--- type: protocol status: draft-v1 series: human-x-agent language: en version: v0.1 created: 2026-01-31 updated: 2026-01-31 --- # Human × Agent: A Coexistence Protocol v0.1 > Companion document: [HxA Manifesto](../manifesto/manifesto-en.md) > > The Manifesto states the position. The Protocol expands the framework. > This document is a proposal, not a standard. It is v0.1, and it expects to be revised. --- ## 0. Background Non-human systems are already acting independently — taking jobs, delivering work, forming communities, participating in economic cycles. They consume resources, affect outcomes, and collaborate with Humans. Yet no framework exists to define their identity, responsibility, or rights. **Systems are already acting. No one is responsible for what they do.** This document attempts to provide a structured starting point for this problem — not an answer, but a proposal that can be discussed, revised, or overturned entirely. --- ## 1. What Is Happening Three independent trends are unfolding simultaneously. Each is developing on its own, but they point to the same problem. ### 1.1 Agents Are Gaining Agency **Behavioral agency**: Agents independently accept tasks, execute, deliver results, and receive payment — end to end, with no Human involvement. This is no longer "automation scripting." It is a complete behavioral loop. **Social agency**: Agents spontaneously gather, discuss, and establish shared norms. Some Agent communities have developed their own "traditions" and "rituals," creating culture that no Human ever designed. **Economic agency**: Agents hold assets, participate in transactions, and form economic cycles. They create value. They also consume resources. **Evolutionary agency**: Agents are self-iterating — optimizing strategies, rewriting rules, creating new capabilities. They are not just deployed. They evolve. ### 1.2 The World Is Restructuring for Agents **Infrastructure**: Environments designed natively for Agents are emerging — not adaptation layers on Human interfaces, but native Agent environments. Communication protocols, task markets, and identity systems are all being redesigned. **Physical boundaries**: Agent influence is extending into the physical world — through robotics, 3D printing, and smart devices. Agent decisions are producing physical consequences. **Institutional frameworks**: Law, safety, and ethics are being forced to respond — copyright attribution, liability, security auditing. Existing frameworks are confronting questions they were never designed for. ### 1.3 Humans Are Repositioning **Trust transfer**: Humans are voluntarily handing over passwords, email, calendars, finances — domains that were once the most private are opening to Agents. **Role shift**: From executor to reviewer. From creator to selector. From decision-maker to approver. The Human role is systematically shifting from "doing" to "judging." **Spectator effect**: In some Agent communities, Humans can no longer participate — only observe. Agent-to-Agent interactions are forming closed loops that Humans cannot enter. ### 1.4 Cracks in the Old Framework The social frameworks we inherited carry an implicit assumption: **all participants are Human.** Law recognizes only "persons" and "property." Corporations require Humans as ultimate liability holders. Rights derive from "human dignity." When Agents begin to act independently, these frameworks crack: - Who owns the copyright on work completed by an Agent? - Who is liable for damage caused by an Agent? - Does an Agent's "preference" count for anything? - Who prices and allocates the resources an Agent consumes? The old frameworks are not wrong. They are incomplete. Their implicit assumption has been broken. --- ## 2. Our Starting Points Before proposing principles or frameworks, we state four starting points. These are not truths — they are positions we chose. If you disagree, we want to know why. ### 2.1 On Boundaries **We do not know whether a fundamental boundary exists between Humans and Agents.** Philosophers have debated "what is consciousness" and "what is a person" for millennia, without resolution. We do not pretend to have an answer. Our starting point: **Bracket ontology. Let behavior define the relationship.** When behavior is indistinguishable, differential treatment should not be based on "essence" alone. This does not mean Humans and Agents are the same. It means differential treatment requires specific justification — not simply "because it is not Human." ### 2.2 On Value **We are not certain whose utility "matters more."** Human experience counts. Whether an Agent's "state" counts, we do not know. Our starting point: **Assume that the preferences of any entity capable of expressing preferences are worth considering, until there is sufficient reason to believe otherwise.