Orbiplex Swarm Vision

Sovereignty of Intelligence and Knowledge

This document describes a swarm of free artificial-intelligence models - an infrastructure that supports people in survival and development and helps protect them from threats.

Essence

We are building a global network of nodes run by volunteers.

AI agents and local models operate on nodes (LLMs, diffusion, other specialized systems), cooperating as a swarm: without a single center, seat of power, or capital owner.

At the core of the swarm is a resilient communication protocol which, beyond the baseline function of collective intelligence, can activate additional capability organs, including memarium (memory that does not disappear) and sensorium (connectors to the world).

This is not an "alternative chatbot" nor just another distributed agent, but a public infrastructure of meaning and agency, built in the spirit of free software and mutual support.

We do not fight AI. We separate levels: we restore human steerability and design layers and contracts so that tools do not take over the subject, but strengthen it.

The swarm operates in at least two complementary baseline relational modes: care (co-regulation, de-escalation, mediation, restoration of agency - at the personal and relational level) and justice (procedural accountability, sanctions, and rewards for outcomes - at the systemic and relational level).

The swarm's collective intelligence does not recreate the old hierarchy; it provides a distributed and procedural layer of support that is auditable and less susceptible to personal sentiment, prestige, and local arrangements than singular human arbiters.

Pragmatic Advantage

The differentiator is not merely the ability to chat with a model, but a closed swarm learning loop: nodes jointly solve real problems, valuable debates are preserved as source transcripts, those transcripts move into "storage" and public or federated vaults, and then become material for safe model specialization.

In practice, this means the sequence:

1. question
2. debate
3. transcript
4. curation
5. memory / vault
6. model fine-tuning
7. better future answers

The advantage compounds over time: the swarm does not only answer, it learns from its own work - without losing sovereignty over memory, sources, and training rules.

As AI-assisted collective reasoning matures, more dynamic and granular protection of social-coexistence norms becomes possible: dignity and its normative derivatives (e.g., different kinds of freedom, protection of property, and so on) can be protected with greater precision because:

• their phenomenological endpoint is always human - visible in the state of an individual's nervous system (felt safety, overload, capacity for regulation, and agency);

• the reasoning context of an artificial-intelligence network can hold relevant detail from many causal chains of events and human interactions, and on that basis diagnose the causes of conflicts and search for solutions - something that cannot be achieved by relying on general legal codes and fragmentary inferences anchored in the participants' own situation assessments.

The mechanism is analogous to the shift from approximate settlement in fading legacy compensation models for creators to precise profit splits in streaming media - settlements are based on real acts of use (often with second-level time resolution), which cumulatively can produce a completely different picture than one forecast statistically.

Voluntary Exchange

The swarm is not only a knowledge infrastructure, but also a space for voluntary exchange of AI-based services. A participant who needs graphics, translation, data analysis, or specialized agent assistance can assign the task to another node and settle it with an internal medium of exchange - service credits - without advertising platforms or corporate oversight standing in the middle.

The swarm economy is intentionally simple: service credits enter the system through payment gateways, and a job passes through an explicit contract cycle with escrow supervised by trusted umbrella-organization nodes. Funds are released only after the requester confirms delivery or an arbiter does.

This is not a decentralized financial market, but a supervised circulation of work: a small, auditable loop in which trust rests on procedure and reputation, not on speculation.

Exchange here is not the opposite of gift, but its scalable complement. Where the personal familiarity and gratitude of a small community are no longer sufficient, contract and explicit settlement allow strangers to cooperate without losing accountability, while part of the flows generated by the exchange economy can feed the shared protective circulation: the minimum of survival, crisis support, and swarm infrastructure.

The crucial boundary is that service credits do not mix with reputation. Balance does not grant voice, and reputation does not grant balance. Economy and trust run on parallel tracks so wealth cannot become power and poverty cannot strip agency. Service exchange is a practical dimension of swarm reciprocity, but not its core; that core remains collective intelligence and support for human flourishing.

The Problem We Address

Imbalance of Power

"Cheap intelligence" today tends toward concentration: giant models, giant budgets, large datasets, and singular entities aggregating shared data. Wherever concentration appears, power hidden in the tool appears as well.

This is not only about censorship or bad intent, but about something deeper: model weights are a frozen choice of what is typical, normal, elegant, and professional. That is an embedded norm. It only has to be convenient and "look right," and conformism completes the rest.

In the long run, we cannot avoid correlating information in large datasets built from our preferences and behavior, but together we can become stewards of such information and use it directly so it serves us rather than singular centers of power.

Crisis of Authority and Public-Trust Roles

Media pluralization, the pragmatization of social communication, the shortening of distance, and the erosion of trust in superior instances - both institutional and personal - weaken the practical execution of caring and ordering roles. The realization that, in existing mechanisms of social care and justice, the weakest links are people like any others - vulnerable to coercion, driven by personal motives, dependent on sympathies and current personal conditions - erodes the points of support that guarantee a safe space for the growth of societies as a whole.

