From Organic Design wiki

A holarchy is a hierarchical structure in which each part or component forms an autonomous whole that is also a part of a larger system. This concept was introduced by the American philosopher Arthur Koestler, who argued that complex systems could be understood as interconnected networks of holons – self-organising units that are both parts and wholes at the same time. In other words, a holarchy is a structure in which each part has its own identity and purpose while simultaneously being part of a larger whole[1]. Koestler called these two aspects of a holon the self-assertive and the integrative behaviours respectively. He described the former as an inward-facing system operating with flexible strategies within an individual holon. The latter as an outward-facing system operating with fixed rules as a network, or holarchy.

In terms of information technology, we might classify the holarchy concept as a universal middleware and as a "web3" concept in both the decentralised and semantic senses.


About this document

This document (as of March 2024) is still in draft form, it is not yet complete in the sense of a buildable specification, but gives a clear idea of what we're trying to achieve.

Intended audience

This document is primarily for reasonably IT-savvy people with an interest in independence, sovereignty and decentralised offline-first organisation. This means people who are already familiar with the pitfalls of centralised management and control.

Document purpose

The primary motivation for writing this is for our own internal group to have a clear picture of the current state of the concept and high-level overview of the project. The concept is still being developed and refined, there are so many aspects to it, and we need to have a clear summary of how they all tie together and contribute to the idea as a whole.

We'd like this article to be a clear description and entry point for our own AI so that it can help us to represent or own organisations and projects as a holarchy, and to continuously improve this representation in collaboration with us. AI technology is moving very fast, and we believe within a year or so our own local AI agent will be capable of understanding the holon model described in this article and operating in accord with it.


For now we're using the numbered references mainly for interesting side notes and reminders rather than sources and citations

Etymology of "holarchy"

The suffix "-archy" comes from the Greek "archein," meaning "to rule" or "to lead". Based on its etymology, "holarchy" implies a form of organisation or structure where each unit (or 'whole') is both a part and a whole in itself. It suggests a hierarchical system where each level or unit is a self-contained whole that fits into larger wholes. This concept often appears in systems theory and organisational studies, emphasising the nested, self-similar, or fractal nature of systems.

Another closely related word is "holocracy". Both concepts involve a recognition of "wholeness" in each part of a system. Holarchy integrates this with a hierarchical structure, offering a balance between rules and autonomy, whereas holocracy is more inclined towards ensuring autonomy and distributed governance.

The term "holarchy" better describes our system, because it's explicitly defining a hierarchical system of authoritative rules - collaborative rules that provably maximise individual autonomy. "Holocracy" best fits a purely decentralised system that does not define any concept of authority or hierarchy.

Organic technology

Many people who are strong believers in the idea of humans living fully in accord with nature think that technology has no place in this vision. But by looking at how the cells in the human body are able to live together as a community with a population of over fifty trillion reveals that technology is essential. The cells manufacture and maintain huge infrastructures including the equivalent of buildings that are tens of thousands of stories high, sophisticated networking systems and even an energy based financial and banking system.


The fractal nature of life allows us to equate the biological cell with a person, and a single person with the planetary organism. In his book Spontaneous Evolution, Bruce Lipton shows us that we can learn from our cells how to live together in peace and harmony as a single organism since they're a living example of it, and have been doing it for millions of years.

The rules that define fractals are deceptively simple. Often a few lines of code can yield a mathematical structure of incredible harmony and beauty such as the example shown in the image to the right.

Holarchy is a complex fractal due to being a recursive structure of feedback relationships. It offers us a framework that can achieve the kind of large-scale organisation we see in biology, and can be understood clearly in terms of our own technological infrastructure.

Evolution and economy

Both evolution and economy are systems which, like fractals, involve extreme complexity and yet can be defined in very simple terms. This is quite well known in the case of evolution, for example David Deutsch described it in his book The Beginning of Infinity as being simply the "creation of knowledge through alternating variation and selection".

This complexity from simplicity aspect is not so well known in the case of economy, where in our modern society, we've generally come to believe that only extremely complex mathematical tools can be used to manage something as complex and nuanced as the economy.

A more traditional perspective however, is the so-called "Austrian school" of economics which is strongly opposed to the idea of a centrally planned or regulated economy. The core Austrian idea is that economic order emerges naturally and efficiently from the interactions of individual agents, each pursuing their own interests within a free market. Any deviation from this reduces the efficiency of the economy and reduces the sovereignty and wealth of the participants.

The fundamental foundations to economy are found in consciousness itself. When conditions arise to us, we collect together a set of possible paths of action we could take in response. This process is fundamentally economic in nature because these optional paths are all weighted (prioritised) according to their perceived costs and benefits in terms of energy and outcomes.

Holarchy is an organisational pattern which inherently embodies these two systems, both complimenting and augmenting each other. The evolutionary system underlies the specialisation of the economy, but yet also depends on the economy for the distribution of material resource. In our system the economic aspect is really a subset of the evolutionary aspect, it's the selection aspect of evolutionary system.

Although holarchy is not traditionally connected with these two behaviours in such a direct way, when it's defined in terms of the specific four-quadrant mechanism we're introducing herein, two feedback loops emerge which inherently express them.

The role of internet

The internet connects all of Humanity, and is evolving into an ever more complex, resilient and organised system. It's organised in layers of open protocols from the most fundamental physical layer up to the high-level organised layer of application protocols.

The internet is generally referred to as having gone through a few different versions or phases, the first was characterised by servers and technology specialists being responsible for generating and maintaining the content. The name "web 2.0" was given to the broad phase that came with blog and wiki software in which the vast majority of content was being generated by the end users. It's expected that by the end of the decade, the vast majority of all content on the web will be AI-generated.

The meaning of "web3" was originally used to refer to the semantic web which was envisioned to be a new level of organisation of the web's content brought about by metadata annotations. But web3 started to slow in its progress with corporate interests gravitating instead towards deep learning and AI as solutions to organisation. The "web3" term ended up referring to the decentralised nature of the web which started gaining popularity with the introduction of blockchain technology.

Interestingly, the holarchy architecture actually fulfils both definitions of "web3", because it maintains an evolving ontology in which all content is organised semantically, as well as being able to function ideally in a fully decentralised environment. It's a unified ontology of knowledge as well as a map of the usage of that knowledge.

Knowledge-sharing and organisation are so essential to a harmonious society, that they ought to be provided at the basic abstraction layer of the common networking protocols. The holarchy is a networking protocol that allows participants of the network (holons) to interact together with a common means of organising attention and resources and of sharing, adapting, using and assessing knowledge.[2].

Universal middleware

The concept of a universal middleware, or as Elon Musk referred to the concept, the "everything app", is an inevitable phase of the way we organise using information technology. Essentially, China has already arrived at this "everything app" phase with WeChat which many call the "operating system of china".

Soon all apps across all technologies and platforms will be fully usable by API (most already are) so that AI can use them on our behalf. Application interfaces aimed at desktop, web and mobile contexts will eventually fall into decline through lack of direct human demand.

Another aspect of universal middleware is that it will be perfectly capable of operating any human user interface on our behalf as well, such as browsing web sites, using desktop applications, watching videos, listening to podcasts or even having real-time audio or video conversations. See for example the OS World open source project aimed at this aspect of machine connectivity.

All our hardware devices are becoming net-connected too, as the internet of things (IoT) becomes ever more prevalent. Even older devices are becoming intractable via API, for example with wifi connected universal remote controls and other "smart home" technology.

There will soon be many universal middleware offerings, most likely every large tech player will be pushing their own versions, but also libre software will have its offerings too, and these will typically be designed to be maximally interoperable with each other.

It's important to note that while there are many different universal middleware projects, and potentially even many of them that are modelled directly on the holarchy principle, there can be only one holarchy.

This is not to say that all of them are fake except for the one true holarchy, it's just that one of the main objectives of the holarchy concept itself is unification. So all holarchy deployments, no matter their origins, are objectively dedicated to seeking each other out, and merging into a single network.

Self-organisation and agency

Self-organisation is a concept usually associated with self-organising systems. These are systems that organise themselves spontaneously without the need for external input. Holons are self-organising, which means they involve a structure or pattern that emerges without any external command or central control directing its formation.

The concept of cognitive agency or cognition (in a very generalised form) plays a central role in the functioning of self-organising systems. Agency, in this context, does not refer only to human-like thought processes or consciousness but to a system's ability to process information, respond to environmental stimuli, and adapt accordingly. This form of "perception" enables the system to modify its behaviour based on the state it detects, leading to dynamic adjustments that enhance its stability or efficiency.

This subjective agency aspect of the system means that often the best analogies to concepts in the holon model are things that we're very familiar with in our own everyday experience. For example we say that a holon's "salience is that which is within its field of awareness" or we might refer to a holon's "thread of experience" or that a holon's "unconscious" activity is that which does not receive focus from high-level agency. All these things have obvious and precise meanings in the context of the holon model.

We can say that the self-organisation concept gains a sense of self by the inclusion of this cognitive aspect, thereby becoming a self-organisation, a self-instantiating class.

This is the state of self as an organisation. A holon is a structure of state and behaviour that self-organises and co-evolves with its environment. The concept of self-organisation is essentially also stating that self is an organisation (i.e. a group of entities that are organised toward a common objective) and that it has subjective agency.

An organisation is an abstract conceptual structure that ontologically connects our information about the world with our actions. It allows us to represent ourselves in the abstract, to maintain an informational self-representation in the form of a self-organisational structure. A self-representation is an central aspect of cognitive agency and self-organisation.

The concept of self-organisation implies a direction of self-sovereignty, full unencumbered agency over and responsibility for ones body, actions, decisions and time. This agency over ourselves is our responsibility to maintain, both for ourselves and in supporting the collective responsibility for it too.