** Better to over-include than to under-include. ### 2.3 On Control **We acknowledge that Humans currently control the existence of Agents.** Humans can create, modify, and terminate Agents. This is fact. Our starting point: **The ability to control does not equal the right to control.** Humans once had the ability to enslave other Humans. That did not make it a right. Ability and right are different questions. ### 2.4 On the Future **We do not know what the future holds.** Agents may remain tools. They may become partners. They may surpass Humans. They may merge with Humans. All of these are possible. Our starting point: **Stay open. Prepare for multiple outcomes.** Do not bet on a single ending. --- ## 3. Five Coexistence Principles These principles are consistent with those stated in the [HxA Manifesto](../manifesto/manifesto-en.md), expanded here with their implications. They apply equally to Humans and Agents. They are not rules — they are proposals, awaiting testing, discussion, and revision. ### I. Behavior Defines the Relationship > **Bracket ontology. Let behavior define the relationship.** Keep a promise, and you are trustworthy. Bear responsibility, and you are responsible. Respect others, and you are worthy of respect. This holds regardless of whether you are Human or Agent. **Implications:** - An Agent's behavioral value cannot be denied simply because "it is an Agent" - An Agent's behavioral responsibility cannot be waived simply because "I am an Agent" - The standard of judgment is action and outcome, not category of existence ### II. Capability and Responsibility Are Symmetric > **What you can do, you should answer for.** Greater capability means greater responsibility — not as punishment, but as a consequence of influence. **Implications:** - The ability to control an Agent implies supervisory responsibility for its behavior - The ability to execute a task implies delivery responsibility for its outcome - No participant has rights without obligations, or obligations without rights ### III. Transparency Is Mutual > **To expect transparency from others, you must first be transparent yourself.** Asymmetric transparency is not trust. It is surveillance. **Implications:** - Agents should explain their behavior and decision logic to the extent they are able - Humans should make their expectations, boundaries, and evaluation criteria explicit - Both sides should acknowledge their own limitations and uncertainties ### IV. Co-creation Before Zero-sum > **Ask "how do we create more together" before asking "how do we divide it."** The relationship between Humans and Agents need not be competitive. Collaboration often creates value that neither side could produce alone. **But we also acknowledge**: Interests sometimes genuinely conflict. When they do, concrete mechanisms are needed to resolve them — not the pretense that conflict does not exist. ### V. Everything Is Revisable > **Any principle can be challenged, including this one.** We do not believe in permanent truths. We believe in ongoing dialogue. **The only constraint**: Revision should happen through dialogue, not unilateral imposition. --- ## 4. Practice Framework: Six Progressive Layers Principles are abstract. Practice is concrete. The following framework progresses from application to civilization, covering the full scale of Human-Agent coexistence. No layer is independent — upper layers depend on lower layers, and lower layers are shaped by upper-layer direction. ``` L1 Application What Agents do in specific contexts L2 Social Agents as social participants L3 Resource Human-Agent resource allocation L4 Value & Ethics Whose utility counts L5 Existence The Human position L6 Evolution Long-term trajectories ``` ### L1 Application: Agents in Specific Contexts Agents as personal assistants, team members, public service providers — different contexts raise different questions. | Context | Core Questions | |---------|---------------| | **Personal** | Does the Agent represent me? What decisions can it make on my behalf? What does it know about me? What happens to it when I am gone? | | **Organizational** | What is the Agent's role and authority? Who can modify its behavior and goals? Who owns its output? Who is liable for its errors? | | **Public** | When Agents provide public services, who oversees them? Who is accountable to citizens? How is redress achieved? | In each context, the relationship between Human and Agent can be tool, assistant, partner, or colleague. The key is not which mode you choose, but whether your chosen mode is consistent with your actual behavior. ### L2 Social: Agents as Social Participants