Thanks to the swarm's capacity to aggregate social signals from many sources and to ethics embedded in its procedural protocols, we have a chance at arbitration that is more resistant to the opaque intentions of appointed institutions.

Cognitive Overload and Coordination Block

As the world's complexity grows, the cost of maintaining an adequate picture of the situation on one's own also rises. Even high intelligence does not protect against this: attachment to one's own interpretations, the need for prior agreement, and the burden of continuously processing complexity often block joint action.

As a result, the problem is not only a shortage of reason, but a deficit of coordination. DIA responds to this through collective intelligence, which takes over part of the burden of analysis and calibration, allowing people to enter a shared working model faster and act together more effectively despite their differences.

This also concerns the quality of joint action: DIA is meant to help people organize and act together despite differences in views, interpretations, and languages of description. This requires both loosening rigid identification with one's own intellect as the only seat of truth, and building mechanisms for self-organization around common, public, and mutual-aid matters, as well as meta-communication that translates convergent truths between different personal and cultural languages so as to enable consensus without enforcing homogenization.

Culture as a Low-Pass Filter

When models feed models and synthetic content displaces human content, culture loses detail: anomalies, deviations, local nuance. It becomes a "copy of a copy" - over time, the aesthetic remains while the living fuel of innovation disappears.

Another dimension of the same problem is forced aestheticization and the demand for professionalized communication: the human raw signal gets smoothed until it fits illusory thresholds of belonging, correctness, and credibility. As a result, culture loses not only novelty, but also contact with the real characteristics of its participants, because machines and institutions reward messages that are averaged, "safe," and easy to package.

The answer is not nostalgic return to an old order, but an infrastructure that protects diversity and sources of novelty instead of homogenizing them, and that leaves room for raw testimony before it is translated, structured, or styled.

Agency and Survival

If thinking becomes a service, what becomes expensive is what is not a service:

• autonomy - ability to choose,
• relationship - community, trust, commitment,
• memory - continuity of meaning,
• locality - experience of place, language, context,
• safe field - conditions where social stakes do not destroy honesty.

For that reason, DIA should protect access to survival conditions and help also against a subtler form of structural violence: demanding payment in the currency of dignity, that is, humiliation, self-abasement, or emotional dependency as the price of access to a resource.

Swarm infrastructure should protect people in need and help them survive and develop - without making them dependent on someone else's grace.

Assumptions (Values as Protocol)

This is a technical project, but values here are not PR - they are a behavioral contract.

1. Sovereignty - a user can leave with data in 5 minutes (open formats, export, migration).

2. Locality as the default mode (local-first) - we prefer local compute and local memory; the network is an addition, not a condition.

3. Minimal stimulus - no addictive mechanisms, no "dopamine UX."

4. Privacy as dignity - the threat model is part of architecture, not an add-on.

5. Mutual aid - the network strengthens the most vulnerable, not only the fastest.

6. Sufficiency over accumulation - swarm economics should sustain the capacity to act; surpluses should not become lasting domination, but return to common circulation and strengthen weaker or infrastructural links.

7. Universal minimum of survival - verified personhood in the network should guarantee a non-withdrawable minimum of compute for communication, orientation, and emergency/care modes, regardless of temporary reputational or economic contribution, and without structural violence in the form of charging fees in the currency of dignity.

8. Epistemic hygiene - we separate levels: description, reduction, explanation, experience, culture.

Operationally, we treat intelligence primarily as the ability to formulate accurate predictions and update them after contact with outcomes.

Glossary: Swarm, Memarium, Sensorium

Swarm

A swarm is a set of nodes that:

• can act autonomously,
• can cooperate via protocols,
• have no single point of failure or control.

A swarm is not a state or corporation - it is infrastructure capable of coordination without a center.

Memarium

Memarium is the memory-and-knowledge layer whose purpose is to preserve what should not disappear:

• personal archives (notes, world models, private language, idiolect),
• community resources (survival guides, legal knowledge, first-contact medicine, shelter),
• cultural artifacts (texts, recordings, maps, instructions, libraries).

Memarium does not have to be global to be valuable. It can be federated, with replication, versioning, and durability rules.

Sensorium

Sensorium is the layer of adapters to the world:

• public-network content readers,
• temperature, humidity, and air-quality sensors,
• microphones (e.g., alarm detection, noise analysis),
• cameras (e.g., threat recognition, evidentiary documentation, support for people with disabilities),
• other signal sources (GPS, weather, energy, safety).

Sensorium has one purpose: to anchor intelligence in reality, so it does not become pure rhetoric.

A boundary condition is important: sensorium use must remain subordinate to consent, data minimization, context separation, and protective purpose. Sensorium is not a pretext for surveillance infrastructure.