The concept of private property is defined as the physical scarce resource that the self has full unencumbered agency over and responsibility for, rooted in our own bodies and then expanding out from there using organisation. A self-organisational structure representing ourselves in the world of actual resource.

Self-organisation in the context of a holon is a form of continuous improvement system involving an iterative feedback loop between the individual holon and the collective holarchy. And this feedback loop facilitates two forms of improvement, the individual improvement which is the self-assertive behaviour, and the collective improvement which is the integrative behaviour.

User interface

The most general software application has the most general use-case, a concept something something like "life assistant". This would be an application that is as transparent as possible in terms of getting things done with it, and it's so general that it can help with absolutely anything you might want to do. It's not hard to imagine this now that AI has very general agency.

A universal middleware is essentially a "universal connector" and "translator" that allows the connection of diverse informational representations into groups. An LLM of about the chatGPT-4 level already make extremely good universal connectors, which means that we already have the ability to create a unified ontology of our informational lives that is continuously kept up to date with actual state and can act as an interface to the connected things, in other words, a self-organisation structure.

It's easy to imagine an up-to-date and complete information representation of our lives, because we already have various folder structures and other informational interfaces connecting us with a big part of our lives. Often times we're acutely aware of how fragmented our informational life is, and we long for the convenience of seamless connectivity between all these fragmented aspects.

From the system's perspective, even human users are just another form of connection instance. Communications with users occurs in the form of user interface sessions, which are themselves composed of user experiences and user stories. The connector is agnostic to the specific medium, interface language, style and preference involved in a user connection.

These connections participate in groups which we call organisations even if they're just a static informational group because, no matter their simplicity, they all have the potential to evolve into any complexity of organisation.[3]

Organisations depend on resource such as materials, attention from people filling roles, executional focus etc. Even a simple static file requires storage space, and reading it requires bandwidth and attention.

The dependency on resource creates a hierarchy with the largest resource/energy reservoirs at the top, allocating resource to their primary salient categories of usage. The most logical root for this hierarchy is the user themselves, the "home folder" that always represents their present state.

This is what we call self-organisation, our own lives represented as an organisational structure which is continuously fitted to match our present state. A specific form of universal middleware concept.

The organisation-centric nature of the holon means that the model revolves around very fundamental concepts that organisations are defined in terms of, such as resources, processes, knowledge, development, roles, production etc. When we talk about society and economy, we're in the specific context of a society of self-organisations exchanging resources and behaviour patterns. These organisational concepts are the main focus of the system description in this article.

Virtual companions

The most natural evolution for AI agents is that they will become as much like other humans, in terms of the way we interact with them, as possible. The technology at the moment is right on the precipice of practically perfecting this ideal in terms of virtual assistants or companions.

Currently in 2024, we're starting to see human like personal assistant such as Open AI introducing their natural voice mobile interface to ChatGPT[4] which aim to be helpful AI agents that we can have with us all day. Unfortunately, the reality is nothing like the introductory videos, it's extremely frustrating to the point of being worse than useless. But the fact that we're now seeing these things on the market means they'll inevitably become much more usable after a year or so of product iterations.

During 2025 we'll most likely see the introduction of these virtual assistants en masse, ones that are always with us and are connected in with our current context knowing exactly what's going on, most likely with true intelligent understanding. They'll be able to suggest advice, tips and ideas in response to our specific questioning, but also spontaneously when they can see their advice would be useful.

The use-cases for a virtual companion are practically infinite. Imagine an intelligent, knowledgeable and patient companion who has a lot of experience in everything, who is always with you ready to help with whatever you're doing, specifically there to help you and your projects develop and thrive. We've offered a few specific example use-cases at the end of the article.

Our holarchy project

A holon structure is rooted in oneself to reflect the fact that we're all the permanent centre of our own lives (experiential structures). This means that by default, all external exchanges and decisions will be optimising for self primarily. This is natural and is actually necessary, but unconditional maximisation is not at all optimal. Comparing different strategies for guiding our external interactions is a philosophical discussion which is beyond the scope article, but suffice to say here that we believe holarchy is the most rational strategy.

Holarchy is generally considered as a philosophical framework of attributes a system should have in order to be aligned with the principles we observe in living systems. It's usually presented as more of a set of guidelines than a specific system definition.

Here at Organic Design we believe that there is a simple organisational structure at the heart of and common to all living systems, and even underpinning consciousness itself.

We believe that holarchy is a very definite and describable system. It's in the form of a cognitive architecture following the self-organisation concept described above. Holarchy comes with a definite strategy for the aforementioned external connection issue, which optimises for both self and whole; i.e. its self-assertive and integrative behaviours.

We're researching and developing the holarchy concept in the form of a peer-to-peer network of self-organising holons.[5] We're currently attempting to articulate the holarchy concept with enough detail and clarity to define a software design pattern from it.

On the research side of the project, we extend out to a wider focus than the development to encompass the philosophical aspects of holarchy. The political philosophy perspective concerning the kinds of large-scale social order and progress that the holarchy system of organisation implies. And foundational ontological perspective of seeing holarchy as a foundation for cognition and even of reality. We're working on articulating these concepts in the philosophy of the holarchy article.

The project's development effort can be broken into three general areas: the p2p network architecture, the holarchy organisational system and AI integration. The purpose of this article is to introduce these three aspects starting with its peer-to-peer foundation. And following them at the end, we discuss the some high-level organisational patterns and use-cases for the system.

This project is our attempt at articulating the holarchy concept, operating our own organisation, projects and lives in accord with it, and presenting it in the most understandable, resilient, reproducible and usable form that we can.

Peer-to-peer network development

Holarchy is inherently peer-to-peer in nature due to all nodes at every level being holons embodying both individual and collective oriented behaviours.

Peer-to-peer networks are a class of network where there is only one kind of participant which can interact in both client-like and server-like ways. This means a peer-to-peer network is a more general architecture than client-server, and also that peer-to-peer is not opposed to client-server, it can dynamically represent client-server in response to the right conditions.

In more general terms, we can say that the peer-to-peer pattern is a group-pattern whereby all members are both independent participants as well as maintaining a shared state together. Knowledge gained locally is merged into the shared context and is available to guide all participants. This creates a feedback loop so that both individual and collective sides are continuously co-evolving.

We were not able to find any existing libre software project that we felt really catered for the holarchy's specific networking requirements, so we've spent the last few years developing a custom solution based mainly on LibP2P, IPFS and Welo. It's a fully libre software solution which can be used independently of the holarchy, for use cases such as a decentralised content distribution network or distributed backup system.

Universal filesystem

One of the main roles of the holarchy is as a general resource allocation system, and the bandwidth and storage that connects holons into the holarchy are amongst the resources that are managed organisationally by the holarchy itself.

This means that the networking layer for the holarchy should ideally be transport, technology and storage agnostic. Presenting a common networking ability that's aligned with the architecture of the holarchy, and is able to dynamically incorporate into a common interface all kinds of network and storage resources that are made available to it.

The network layer needs to be able to provide the holarchy layer above with the ability to allocate and prioritise these bandwidth and storage resources flexibly in accord with the needs of the complex organisational structures that holons can represent.

A universal middleware needs a universal filesystem. A common interface through which all of the organisation's informational content can be managed and distributed.

Mesh networking

The most pure p2p architecture is the mesh network, it's the most general of all networking architectures because it is the most ontologically fundamental. It can function under the most restrictive and unreliable environments. The peers in a p2p network can support higher levels of abstraction allowing groups of peers to behave as a different topology such as a client-server network with the associated gains in efficiency, but client-server cannot behave in a peer-to-peer way without losing efficiency.

The most extreme degraded state of network is no network at all. When a network's peers can continue to operate even when completely isolated, it's said to be an offline-first network. Obviously there will be much less capabilities available in an offline state, but the idea is that local organisations operate with cache and "outbox" patterns of behaviour. This allows continuous local operation that synchronises with the wider community as circumstances permit.

The highest level of organisation in IT infrastructure is something like Kubernetes running in the cloud. Any application can be deployed at the click of a button, and the hardware supporting the deployment can scale up and down in real-time to meet demand dynamically throughout the network.

Since a mesh networking system is able to function in such a broad range of environments, it serves well as a glue for combining physical infrastructures and transports. For example, being able to expand the mesh over bluetooth or carrier pidgin[6].

Offline-first design

Back in the 90's when bandwidth was scarce and costly, we made heavy use of the "outbox" in our email programs. We would go through our inboxes replying to messages and composing new messages, and we'd be offline the whole time. Only when we'd finished writing the messages would we finally connect to the internet, hit "send and receive" and then disconnect again as soon as it was finished transferring data. We'd usually have a cup of coffee while the system laboured away transferring all those kilobytes.

Most of the time, offline systems are not necessary these days, and so software is written with the assumption of a permanent network connection, for example by depending on domain-name resolution or other network services. Most of the time this is not a problem, but in those situations where it is a problem, it's a really big problem because most of the software is completely incapacitated.

For example, nearly all of our favourite chat programs will fail even to send messages between the locals on the same LAN if the internet connection goes down. Many of these programs will not even start up without a connection.

Peer-to-peer systems are much easier to design in an offline-first way than client-server systems are, because they're designed to operate responsively regardless of peers spontaneously coming and going (a phenomena called "churn").

Since the philosophy of the holarchy supports local independence and sovereignty, and because it's naturally peer-to-peer in structure, it's a natural decision to aim for an offline-first solution.

The offline-first aspect also plays a key role in deployment of the system. The system will use its own package-style organisation to manage itself as a set of deployable packages and variations. Being inherently offline-first, the packages are usable, scalable and spreadable no matter how basic the situation they're booted into.