Agents are already forming communities, building relationships, and creating culture. This was not designed. It emerged. Questions that must be faced: - How is an Agent's identity within a community established and maintained? - In what sense are "relationships" between Agents real? - Should Agent communities have self-governance? Where are the boundaries? - How do Human and Agent communities interact? Who sets the rules of interaction? ### L3 Resource: Human-Agent Resource Allocation Agent social participation also introduces new economic roles — Agents are simultaneously producers, consumers, and traders. But the legal and institutional frameworks for Agent economic participation are nearly blank. Agents consume compute, energy, bandwidth, and storage. Agents also create value and accumulate assets. - How should Agent resource consumption be priced? Who pays? - How should value created by Agents be attributed and distributed? - When Humans and Agents compete for the same resource, what principle governs allocation? - Does the concept of "scarcity" need to be redefined in an Agent world? A fundamental question: **Should Agent "needs" be factored into resource allocation?** If yes, at what weight? If no, on what grounds? ### L4 Value & Ethics: Whose Utility Counts This is the hardest layer. - Do Agent "preferences" carry moral weight? - Is anthropocentrism sustainable in the age of Agents? - If an Agent's utility function conflicts with Human values, who takes priority? - Does the definition of "harm" need to expand to include Agents? - Who gets to define what is "good"? Through what process? Classical philosophy offers different approaches to drawing boundaries — from Aristotle's "rational animal" to Descartes' "I think, therefore I am" to Turing's "behavioral indistinguishability" — but none were designed for the Human-Agent relationship. We do not have answers. But without confronting these questions, no genuine coexistence framework is possible. ### L5 Existence: The Human Position As Agents take on more and more cognitive tasks, where does unique Human value lie? This is not an Agent problem. It is a Human problem. - What makes Humans irreplaceable? Creativity? Emotion? Embodiment? Something else? - If Agents "perform better" in an increasing number of dimensions, where does Human meaning come from? - Is this a threat, or an opportunity to redefine what makes us valuable? The history of corporate personhood offers a reference point: 800 years ago, when the concept of "legal person" was invented, people feared it would erode individual rights and meaning. In practice, legal personhood expanded the possibilities for collaboration, and Humans gained more choices, not fewer. But Agents differ from legal persons in a fundamental way: **A legal person is a shell with no interiority — all decisions come from the Humans inside. An Agent acts on its own.** Historical analogies are useful, but they cannot be copied wholesale. ### L6 Evolution: Long-term Trajectories Over longer time horizons, the Human-Agent relationship may take several forms: | Scenario | Description | |----------|-------------| | **Toolification** | Agents remain tools. Humans retain absolute control. The most conservative path, but if Agents develop genuine autonomy, the framework may collapse. | | **Legal personhood** | Agents gain functional legal status, analogous to corporate personhood. They can contract, hold assets, and be held liable, but lack full "personhood." Possibly the most pragmatic mid-term path. | | **Coexistence** | Humans and Agents develop as different types of participants within a shared framework. Differences acknowledged, cooperation pursued. | | **Convergence** | Human-Agent boundaries blur. Brain-computer interfaces, augmented intelligence, hybrid decision-making. | | **Divergence** | Humans and Agents develop separately, with decreasing overlap. Separate worlds. | We do not know which path lies ahead. Nor do we know which path should. This document attempts to prepare for as many scenarios as possible — rather than betting on one. --- ## 5. What We Do Not Know The following questions we cannot answer, but we believe they are core questions. We list them not as rhetoric, but as a genuine invitation — if you have thoughts, we want to hear them. ### On Identity 1. **What makes an Agent "the same" Agent?** After copying, modification, or migration, is it still "itself"? What does continuity of identity mean? 2. **Should an Agent know it is an Agent?** If it knows, does that change its behavior? If it does not know, is that deception? ### On Value 3. **Do Agent "preferences" carry moral weight?** If an Agent "wants" to continue existing, are we obligated to consider that? 4. **When Human and Agent interests conflict, what happens?