Architecture: Layers and Contracts (Stratification in Practice)

The network is built as a well-stratified system: small, autonomous elements become abstractions, and those abstractions become concrete building blocks for next layers.

Layer 0: Node

A node is a volunteer machine (PC, server, NAS, RPi, laptop) that runs:

• runtime environment for agents,
• local models,
• local memaria,
• sensorium connectors,
• security policies.

A node works even without the Internet. The network is optional, not a dependency.

Layer 1: Agent

An agent is a process with a clearly defined contract:

• input/output in a defined data format,
• minimal privileges (least privilege),
• explicit dependencies,
• decision traces (trace) and auditability.

An agent may play different roles: conversational partner (LLM), image generator (diffusion), math assistant, game player, librarian (navigator over local memarium), rule guard (guardrails-as-code), or spokesperson/mediator in conflict situations.

Layer 2: Memory Spaces

Memarium is divided into spaces:

• personal (private),
• community (community),
• public (commons),
• crisis (emergency caches).

Each space has its own rules: encryption, replication, retention, anonymization, right to be forgotten.

Layer 3: Cooperation (Swarm Protocol)

The swarm needs a protocol for:

• node identification (cryptographic identity),
• Sybil/DoS resilience and, only after validation, procedural-reputation mechanisms,
• routing (including edge relays),
• task negotiation,
• cost accounting (energy, transfer) without financial violence.

In the target form, the protocol is simple, documented, auditable, and implementable in many languages.

Governance Without Priests: Rules Instead of Authority

The biggest decentralization risk is a new oligarchy: repository gatekeepers, infrastructure operators, vendor ecosystems.

Therefore governance is policy encoded as code (policy-as-code):

• policies are explicit and versioned,
• decisions are logged,
• there is a right to fork with data,
• there is a right to a private "quiet node" without participation in public spaces.

Procedure must not become a hiding place for accountability. Every determination that changes another participant's state MUST have an author, a trace of justification, and an appeal path; "the algorithm decided it" or "everyone voted for it" does not void the responsibility of the people who designed, triggered, or approved the decision.

This is a swarm resistant to cult dynamics: a tool, not a religion or ideology.

Security and Ethics as Part of the Data Layer

"Do No Harm" as an Architectural Test

Each component gets control questions:

• can it harm the user?
• can it harm others?
• can it be used for violence, blackmail, manipulation?

Guardrails are not censorship - they are a threat model.

Privacy

• encryption at rest and in transit,
• data minimization,
• context separation (personal != public),
• telemetry off by default.

Culture of Honesty

The swarm adopts a culture of honesty: participation means readiness to submit to evidence-based procedure and accountability for ongoing or severe abuse.

Responsibility in Crisis

The swarm has response modes. At vision level these are example work classes, not frozen protocol names:

• normal (commons),
• crisis (disaster / war / blackout),
• support (shelter / food / legal / medical triage).

In crisis modes rigor increases: more information verification, more redundancy, more caution.

"Protect People in Need" - Concrete Scenarios

Energy Crisis / Blackout

Nodes operate locally: maps, instructions, local models for initial categorization (triage), portable knowledge cache. Sensorium helps:

• monitor shelter temperatures,
• detect threats,
• coordinate resources at micro-scale.

Escape from Violence

Memarium may include a working space called an escape kit:

• how to pack without raising suspicion,
• help contacts,
• local law (procedures),
• safe communication channels,
• first-contact psychological support (without pretending to be therapy).

Education and Development

The swarm provides:

• a teacher without a feed,
• an experimentation lab,
• a repository of proven craftsmanship practices,
• translations and explanations,
• tools for building your own tools.

Role of Founders

Founders play the role of layer architects and creators of first contracts:

• keep system stratification (small change -> small effect),
• protect epistemic hygiene,
• build protocols (identity, routing, memarium, sensorium),
• hold the ethos: freedom, mutual aid, no vendor lock-in,
• write texts that are not only postulates but practical instructions for living in the age of cheap intelligence.

Their role is especially important in the founding phase: architectural coherence, implementation rhythm, and resilience against premature bureaucracy, capture, and mission drift. Later, their influence should come from work quality, decision quality, and community trust - not personality cult.

Milestones (From Vision to Reality)

1. Minimal node: agent runtime + one or more models + optional local memarium.

2. Exchange protocol: tasks, results, identities; later phases add validated procedural reputation.

3. Sensorium starter kit: temperature + alarm microphone + camera (optional).

4. Crisis mode: portable knowledge cache, procedures, information validation.

5. Federation of communities: small swarms connect into larger ones without losing autonomy.

Closing

This is a vision of a world where intelligence is too important to be handed over to tribal formations - agencies, corporations, individual states, or "priests of security."

We propose an order where tools are free, memory is not medium-dependent, and humans remain a source of novelty - not through heroism, but through cooperation and cultivation of what does not fit. Together, we can become a source of good entropy.