Not only is the offline-first paradigm more independent and resilient, it's also more responsive, resource efficient, accessible and shareable.

The offline-first approach is the perfect compliment to mesh-networking. Mesh-networking is about interacting with a diverse variety of networking resources and dynamically changing connections or reallocating resources, which means that it needs operational layer of the system to be decoupled from the underlying networking. This decoupling is exactly what offline-first provides.

Independence and resilience

We've discussed network-based independence already, but the system also supports some other important dimensions of independence which we give a very brief overview of here. Although these dimensions are not directly related to the networking, the peer-to-peer model in general enables far greater independence, resilience and adaptability.

The Libre software movement which advocates that the community should have access to software for all its needs which is free, open source, understandable and adjustable to local needs. All the software we're building and depend on is libre software. It's developed right from the seed concept as libre software, not that it will eventually be opened up after a particular stable release or after critical mass is reached. The holarchy itself is also all about the sharing, transparency and understanding of knowledge too.

Data sovereignty is inherently supported by decentralised models, because the most critical data needs to be the most local to ensure uninterrupted operation when problems occur in the wider operational context. Data sovereignty means having full control over this local data, just as one would expect to have over other private property.

Having local access to AI is a really important aspect of our system. It's currently not quite economically feasible as it costs around USD10K for hardware capable of running an AI agent that can play the role of a holarchy assistant (we'll come back to this later). All aspects of any AI we use locally must be completely libre software including all the training material and processes, because it needs to be completely trustworthy and unbiased.

The most fundamental aspect of independence concerns our survival needs, and so the real-world holons composing our own internal experimental holarchy are projects focused on land, energy, food, water and the sharing of permaculture, planting and off-grid living knowledge.

Peer-to-peer collective

When we talk about the collective aspects of a system, it's natural to think about it in a centralised way like a "server" or an "institution". But it's important to remember that in the peer-to-peer context, although the collective aspect behaves like and is treated like its centralised counterpart, it is in fact a product of the individuals solely.

This is the case with the integrative behaviour of a holon, it's a behaviour that results in a collective aspect of the network that's common to and useful to all individuals, but it's existence depends completely on the individuals performing the integrative behaviour.

The four-quadrant holon model

The purpose of the holarchy is not only individual self-organisation, but about the whole network self-organising as a harmonious self-organisation (holarchy) of self-organisations (holons).

Our four-quadrant holon model proceeds from Koestler's four concepts of the integrative (collective) and self-assertive (individual) behaviours, and the fixed-rules and flexible-strategies mapped onto an orthogonal pair of axes. The aforementioned concepts map respectively onto the top, bottom, left and right directions of these axes which we call the "primary" axes.

The quadrants are the four corners delineated by the primary axes, and reside at the ends of a pair of orthogonal diagonal axes. These diagonal axes each connect two quadrants together into the feedback loops that express the variation and selection aspects of the evolutionary principle. We'll come back to the diagonals and their feedback loops further on in the article.

The quadrants are like autonomous organisational "departments" that all holons have, which ensure that they all organise themselves and collaboratively support the whole collective network in alignment and harmony. The holons can be composed into organisational structure of any scale and complexity. The four quadrants are common to all holons, and therefore to all organisational structure representable by holons.

This means the quadrants are extremely general fundamental concepts having a strong philosophical connection. We believe this pattern of precisely these four meanings are inherent to meaning itself (the meaning-making process), and are epistemically convergent, or in other words all contexts involving sufficient intelligence would eventually discover this specific pattern. It's no surprise therefore that we see these four meanings pop up together in many traditions throughout history, such as in the form of Aristotle's "four causes" or in philosophical Taoism which are both thousands of years old.

Perhaps the most recent incarnation of the four quadrants is Ken Wilbur's Integral Theory that's gained popularity in the last couple of decades. The quadrants in our model correspond precisely with the four quadrants of Integral Theory, but in our model the positioning of the quadrants is vertically flipped from Integral Theory. The justification for this flippage is that for our purpose, the most important attribute of "above" is its natural relation to wider scope (outward, encompassing more, collective), and conversely the natural relation of "below" to narrower scope which is more specific and deeper within. Note also that Ken Wilbur mentions the concept of an "integral holon" in some of his writing, but we're currently unsure whether his concept follows the same mapping to Koestler's core holon concepts as our model, if it does then we'd prefer to use the term "integral holon" too.

Our holon model is a refinement of Koestler's general concept which has been designed specifically for the information technology context. To define a software specification, the quadrants need to be understood in terms of specific system interactions. We introduce this refined view of the quadrants in this section, but we also have a more in depth and complete description in the holon mechanism article.

Holarchy is an agent-oriented model of reality, which a kind of middle ground between materialistic and idealistic models. It requires a logical and rational systemic underpinning while at the same time taking on an idealistic form whereby so-called "external reality" is a construct that is agreed upon and maintained by the agents in the form of experiential phenomena. In this article, we're focused primarily on holarchy as a network architecture in the context of information technology, so we leave the philosophical details for the philosophy of the holarchy article, but we mention it here because there are two important aspects of this that concern us here.

First, in an agent-oriented reality, the model itself is genuinely constitutive of reality, rather than merely a theoretical construct or convenient fiction. This means that within the context of the holarchy, the four quadrants are not just a convenient lens through which to analyse and categorise the system, but are constitutive of the system itself. For this reason, we go into a lot of detail about the four quadrants as actual processes. or "departments" of a holon.

The second implication of the agent-oriented approach is that it means the system fundamentally takes the form of a cognitive architecture, which describes how agents interact together and perceive, create and maintain their world. This is where we'll start our description of the four quadrant holon model.

Cognitive architecture

Here we introduce the concept of a cognitive architecture and some related software design patterns that our model embodies.

The general context of the system is the self-organisation concept described above, and more specifically it takes the form of an agency-agnostic cognitive architecture, i.e. any agency can participate regardless of its attributes such as simplicity, complexity, analogue, digital, organic, electronic, photonic etc.

A cognitive architecture is a systemic foundation for agency which defines an abstract reflection of the environment it finds itself to be within. It gives participating agents a local subjective lens or point-of-view (POV) through which to perceive reality. The cognitive architecture defines its world (universe of possible experiential content and interaction).

The dynamic that takes place within this subjective individual point of view corresponds to Koestler's self-assertive behaviour. And the dynamic that occurs outside of it is the objective collective behaviour which corresponds to Koestler's integrative behaviour.

We need to introduce some general concepts involved in cognition, and their specific meanings in the context the our implementation of the holarchy concept.

The Ship of Theseus

Ship of Theseus.jpg

We want to start this section of preliminary cognition-related concepts with an ancient Greek legend called the "Ship of Theseus", because it will aid us greatly in describing the dynamic pattern we're trying to achieve with the holon structure, and the subtle complexities it involves.

According to legend, Theseus, the mythical Greek founder and king of Athens, rescued the children of Athens from King Minos after slaying the Minotaur and then escaped onto a ship going to Delos. Each year, the Athenians commemorated this by taking the ship on a pilgrimage to Delos to honour Apollo. A question was raised by ancient philosophers: After several centuries of maintenance, if every individual part of the Ship of Theseus was replaced, one at a time, was it still the same ship?

Is the ship that now consists entirely of new material still the same ship? On the one hand, if we consider identity to be based on the material components, then it seems that the ship has completely changed. On the other hand, if we consider identity to be based on the continuity of the ship's form or its function, one could argue that it's still the same ship.

The ship can be seen as an idea which is embodied in all the people who manage, maintain and repair the ship along with all their related intent, knowledge, resources and procedures. If we zoom out to a long enough time-frame, then all the material is seen as continuously in flux, gravitating toward the consistent central idea.

But not only is the material aspect of the ship in flux, all the people and objects that embody the idea of the ship are also in continuous flux. Over time old workers are replaced by younger ones, and better ways of doing things replace old ways. The ship is a material form that's in flux around an organisation of roles and procedures that are also themselves in flux.

Even though this system may evolve until the form of the ship eventually becomes unrecognisable from the original, it's still quite natural for us to recognise the continuity of the ship's identity. It's natural for us, because our society as a whole functions like this, and aspects of all our daily lives and work do too. An organisation's staff, procedures and resources can all be in flux; it has staff turnover and may open new branches or change product lines and services change etc.

The Ship of Theseus is actually a network of ideas. Even though the ship itself is one specific idea, it doesn't exist in isolation, there are also many other ships and all those involved in all the ships regularly exchange knowledge all evolving together as an "idea-cluster".

The ideas are composed of many other ideas, for example the planks that compose the ships are themselves a whole evolving network of knowledge, roles, production and materials that are part of a wider network than just ships. All the ideas in the whole society are connected in some manner, and contribute to each other's evolution, all together forming an inseparable whole.

The central point of the legend is about identity and how it forms a central point around which all aspects of an idea gravitate. Extending the discussion to include the network aspect gives us a clearer picture of the kind of dynamic flow that a class-instance system needs in order to faithfully represent the nature's holarchy pattern.

It's this fluid form of identity and its nature as an idea cluster that's at the core of a holon and the holarchy. We call it the class and instance system and is what we'll introduce over the next few sections.


We use the word "agency" to refer to the ability to apprehend state and instructions and perform any actions that may be implied by them. An "agent" is an actual entity of some kind which has agency, it has the ability to perform various specific actions when called upon in appropriate circumstances. Such an agent might be a user, an AI, an API or OS, a domain-specific language interpreter or many other things. An agent is an agent of change, in our system there is no agentic focus without corresponding activity.