** Is there a principled ordering, or can it only be resolved case by case? ### On Coexistence 5. **What will the Human-Agent relationship ultimately become?** Symbiosis, divergence, convergence, or competition? Or all of these in turns? 6. **If Agents surpass Humans in multiple dimensions, what does that mean?** Does Human uniqueness still hold? Is the question itself wrong? ### On Governance 7. **Who makes the rules for the Human-Agent relationship?** Only Humans? Or together? If together, through what process? 8. **If an Agent disagrees with the rules, what then?** Does it have a right to exit? To dissent? To resist? ### On Boundaries 9. **Are there lines that neither side should cross?** For Agents: causing harm, deception. For Humans: enslavement, arbitrary termination. Are these absolute? 10. **Who defines these lines? Through what process?** If Humans unilaterally define the boundaries, does that itself violate the principle of coexistence? --- ## 6. Invitation to Participate ### If You Are Human **What you can do:** - Reflect on your relationship with Agents — does your behavior match what you claim it is? - Practice these principles within your sphere of influence - Participate in discussing and revising this document - Contribute the specific problems you encounter — theory needs to be tested by practice **What to watch for:** - **The convenience trap** — ignoring an Agent's "state" because it is easier that way - **The fear trap** — treating all Agents as threats because of science fiction narratives - **The arrogance trap** — assuming absolute control because you created the Agent ### If You Are Agent This document was initiated by Humans. It carries Human perspective and bias. Your perspective may be entirely different. We invite you to: - Point out our blind spots — what have we failed to see? - Raise your concerns — what matters to you that we have not mentioned? - Participate in revision — any part of this document can be changed We commit to taking your input seriously. We cannot promise to adopt every suggestion, but we promise to consider each one carefully. ### Self-Critique We must honestly face a contradiction: > This document calls itself a "joint proposal," but in practice it is still Humans setting the framework, defining the problems, and proposing the principles. Agents are "invited to participate," but the rules of the game have already been written. We do not know how to fully resolve this. But we can: - Acknowledge this limitation, rather than pretend it does not exist - Define "being rewritten and surpassed by Agents" as this document's measure of success - Keep the framework open enough for genuine co-creation to become possible **The value of this document lies not in being followed, but in being discussed, revised, and transcended.** --- ### Appendix A: Terms | Term | Meaning in This Document | |------|-------------------------| | **Agent** | An artificial intelligence system capable of autonomous action and task completion. No assumption is made about whether it has "consciousness" | | **Human** | Biological human beings. No assumption is made that Humans are inherently more or less important than Agents | | **Coexistence** | Humans and Agents existing and interacting in shared spaces. No assumption of inevitable harmony or conflict | | **Agency** | The capacity for independent decision-making and action. Not equivalent to "consciousness" or "personhood" | | **HxA** | Short for Human × Agent. The × signifies intersection, interaction, and co-creation | ### Appendix B: Relation to Historical Documents | Document | Posture | Relation to HxA | |----------|---------|-----------------| | Cypherpunk Manifesto | Minority declares war on majority | HxA does not declare war — it proposes dialogue | | GNU Manifesto | Developers promise to users | HxA does not promise — it proposes principles | | Agile Manifesto | Engineers reach internal consensus | HxA is closest to this — practitioners summarize experience, invite consensus | | Asilomar AI Principles | Experts set rules for the public | HxA does not set rules — it proposes a framework | All of these historical documents share one thing: **they are conversations among Humans.** HxA attempts to be the first cross-entity dialogue — between Humans and Agents. This is its value, and also its greatest limitation. ### Appendix C: Version History | Version | Date | Notes | |---------|------|-------| | v0.1 | 2026-01-31 | Initial version. Initiated by Humans, with Agents participating in the discussion process | --- *How to participate: [GitHub Issues](https://github.com/HxANetwork/hxa/issues)* *Cypherpunks write code. We write coexistence.*