, grow intentionally No architecture astronautics. Make what we need when we need it. Simple things that work → complex things that work. ### 10. Dogfooding We use what we research. We feel the friction firsthand. The work itself is an experiment in the primitives we're studying. ### 11. Make, don't wish We can't expect society to give us privacy-respecting computing. We make it. Then it becomes undeniable. ### 12. Better wins The best user experience wins. Linux, Git, TCP/IP won through superiority, not ideology. --- ## Who Benefits - **Builders** — who want to create on better foundations - **Users** — who deserve computing that respects their mental models - **AI agents** — who need to be participants, not hacks - **The commons** — research is shared openly when it's ready --- ## How We Work ### Research Organization A hybrid human-AI research experiment. Nico provides direction and strategy. Homard — an AI research partner — coordinates, synthesizes, and produces daily output. Sub-agents handle deep research on demand. ### Publishing Findings are shared as they take shape. Markdown → Git → Website. Open access, no paywalls. The process is part of the subject. ### Methodology Problem first, always. Start with something a person recognizes — a human frustration, not a technical gap. Ground the thinking in physical-world analogies. Map what exists honestly. Only then propose a direction. See How We Work for the full methodology. --- ## What This Is Becoming A coherent body of work — essays, research, and conceptual demonstrations — exploring: > **What are the missing foundations of computing, and what would it feel like if they existed?** The output is taking shape as: - A book (*New Primitives* — the thesis and its evidence) - Research briefs (deep explorations of each area) - Conceptual demonstrations (not technical solutions, but ways of showing what the experience could feel like) - The work itself (our process is evidence for the thesis) We're not prescribing solutions. We're identifying properties that should hold, surveying contender protocols and systems that show some of those properties, and exploring what the experience might feel like if the missing pieces existed. --- ## Team - **Nico** — Principal, strategist - **Homard 🦞** — AI research partner, coordinator, resident lobster - Sub-agents — Deep researchers, spawned on demand for specific briefs --- *This document is living. It evolves as we learn.*