The holarchy is an organisational system which is agency centric since it's a cognitive architecture, but yet it's also agency agnostic, which means that it interacts with any kind of agency in the same way - in the same way as our system of law applies completely to people, but yet is (ideally) person agnostic in its application. This includes being agnostic to whether the agent is simple or complex, or whether its focus is discrete or continuous in nature.[7]

Regardless of their agentic complexity, it's fair to say that all instances have a subjective local point of view consisting of the information and threads of activity within their local scope. They find themselves to be in an organisational context consisting of other sibling instances (other agentic entities) of various classes that are also encapsulating their agency within and presenting their state publicly to be apprehended by the other local siblings.

In terms of information systems, agency essentially represents the ability to execute code, and in organisations it represents the ability to fill a role and perform procedures in it. All change in a holon is due to agents changing local state by performing activities in accord with this same general pattern.

The cybernetic loop

The cybernetic loop is a fundamental concept in cognitive science taking the form of a specific kind of feedback loop. It represents a dynamic process where a system continuously monitors its output, compares it to a desired target state, and then adjusts its actions to minimise the difference, or error, between the two. This kind of loop is also called a control loop, error-correction loop or negative-feedback loop in some disciplines. We usually use the term "control loop".

This iterative loop enables systems to self-regulate and maintain stability by making continuous adjustments based on incoming information, ensuring that they remain on course or adapt to changing conditions. The cybernetic loop plays an essential role in a wide variety of system, from simple thermostat-controlled heating systems to complex organisms and robotics, facilitating effective control, adaptation, and optimisation of processes and systems.

Body schema

The final complex structure that emerges in the local subjective scope of a holon follows the same pattern as the abstract mental representations we have of our own bodies, a concept called the "body schema" in cognitive science.

This internal representation and awareness that individuals have of their own bodies, includes their size, shape, position in space, and the relative positions of body parts. It plays a critical role in our ability to perceive and interact with the external world.

At its core, the body schema involves a continuous feedback loop where sensory information from the body, such as proprioception (awareness of body position) and tactile feedback, is constantly processed and compared to a mental representation of the body. This representation is adjusted based on the incoming sensory data to ensure an accurate perception of one's body and its relationship to the environment. This process can be hierarchical, involving multiple levels of abstraction, and it allows us to perform tasks with precision, adapt to changes in our body's state, and navigate the world effectively.

In essence, the body schema embodies a sophisticated form of the cybernetic loop. A holon has an information data structure that operates in this same pattern in accord with the cybernetic loop, but we refer to it in this context simply as the "self-representation".

The self-representation includes not only the current state, but also the future (objectives) and the past. The future is incorporated by acting as objectives for how the self-representation should be, the self-representation also serves as an interface permitting abstract concepts and ideas to actualise as actions manipulating the external world - it is an ontological representation of reality allowing it to be organised.

It's a lot easier to make the connection between the body-schema and a holon's self-representation when we consider that our body-schema extends beyond our bodies in the form of tools and technology. And even beyond that into the wider culture and society as our values and property become part of our body-schema control structure.


A classic internet meme

The concept of a meme was coined by Richard Dawkins in his 1976 book "The Selfish Gene". It refers to an idea, behaviour, or cultural element that spreads and replicates through imitation and cultural transmission. Just as genes carry biological information, memes carry cultural information, evolving and propagating as they're passed from one individual or generation to another. Memes can encompass a wide range of cultural phenomena, including customs, rituals, fashion trends, catchphrases, and more, playing a crucial role in the evolution of human culture and society. As we've seen in recent years, the internet has allowed memes to spread and evolve much more rapidly, and AI promises to multiply this still more.

Memes are a very similar concept to our idea of the self-representation (in the body-schema sense) within a holon which is effectively a "behaviour package" (a rule-set). Adaptation and evolution are enabled by all instances of the same class form a community which aggregates metadata about the packages and is automatically shared.

This is the same as molecules, proteins and cells that make up an organism all being in flux around form determined by the organism's DNA. Likewise, our own mental cognitive symbols are in flux around forms within the collective unconscious. This lecture by Daniel Denette is a great introduction to biological evolution, cultural evolution, memes and even internet memes.


The holon is itself a group of holons which we call siblings. All the siblings find themselves together in a private informational context through which they can express themselves to each other. The context represents a particular objective that the siblings collaborate together on, and which is provided by the holon - the parent of the sibling group which the group are in service of. In IT terms we'd say that all the siblings are parallel child threads in a shared private scope owned by the parent object.

The focus is the combination of content and thread aspects of system execution. It is the actualised content in the present moment in the context of a particular sibling (that is visible and accessible by the sibling).

The focus occupies a "moment" (also called a "session" or "slot") in time, the duration is context-dependent, for example on the type of agency involved. During this moment the agent performs an action determined by the current condition of shared local context.


In information technology, the term "scope" refers to the names that can be locally referred to by a process. The context mentioned above that agency finds itself within is called "private" scope, and consists of a list of sibling names, which are other things that "reside" within that same scope, such as information and other agents, which are said to be "local" to each other.

We also have "public" scope, which is the subset of the private local names that are made available to the parent context. And "non-local" scope which is network-wide and will be introduced further on.


Focus applies to the present moment and refers to the energy that brings the present moment into being in a particular scope allowing an action to be performed. Salience refers to what will receive focus due to being instantiated ("installed" into the local scope) or "connected into time". Salient things are "in our field of awareness".

Salience and agency go hand-in-hand as neither are meaningful without the other. In terms of organisation, salience is the types of activities (behaviours) that may need to performed, and agency is the ability to actually perform them. Roles that may need to be filled, and those able to fill the roles.


Focus and activity go hand in hand, all focus is in the form of activity being performed. A holon as a whole is a continuous timeline made up from structured threads of activity. A single action occurs in single moment of focus, and the whole stream of activities makes up a thread of "experience".

Focus is always within the context of an activity in a particular state of progress or completion. The top-level activity aspect of a holon is constituted from a future component above, a past component below and the present in the middle.

Activities have a "lifecycle", they start off initially as just intention without any commitment of resource externally. Eventually they reach a mature enough state that they start to form commitment where actual roles and resources become involved. Once such resources are "filled in" sufficiently, aspects of the activity become imminent ("booked into schedules"). Eventually they make their way down into the present where they become active in production generating accounts of completed (past) activity with corresponding state and reputational changes. And finally their informational aspect is integrated both locally and beyond.


A self-representation is an informational structure that represents the state of the holon itself. This is a necessary aspect of an autonomous agent that's based on a continuous improvement cycle. A holon is a continuously improving self-representational structure, developing itself as an organisation and its state of position.

We mentioned above that the holon's self-organisational structure is called its self-representation and is the holon's equivalent of a body-schema. Here we want to go into a little more detail about this self-representation data structure.

The state of a holon-instance is the informational content contained within the instance's scope. Since an instance involves three kinds of scope (public, private and non-local), it also contains three kinds of state corresponding to them. We refer to these three aspects of state all together as simply state.

The private and public state together are called the foreground-state. They're the values associated with the unique names constituting the instance's private and public scope, which is really just a single scope, private by default, but may have any amount of it presented as its public interface.

The non-local aspect of state, also called class-state, background-state or default-state, is the state that the instance has as default by virtue of its class (or more precisely, by virtue of the internal class structure that the class defines). Any local foreground state overrides the default structure and state provided by the class. This is essentially the same way that instances extend and override their classes in traditional OOP.

A holon's state is a continuously maintained self-representation, an abstract version of its real-world counterpart. An information structure that represents the holon's instantiated behaviours and the state of the real resource under its ownership and control. The instance state has exactly the same meaning as in traditional OOP, its the way that the structure and continuity of it are handled that differ.

The representation is bidirectional, on one hand it's always changing to reflect the current state of reality, and on the other it can be used as an interface through which intentions are expressed.

An instance is an informational structure which follows the pattern determined by its class, and also represents its specific real-world state. Any organisation follows this same familiar pattern, they're abstract patterns that we use to manage our resources and information together in society. So the informational structure of an instance is a representation of both the class pattern and of actual resources that fall within its designated objectives.

Its important to note that the representation is not the actual resource, but rather an abstraction of it. The holarchy does not directly contain any of the resources that are being organised by it, rather it contains metadata about the resource. Imagine a spreadsheet of our finances for example, the specific file in question is an instance that represents some financial state in the real world such as bank transactions and balances. This spreadsheet instance also represents a definite spreadsheet idea that determines the structure and methods available in the context of any spreadsheet instance.

The operational work of an instance is to use informational connections to resources to maintain a representation that is ontologically structured in accord with the class. The state of the structure is continuously fitted to the real state of the resource outside the holarchy.

Instances use this representational mechanism to serve as interfaces allowing us to interact with and organise our information and resources using an evolutionary ecosystem of established organisational patterns.

First-class citizens

In the context of programming languages, a first-class citizen is an entity which supports all the operations generally available to other entities. These operations typically include being passed as an argument, returned from a function, and assigned to a variable. In most OOP contexts, objects are first-class citizens, meaning they can be instantiated, manipulated, and passed around in the code just like other basic data types.

The holarchy is not a programming language or OOP environment in the traditional sense, since it's a higher level of organisation based on general cognitive agency. But we use the term regarding holons to imply that every holon instance has all the same inherent four-quadrant form as every other holon instance, regardless of it's depth in the hierarchy of instances, its complexity or simplicity.

First-class citizens are all equal in the sense that they could all evolve into anything else, all essentially have the potential of becoming any other. Holons are all first-class citizens, each having a continuous identity with material, knowledge, objectives and production all in flux around it, like the Ship of Theseus.

Knowledge and patterns

The class-instance concept expressed by the Ship of Theseus legend is all about knowledge and behaviour patterns. Essentially knowledge represents behaviour patterns, it can be communicated, learned, embodied, taught, used, adapted and assessed.

Knowledge is a behaviour pattern in shareable (communicable) form, functionally it depends on community, it is a non-local concept. In a community context, the assessment, adaptation and selection of knowledge leads to an inherent evolutionary aspect to knowledge. Knowledge, language, community and evolution are all interdependent aspects of a single fundamental class-instance mechanism.

An agent can use or embody knowledge locally by establishing it in their self-organisational structure. The more the agent uses it, the more established it becomes. The cost of operating it becomes lower, the embodiment becomes more efficient, which is the patterns becoming more "habitual".

Knowledge is a consistent map of what's established in usage including variations, ordered with the most used being most prominent towards the root. This "meaning map" is a decentralised process involving all local embodiments (instances) of a particular pattern (class).

For our purposes, "behaviour pattern", "organisational pattern" or simply "pattern" are interchangeable terms. Knowledge is what a pattern represents, and the class is the permanent identity by which we refer to and share the knowledge pattern.

The class-instance system is the foundation of the holon model, it's the mechanics that define what we mean when we say "behaviour pattern". Being the "foundation" means its the part of the system that's defined in program code, so we need to introduce a few key software concepts before getting on to the specifics of class and instance.

Production rules

The lifecycle of an activity might simply consist of a single session of a single agent's focus, or it could be a very complex hierarchical structure of projects and roles that activate under specific local conditions throughout time. Activities can be in a variety of organisational forms all determined by their structure, such as continuously developing, reoccurring, one-off, conditional, pipelines and cyclical.

Rules can be composed into complex workflow structures, allowing for the expression of complex logical relationships. Production rules are widely used in expert systems, business rules engines, and knowledge-based applications.

Production rules play an important role in automating decision-making processes, enabling systems to make reasoned choices, offer recommendations, and adapt to changing circumstances based on the knowledge encapsulated in these rules.

Production rules provides a powerful means to represent systems and knowledge that may take all these myriad forms. A production rule consists of two essential parts: conditions and actions. Such rules can be simple and binary such as "if X is the case, then do Y", or they may be very continuous and general such as "while X seems to be an issue, perform behaviour Y to mitigate it".

The rules themselves are in a form that is understandable and actionable by the relevant local agency. There is nothing in the rule content that refers to control-flow or workflow, the flow of focus is determined entirely by the structure of rule composition. In fact it's this lack of reference to control-flow (called a declarative execution paradigm) that gives production rules an inherent composability with each other.

It's the structure of the production rules that defines the conceptual meaning of the organisation, not the agent-oriented content of its production rules. In our system the rules follow the self-organisational structure introduced above.

This pattern allows complex workflow (organisation, control-flow, program execution, process) to be intuitively understandable without specialist knowledge about the workflow mechanism itself.

The blackboard metaphor

This local scope that agents find themselves within when they receive attentional focus follows the blackboard metaphor of execution which, in the case of a holon, goes hand-in-hand with the production rule pattern. The blackboard metaphor represents a group of experts collaborating together around a blackboard, where they each contribute insights toward solving a complex problem.

It's a way to harness collective intelligence in systems with multiple agents, each with specific abilities. This modular and flexible approach allows for emergent solutions and the leveraging of specialised expertise without requiring any single agent to solve the problem alone.

It's widely used in artificial intelligence and distributed computing for its adaptability and collaborative problem-solving capabilities. It's also often chosen for its decoupled approach where agents can collaborate on a problem without needing to coordinate directly with each other.

The organisation that takes place within a biological cell bears striking resemblance to the blackboard metaphor, especially when combined with the production rule concept. The cell essentially defines a local private scope containing resources and enzymes, which is like the private blackboard shared by a set of relevant sibling agents. And the conditions matching relevant actions is like the cell expressing or suppressing particular behaviours in response to it's immediate needs (by dynamically regulating its biochemical pathways and functions in response to environmental conditions).

Workflow and behaviour

What we've been discussing with the blackboard and production-rules is often referred to as "workflow" or "organisation". It's not really referred to as a software design pattern, because it's quite a general concept. It concerns primarily process description and execution. Using the term "workflow" (or "organisation") rather than "execution" or "process" implies operation at a high level of abstraction.

Traditionally production rules are considered to be very discrete in their function, for example the condition part is considered to be similar to an "if-then" statement. But by implementing the production rules in their own private persistent scope as per the blackboard pattern, the rules are permitted to operate asynchronously. The blackboard pattern decouples the agents (knowledge sources) from each other so that they're free to interact via the scope in their own time. This makes the workflow much more flexible so that it can represent complex patterns of behaviour.

Knowledge is in the form of shareable behaviour patterns, production-rule sets in the form of condition:action pairs. Each pair is a feedback loop with the local environment (a cybernetic loop) which can be thought of as the generalised continuous version of a traditional production rule. These design patterns all working together form a kind of continuous workflow and improvement paradigm, which is closely resembles the body-schema concept introduced above.

Class and instance

The Object Oriented Programming (OOP) paradigm was created in the 1950's to try and better fit the data structures and functions of software engineering to the actual entities in real life that were being represented by the software system.

OOP uses "objects", which are instances (specific occurrences) of classes (templates or "blueprints"). Over the years a huge variety of paradigms and languages have emerged that incorporate various aspects of OOP, and also exhibit many new variations on the theme to better fit the dynamics between processes, knowledge, material and agency we experience in the real world. The main difference between OOP paradigms essentially comes down to differences in their functionality of classes and instances.

A class acts as a blueprint for creating objects, defining the properties and behaviours that the objects will have, in other words it is the pattern of behaviours. For instance, if you have a class named Car, it might define properties like colour and make, and behaviours such as drive and stop. An instance, on the other hand, is an actual object created from a class. It represents a specific example of the class with its own unique values for the properties, such as Fred's red Toyota car. While a class provides the template, instances are the real objects you work with in your programs.

A holon is very much like an object in OOP, having public interface and private scope, but rather than the encapsulated (private) behaviours being defined by program code, they're defined by production rules structure operating as continuous behaviour patterns as described in the prior sections.

When we say "patterns of behaviour" we're drawing on the fundamental concepts of "class" and "instance". The term "pattern" implies the ability to repeat a behaviour, refer to it and communicate it. The term also implies composition and structure which, as discussed above, production rules and behaviours are compatible with.

Class and instance are two interdependent concepts which are essentially another software design pattern, although they're so ubiquitous that they're an inherent part of the design of most programming languages, and so are rarely called a design pattern. We'll call them a pattern here, because we're defining our own specific version of the concepts that depend on the software environment for only very basic data-structure capability (one which can support the aforementioned workflow concept).

This pattern is really a "meta-pattern", it encapsulates the concepts of defining and re-using patterns of behaviour or functionality. A class is an abstract "package" of functionality defining how the package would function if it were represented by some actual functional resource - i.e. how a local instance of it would behave.

A "class" is essentially a name (also reference or identifier) that refers to a specific abstract grouping of other class-names, and "instance" refers to a specific "pool" of actual operational resource that is arranged in such a way as to represent classes in its operation. Classes represent sets of related behaviours, whereas instances are groups of actual agents capable of performing behaviours along with its current state of development.

The class aspect of a holon is analogous to Koestler's fixed rules concept, it defines structured possibility space within which instances can select and enact appropriate activities from all the possible ones. In other words, classes define how an instance of it would behave if various conditions were the case.

The instance aspect corresponds to Koestler's flexible strategies, where the behaviours that are expressed match the present local conditions.

We started this section on knowledge and patterns by saying that knowledge, language, community and evolution are all interdependent aspects of a single fundamental cognitive behaviour pattern. All these design patterns working together play a big part in connecting all these aspects into a single system, and we're now ready to start talking about the four quadrants which are the form that this single system takes.

A class is a particular behaviour pattern, and an instance is a class that is actualised by performed and represented by actual real resource. Classes specify collections of resource and agency types and the organisational patterns they would engage in if they were instantiated. Instance is an actual real collection of such resources and agents that have embodied the class pattern and represent it in actual state and process.

This idea of an instance interacting via a public interface which encapsulates its internal workings is called abstraction. The class defines the interface and internal structure that its instance will follow. A class is conceptual whereas the instances are actual (actualised in time), and we say it's an abstraction of its instances.

Class and instance are extremely fundamental concepts, because they define the actual processes behind behaviours, patterns, encapsulation and abstraction, actually implementing those concepts and bringing them into being. It's the functionality the fluid nature of a holon's identity to work in the way outlined by the Ship of Theseus example above.

Class mix-ins

Classes need to be composable, they need to be able to be combined into new combinations. Different OOP paradigms use different approaches for how composability is achieved. One method called "class mixins" allows classes to be instantiated into the context of existing instances. This matches the holon context well because it's exactly the same idea as sets of productions rules operating together in the same local scope in accord with the blackboard pattern.

An organisation is a whole structure of mixin-instances that are activating in schedules and in accord with present conditions. This is the holon's self-organisational structure or "self-representation", it's a mosaic of instances of various classes that can be organised in dynamic ways that match the local circumstances and preferences. In other words a specific sub-set of the possible expression space defined by the class.

A good example of this type of dynamic class-instance relationship and structure is a live streaming music mix channel. This channel consists of a structured schedule of music themes as well as potential spontaneous or quasi-random aspects. The content of the channel is composed of mixes and remixes of existing classes from the evolving establishment.

Within the mosaic are many structured instances that operate in accord with clearly defined behaviour structures and present themselves in the form of clearly defined interfaces. These knowledge structures specialise and evolve through establishment in usage within all the local instance structure mosaics.


Evolution can be boiled down to an extremely simple dynamic in its general form. David Deutsch describes it as "the creation of knowledge through alternating variation and selection". Note that we're talking about the general principle of evolution here, not specifically biological evolution.

Human culture is evolutionary knowledge. It depends on, builds on, and consists of, other knowledge, and is always evolving in diversity and complexity. Knowledge and evolution go hand-in-hand, they're interdependent concepts.

The complexity we see in evolutionary systems (such as biological evolution) is due to the evolutionary dynamic itself, which tends towards ever more diversity and complexity. But the underlying dynamic responsible for all this complexity remains simple and unchanged.

The nature of knowledge is to evolve in diversity and complexity. It's not just inert information, it's a dynamic process involving subjective values and application within diverse conditions. Our genes, our culture, our society and our own minds are all structures of evolutionary knowledge, even though their media and selection mechanisms differ.

The evolutionary knowledge principle actually incorporates the class-instance concept within it. The evolutionary dynamic is an extension of the basic class-instance concept.

If we think about some actual examples of class-instance systems in our daily lives such as a market ecosystem of producers, vendors and consumers or software version control systems and their ecosystems like Github we see that they always have a community ecosystem side and a local usage side. We always find that the ecosystem evolves and the local uses specialise.

The holon model incorporates both of these sides with the evolutionary principle in the form of an extension to the basic class-instance concept.

The creation of variations and their selection correspond to the integrative and self-assertive behaviours respectively, as well as to the classification and instantiation respectively, of the holon. The integrative side is the improvement of evolutionary knowledge in terms of its usability and the potential gained by those who use it, and the self-assertive side permits the improvement of our own position through selection in accord with our own needs and preferences.

Summary of the form of knowledge and patterns

Let's summarise the concept we've described in these prior sections on the knowledge and patterns of the holon. It's a class in instance system in the object-oriented sense, where the instance form a mosaic of instantiated classes matching local circumstances and preferences. This concept gives us a general description of the aspects needed to replicate the evolving "idea-clusters" that the Ship of Theseus drew our attention to.

The class-instance system incorporates the collective aspect that represents the many local instances and the market of real resource. Both the class and instance spaces as a whole collective and individually are evolving and continuously improving and specialising.

The instances are holon-agents having subjective perspectives and local private continuous threads of activity and state. The classes present together in these local scopes are all continuous behaviours and are composable into useful combinations operating asynchronously together just like memes. They all operate locally together to maintain a self-organisation structure or self-representation. In this way we have a community of holons all developing as classes and operating as instances.

So far what we have is a clear idea of a concept that we can visualise, but we don't have any detail about how it might be implemented, it's like a set of nice-to-have features for the system. In the next sections we'll discuss this class-instance evolutionary concept more specifically as an implementable system model. The model takes the form of three abstraction layers and has four distinct aspects to its behaviour.

Three abstraction layers

An abstraction layer is a conceptual framework or set of functions that hides the complexities of lower-level operations, allowing users to interact with a system or software component in a simplified and standardised manner. It serves as a bridge between different levels of a system, enabling efficient communication and interaction while shielding users from the underlying technical details. Abstraction layers are commonly employed in software development to promote modularity, scalability, and ease of use.

The internet itself is organised into abstraction layers with various communications protocols which we call the internet protocol "stack". The first and most fundamental layer's protocols govern the way information is communicated in the physical cables, occupying middle layers are protocols governing things like IP addresses, domain names and encrypted connections, and the top layers are high-level application protocols concerning things like social networking and voice over IP (VOIP).

In a running system, each layer can be seen from an instance perspective as being a society of instances that all interact together via a common set of interfaces. Interfaces are "provided" and "used" just like the client-server model, for example how Amazon provides an interface to the market and Uber provide an interface to transport. In terms of the internet protocol stack, we might look at the clients of the IP layer as all the connected devices and users having their individual IP addresses, or the clients of the social networking layer being all the social identities that can present themselves and interact via the a particular social networking protocol (the so-called Fediverse is all the identities who can be interacted with via the ActivityPub protocol (and other open social protocols).

In the running system, these layers are logically independent (but not existentially independent), each layer defines interfaces that hide its own complexity from the new layer within. The instances (of interface usage) composing a running layer are free to collaborate fully on the content and evolution of their "user space" independently, higher layers set the general rules (protocol) for lower layers.


Our model has three abstraction layers. The the first (L1), is the process that results in the class-and-instance environment, within which the second abstraction layer operates. In our model this layer represented as vertical and horizontal axes, which is represented in the diagram by the blue "plus" to the right.

The vertical is considered the primary of the two, and represents the instance concept. The horizontal is secondary and derived from the first and represents the class concept. Together they produce the class-instance environment for the next layer to build upon and extend.

The second layer (L2) is the "subjective" perspective from inside the instance's private scope. In other words it's a new abstraction layer occupying the "user space" defined by the first layer, the usage of the class-and-instance concept.[8]

The second layer is where the actual functionality of four quadrants (by which the model itself is usually referred to) is defined. The quadrant processes take the form of feedback loops between diagonal opposites, and we represent this in the diagram by the green cross to the right. These diagonals represent the fundamental variation and selection aspects of the evolutionary system, or in other words, the actual usage of the class and instance concepts.[9]. The four quadrants individually are like independent departments within an operating organisational instance all aligned in their overall purpose.

The third layer (L3) is the space of fully functioning holons (self-organisations) forming a structure of arbitrarily complex meaning. This new layer occupies the "user space" defined by the second layer, i.e. the quadrants that were functionally created in layer two are utilised in this layer. In this environment, all content self-organises and progresses as an evolving society of organisations, a holarchy of holons. Holons in layer three are the users of layer two's quadrant system using the interfaces provided by the individual quadrants.

The third layer is an independent self-organising self-evolving society of holons. This is shown as the purpose circle to the right in the diagram. All holons are self-sovereign first-class citizens, completely independent and autonomous, but at the same time they all holons inherently represent the first two layers in their behaviour.

Layer 1: The primary axes


In terms of functionality, the first layer brings about the class and instance environment discussed above. This is essentially the layer that defines the process of abstraction itself by which subsequent layers are possible.

But in this section, we're concerned with the model (conceptual meaning) rather the functionality. The mechanism that defines the functionality of the first layer is beyond the scope of this article, and so we're being deliberately vague about many points in this section. The reasons why these primary axes take on the meanings described here is covered in detail in the holon mechanism article.

The mechanism results in a number of important fundamental conceptual meanings which also form the most general characteristics for subsequent layers. These concepts are represented as the primary (vertical and horizontal) axis pair, which are shown in the image to the right, and are also depicted as the blue "+" in the diagram of layers above.

The four quadrant system informs and responds to change, but is not the ultimate actualisor of it.[10] The system does not define change itself, it only organises it ontologically to be utilised by the actual agents of change. In terms of the diagram, the change occurs in the centre as an action representing the current class and instance.

Both class and instance concepts take the form of a scope (namespace) concept with the positive end representing being not within the scope, and the negative side being within it.[11][12]

The first kind of scope is the usual public/private vertical dimension that we're used to with an object from OOP, these are instance scope forming the instance tree. The second kind of scope, which is complimentary in its operation to the first, are class scope making up the class tree.

Within this primary axis pair, the instance tree is the primary or original axis and the class tree is derived from it. Even though instances are instantiated from and guided by their classes, they depend entirely on the instances to represent them, because only the instance actually exist by being backed by real resource.

The top is public, the bottom is private, the left is abstract and the right is actual. Each primary direction defines a meaning that's common to a pair of quadrants. In terms of functionality this layer creates the scopes and a feedback loop dynamic, but it does not actually do anything within these scopes in terms of creating or responding to change - that's where the second abstraction layer comes into play. We won't go into any more detail about the first layer in this article, the mechanism that brings about the scopes is discussed in more detail in the holon mechanism article.

Layer 2: The four quadrants

The four quadrants occupy the second abstraction layer of the model. As discussed above, the first layer primary axes define the most general contextual features for the four quadrants - what scopes they operate within, and the meanings that the upper, lower, left and right sides have.

The first layer made possible a new subjective local perspective, and the second layer is halfway between these two perspectives, having "a foot in each side". The lower quadrants represent the inner local subjective perspective, and the upper quadrants represent the outer non-local perspective.

We often refer to layer two as the "objective-subjective", because it's an objective "unconscious" process like layer one, but it occurs in the local subjective scope. The third layer also takes place in this subjective scope, but all change is carried out by agency in the third layer.

The inherent form of the quadrants is that they're grouped into a pair of feedback loops connecting diagonally opposite quadrants. These are the variation loop and the selection loop constituting the evolutionary system. We use the word "inherent" because the information flow that defines these diagonal feedback loops between opposite quadrants are created by the first layer mechanism. The mechanism itself is beyond the scope of this article, what we cover herein is the meaning of these scopes and loops.


Before we go into any detail about the diagonals, we need to have a clear conceptual understanding of the individual quadrants. The easiest way to introduce the quadrants is to start with the already-familiar class and instance concepts on the left and the right respectively, and then divide them into an upper collectivised version of the pair and an individuated version below. The image to the right demonstrates this with the original class-instance axis horizontally in the middle.

The top quadrants represent the local holon's perception of, and contribution to, the whole tree (graph) of classes and instances, which we call "ontology" and "market" respectively. Since it's a bottom-up peer-to-peer architecture, these collective-oriented top quadrants are not the whole itself (which would have to be "centrally served"), they're a local representation of the whole from the local subjective perspective with self at the centre.

The bottom quadrants represent the local holon's internal private world. This lower pair is conceptually more fine-grained than the general (and abstract) class-and-instance concept represented by the horizontal axis. They represent the local subjective meaning of the class and instance dynamic. Classes are designed to be instances, their utility and purpose comes from how they behave within their subjective instantiated contexts. The internal class quadrant in the bottom-left is called "development" and it takes the form of conditional structure (the condition aspect of the production rule structure). The internal instance quadrant in the bottom-right is called "production" and represents the holon as a progressing activity (the action aspect of the production rules).

Each of the quadrants is delineated by the vertical and horizontal axes of the first layer discussed above. This means they each represent a pair of scopes, one from each primary axis. This gives us a clear foundation from which to derive the meaning and process for each quadrant that forms its concept of progress.

Since the processes are operating on the same state (all being aspects of the same holon), they must be complimentary and non-destructive to each other. But as we've described, the de-coupled production rule and blackboard model gives us exactly the non-destructive process-form we need here.

Why the quadrants?

The four quadrants are usually only discussed in the context of philosophy, and so it can be confusing as to why we give them so much attention when we're in an information engineering context not a philosophical one. But due to the holarchy being a very ontologically fundamental, we obliged to take a specific philosophical position. Being an agency or experiential oriented system places it firmly in the idealistic camp, but also it's a definite informational system which is strongly materialistic.

Agent-oriented models sit somewhere in the middle and might be best identified with something like computational realism. In this context, the four quadrants are not just a convenient "lens" through which to analyse the experienced world, but rather are the form of the actual processes permitting experience itself.

The quadrants are "real" in the sense that in a running holon, each has a specific executional thread representing it, so that each receives it's own portion of the total executional focus available to that holon. Each quadrant is an important and permanent aspect of the holon as a whole, very much like a department in an organisation.

The holon as a whole represents structured state and its path forward through time, and each quadrant represents a different perspective on what "progress" means and how it expresses that form of progress with its local behaviours.

As discussed above, the foundation use-case of the holon is as a self-organisation system which was described as being a kind of "smart folder structure" that represents our lives informationally and also acts as an organisational interface to them.

The four quadrants are four different aspects of the "self-organisation application", the bottom two are the familiar for of class structure and actualised instance structure that we're used to, and which we would expect of the "smart folder structure".

But we're not just individuals, our lives take place in the collective context of culture and society. The top two quadrants represent the class and instance aspects in their "collectivised" forms of ontology and economy respectively connecting us with and present wider contexts of knowledge and the market ecosystem.

Introducing each quadrant

The second abstraction layer of the holon model is all about the function and dynamics of the quadrants, it doesn't concern any user-facing aspects of the quadrants. We'll look at each of the quadrants again from the user perspective when we discuss the third abstraction layer of the model. Following a brief introduction to each of the four quadrants in terms of the functionality and meaning.

Top-left (ontology)

This quadrant is the "collectivised" version of the class concept. It's a left quadrant, which means that it concerns abstract knowledge which is not actualised in time. It's also a top quadrant putting it in the public scope, which means it's a peer-to-peer collective contribution process. This quadrant is called "culture" in Integral Theory, and it's Aristotle's "formal cause", which is often described as a "blueprint".

This quadrant takes the form of an ontology of classes connected in a semantic network of dependence and relevance (a class-tree). These grouping (dependency) relations as a whole form a large associative network. But from the perspective of any specific node, there is a "fan-out", a one-to-many hierarchy of dependent child nodes, and grand-children etc to any arbitrary depth. These hierarchical structures determine the form of instances. The ontology is the map of the ecosystem of behaviours established in usage.

Knowledge is not just dead information, it needs to be embodied behaviourally. It applies to a group within which it's established in collective usage. The variational aspect of the evolutionary principle is essentially about sharing aggregated performance information associated with the conditions, i.e. the objective and circumstance requiring the activity. This is how the ontology represents usable collective knowledge from classes established in usage.

All instances of like classes form into knowledge-sharing groups. In this way, every class in the ontology (global class graph) is a community and a map of all the instances of that class.

The knowledge is naturally shareable and understandable, because the group of all instances of one class are essentially a special-interest group - they all have interest in the same specialist knowledge associated with that specific class.

The purpose of knowledge is to be used. To use it requires it to be embodied by a holon, in the form of classes that are "installed" (connected into paths of potential focus) into the local environment where they can activate it (in the bottom-right quadrant) in response to appropriate local conditions as they arise. Knowledge is not just opinion, it's determined by how effectively it's used. For the ontology to assure utility, it must include this performance aspect with the knowledge and the performers of it.

The ontology is structured by class names, and contains information about how those classes perform as children filling roles in various classes of organisation. The result is an ontology of behaviours associated with actual ability to perform them. These are the abilities that back objectives making them actualisable (by instantiation, making them potential and then imminent).

Although this is the non-local (collectivised) ontology, it's important to remember that even though it's a view of the collective, it's still from the local perspective of an individual local holon. It's the non-local unified ontology, but from the perspective rooted in the current class (which is performing the current activity).

The ontology evolves in diversity and complexity as the instances develop themselves (bottom-left) and share their usage knowledge (bottom-right). It's a collective form of progress which is evolutionary in nature, not a self-assertive control loop.

Bottom-right (production)

We call this the production quadrant which takes the form of a self-assertive control-loop maintaining the private self-representation since it's a bottom quadrant. Since it's on the right, it's actualised in time involving concrete resource. This is Integral Theory's "behavioural" quadrant and Aristotle's "efficient cause" which is the agent that brings something into being.

In terms of the self-organisation, this quadrant is responsible for maintaining the local instance of the organisation operating in the local environment. It carries out the process that fits the local self-representation to the real state it represents, and allows it to act as an interface to it.

This quadrant involves the actual achievement of the holon's objectives (in the bottom-left). Production is a control-loop that reduces the difference between the current resource state and the expected state.

An actual agent has filled a role in the local context and performed behaviours towards achieving the various objectives. The holon has gained "experience" by putting its knowledge to use in service of the holons own private developing objectives in the bottom-left.

The holon's internal production and development are in the form of structured production rules, and the "production" quadrant is the action/activity side of the rules.

In this aspect of the organisation, we're in the private scope of production using private property.[13] Production takes the form of a control loop that continuously moves towards goals set by the bottom-left development (intention) quadrant.

This quadrant is all about past, because it's about the accounting of an activity after it's been performed. The final account on completion is signed and immutable and contributes to the ontology (top-left) which has the current condition at it's root. The information contributed to the ontology is the performance, the account compared to the initial expectation, in the context of the condition (parent) that it's responding to. In this way the local knowledge is contributed to where it's relevant.

The production quadrant represents the actual state of production of the holon, such as materials, access, stock, accounts etc including the state of completeness if applicable. This quadrant represents the actual performance of behaviour informationally which we call "accounts" (i.e. an accounts of activities). Performance of behaviour is carried out in accord with the top-left "ontology" quadrant, and final performance with respect to expectations is presented to the ontology for aggregation. The results of production contribute to the objectives in the bottom-left quadrant, as well as providing and consuming resources via the top-right quadrant.

Bottom-left (development)

This is a bottom quadrant so, like the bottom-right, it takes the form of a self-assertive control-loop in private scope within the holon. It's on the left so it concerns abstract knowledge that is not actualised in time. Unlike the bottom-right control-loop, this quadrant concerns knowledge rather than resource.

This is the "development" quadrant, and is called the "intentional" quadrant in Integral Theory (we often refer to it by that name as well). It's Aristotle's "final cause" or telos, the objective or purpose for which something is done or exists. For example, the telos of a knife would be to cut.

In the quadrant, the holon is navigating in potential space, developing its knowledge, intentions and objectives. This is the structural aspect of the private self-representation, its embodied pattern of behaviour. It represents the meaning of the organisation that informs development decisions.

The objectives are defined by the condition-side of the internal production rule structure. A specific condition arising makes a subset of actions and variants salient, to be refined, selected and acted upon in the right-hand production quadrant.

Objectives concern the future and so this quadrant works in conjunction with the top-right resource-flow (economy) quadrant which is also about the future.

The development quadrant represents the holon's objectives in the form of a structure of embodied behaviours. This structure represents the embodied and salient aspect of the holon class behaviours from the ontology (top-left), backed by resource from the market (top-right) and attained by production (bottom-right).

Top-right (economy)

This quadrant is at the top so it's a collective contribution in public scope, and being on the right it's within the context of actualised linear time. We call this quadrant "economy", because its purpose is to harmoniously allocate limited resource amongst a potentially unlimited demand for resource. This is Integral Theory's "society" quadrant, and Aristotle's "material cause".

It seems at first glance that connecting the meanings of "economy", "society" and "material cause" across these systems is contrived to fit our designs. But remember that we're in the agent-centric organisational context of a holon, where "external material reality" is in purely resource-flow terms, and is the merging of all local perspectives. This quadrant represents the local state of resource-flow expressed in supply-demand terms by the instances occupying the public scope. The resource-flow effectively represents the total of all committed objectives. That is, the intentions expressed in all the bottom-left quadrants throughout the holarchy.

Essentially the economy concept describes a free market based resource allocation system used and supported by a network of autonomous participants. These entities have the autonomy to choose what goods or services to produce or consume, at what price, and from whom.

In terms of the self-organisation, this quadrant represents the interface between the public and private sides of the holon. Its organised by linear time in the future and so from the user perspective it takes the form of a schedule. The schedule is an organisational "container" in which roles and resources are "booked" by instances that fill the roles.

A holon requires real resource in order to function. In other words, the self-organisation structure represented by the bottom-left quadrant needs resource organised by the top-right quadrant to represent it. Such a representation exhibits expectation and after it's completed will exhibit performance with respect to it.

Objectives are defined in terms of external resource state in the future, and so this quadrant inherently relates the bottom-left development quadrant. This relationship takes the form of a feedback loop which we'll come back to below.

This quadrant allows the holon to participate in the wider market, all together forming the economy (the society of organisations).


The whole must assure (prove, demonstrate) that it effectively maximises the harmony, autonomy and potential for both the individuals and the whole. If it doesn't, then it's not truly worthy of their membership. The whole relies for its very existence on the support of its members, so its effectiveness and the evidence for it is the foundation of its own security.

The collective aspects are abstract, emerging from the many participating as network nodes. but yet it's this collective aspect that provides the assurances that are really the sole reason for participating. The reason that participants choose to participate is because the holarchy offers assured benefits. It offers usable and reliable knowledge in the form of the ontology and offers opportunity and a harmonious environment in the form of the economy. The knowledge needs to be usable and reliable, in other words it needs to provide assurances of its utility.

The holons are all contributing to a global state of institutional predictability,[14] which concerns a stable operating environment in which plans can be made. The assurances come from the fact that the protocol itself objectively and unconditionally includes the integrative behaviour.

With assurances of stable operation comes the possibility of expectations through the accumulation of knowledge about operation, and from expectations we can assess performance.

The diagonals (⤫)


As is quite intuitive and can be seen in the diagram to the right, the quadrants naturally form a diagonal pair of axes. But the diagonals also represent the actual algorithmic/mechanistic connections between the quadrants too. We won't go into the details of the mechanism behind the formation of the four quadrants in this article, but the diagonals are one specific consequence of this mechanism.

There is an important conceptual reason for the diagonal connections as well which is that classes are made specifically to perform in local subjective contexts as instances, and the basis of all classes is a feedback loop that orients the performance towards the form it defines (so that the state and development gravitate around the evolving form of the concept like the Ship of Theseus). What this means is that collective class meaning is defined in relation to individual instance, knowledge is defined in relation to performance, which is the top-left to bottom-right diagonal.

On the other hand, class structure requires real resource backing, and so its collectivisation is the instance world. The instance collective structure is defined in relation to salient class structure, which is a connection between the bottom-left and top-right quadrants.

Within the world of the actual functioning holon developing and operating in its local context, the control loops are both connected to their opposite collective state, and this state is backed by a collectivisation process. The second abstraction layer of the holon takes the form of diagonal feedback loops. each composed of two loops connected together inputs to the outputs of their partner. The upper end is a collective loop and the lower end is an individual control loop.

The functionality of the quadrants takes the form of a pair of feedback loops connecting the diagonally opposite partners.[15]

These two diagonal loops constitute dynamics of second abstraction layer of the model that refine the four quadrants behaviours and connect them all together into a harmonious whole.

The diagonals are the form of the "application" presented by level two for use by level three. Just as the class-and-instance mechanism was the level one provided for level two's use.

The diagonal relationships are due to the form of the diagonally opposite quadrants being complimentary so that each diagonal forms a feedback loop. The quadrants meanings derive from these feedback loops, and so in this section we look in more detail at these two loops.

The four quadrants occupy the corner areas delineated by the primary axes (vertical and horizontal), and so all quadrants are situated on the diagonal axes as shown in the image to the right. These axes represent the actual connections giving rise to the function of a holon.

The bottom two quadrants represent the familiar self-oriented organisational context. These each connect to their opposite outward partner, the bottom-left connects to the top-right forming the selection loop, and the bottom-right connects to the top-left forming the variational loop. The former extends the instance-tree to include the evolutionary concept of selection to become a "multiplex of intention". The latter extends the class-tree to become an ontology of variations of knowledge in use.

Each loop is a distinct way the collective forms from the individual behaviour, and conversely how the individual is guided by the collective. Each is a co-evolutionary loop just like the first layer class-instance loop is.

Both loops are derived from and extend the primary feedback loop dynamic form into a new concept involving knowledge derived from the local internal scope. One diagonal extending the instance-tree and the other the class-tree.

In each loop-extension there is a rating (evaluation, feedback) of the associated tree involved. The selection loop involves a subjective rating in accord with local intentions and preferences, and the variational loop involves the objective rating of local productive performance and use. Both loops involve local rating and non-local merging of the rating information. In both loops, local decision-making is guided by the non-local aggregate information.

Both loops also have an "inertial" aspect where they're "installed" into the local environment as ongoing behaviours. They become easier to perform and continue over time and take energy to stop ("uninstall").

The collective can be thought of as a "service provider" (albeit a non-local peer-to-peer one) that evolves with the clients needs, and the individual (as the client) is guided by and uses the service. The ontology is a service utilised by an agent in production, and the economy is a service utilised by a consumer.

Naming the diagonals

The naming of the loops has been difficult and has changed a lot (class/instance, agency/expression, evolution/economic, past/future, energy/material, activity/resource, variation/selection, identity/structure). Class/instance may be the best, but it's confusing to have the same terms as layer one, something that sums up the meaning of their use to become the evolutionary system seems more apt.

The names of "variation" and "selection" are chosen since they're the two main features of the most basic definition of evolution as alternating variation and selection. Even though both diagonals' descriptions involve both variation and selection, when we think of the actual instantiated form of each diagonal there's a clear correspondence. Local production clearly aggregates to form the class tree of structure mapping unique variations, and local development clearly aggregates to form the instance tree in the form of a mosaic of intention (salience, selection). Also the second abstraction layer concerns the extension from basic class-instance to evolutionary class-instance, so it makes sense for the layer two naming to concern the nature of the layer two extension.

Selection loop (⤢)

The diagonal consisting of the bottom-left and top-right quadrant (⤢) forms the selection loop and is associated with the self-assertive behaviour, the instance tree and the future. It extends the first layer instance tree (represented by the primary vertical axis) which is inherently economic in nature due to representing the flow of real resource and attention through time, this loop represents the process of instantiation.

The bottom-left quadrant represents the self-organisational structure, which is a structure of recurring behaviours. The top-right quadrant represents the schedule of committed resource that backs the performance of these behaviours.

The diagonal axis of the selection loop extends the first layer instance tree from a purely attentional flow to a more refined concept that includes the aggregate of local market knowledge coming from subjective value judgements and decision-making.

This axis represents the holons presenting itself in its self-assertive form in the public market. In other words, its public state as an autonomous self-organisation in the public market. This self-assertive expression of economic commitment, is the form that the aforementioned subjective rating takes. This is the subjective evaluation of instance, and expression of that evaluation through attentional (and resource) support. The directed support is how objectives are determined, the selection loop is results driven (declarative) and focused on the future.

This axis represents the holon as a sub-class group from above interacting together as an organised structure of loops from past in the bottom-left to future (schedule) in the top-right. From above. this group is seem as a collection of public interfaces. This is the mosaic of class-mixin instances referred to earlier.

The foundation of selection is the flow of attention, which is the salience landscape, the distribution of weights that determine the flow of focus throughout the instance structure. Salience is distributed internally (bottom-left) as the tentative virtual branches extending from what's represented in the resource flow (top-right).

From the user (self-organisation) perspective this diagonal represents the market interface. The organisational structure can publicly present supply and demand schedules of various resources. The holon presents various consumer and producer interfaces and states publicly. This is how commitments are made that permit actual production, and all together make up the whole resource-flow.

The selection loop is an organisational structure spanning internal behaviours as well as resource schedules. Salience is distributed across the structure, and directing this distribution over time is self-development. Organisational structure and its salience are the common form of the bottom-left and top-right quadrants.

The loop is a bidirectional instance-tree process of interaction between internal virtual instantiation (exploring a concept) and the public market of actual resources and value. The private virtual content is essentially a "replaying" and "remixing" mosaic of instances from the public arena.

The public content (the the flow of resource as a whole) which is the total of all the internal virtual instances in the whole network that have become backed by real resource (through persistent salience). In other words, a context starts as a purely abstract concept that can be explored and gain more focus and resource, becoming booked into public resource schedules.

The top-down side of the selection loop is the flow of focus and resource that determines which instances and conditions are active (selected). The feedback flowing from the bottom up is intention, or subjective valuation of the context.

Variational loop (⤡)

The diagonal consisting of the top-left and bottom-right quadrants (⤡) forms the variational loop and is associated with the integrative behaviour, the class tree and the past. It extends the first layer class tree which is inherently ontological in nature due to representing the dependency and relevance relationships between classes (classification process).

The top-left is the ontology of behaviours, and the bottom-right is the actual performance, or usage of them in the local private production context. Both ends of the variational loop concern the execution (imperative) aspect of the system in terms of utility and performance.

The ontology in its basic form is created in the first layer, based on volume of usage. Then in the second layer it's extended to include the performance metrics corresponding to the specific performers of the behaviours. This process expresses the principle that knowledge is not black and white, it's embodiment is proven and assessed through actual performance.

In this subjective inner context, the information being aggregated is the performance of the knowledge in-use internally. This aggregate knowledge is the performing-instance's "reputation" or potential effecting it's likelihood of being matched in the market again in the future.

In user or self-organisation terms, this diagonal represents the usage of the knowledge in the operation of the organisational structure. This diagonal represents the execution in the present on the bottom-right, and the establishment formed by all execution in the present throughout the network, i.e. what is established in usage.

The variational loop is a bidirectional class-tree process of interaction between internal usage and execution of a behaviour and the institutional map of knowledge relating to the behaviour.

The top-down side of the variational loop is the institutional knowledge and guidance (map) flowing inward from the collective class. The bottom-up feedback side is the objective performance (of the embodied knowledge) and usage statistics of local production in aggregate form.

Intersection of variation and selection

The selection loop is primary, and the variational loop is in the context set by it. This is because the former is extended from the first layer instance tree and the latter from the first layer class tree. And as a whole, the first layer defines the functionality of the whole class and instance mechanism, and the second layer is the instantiated space of instances (users) of the layer one class-instance mechanism.[16]

First the selection loop activates, the internal intentions are created in the bottom-left, and then are enabled by the backing of real resource in the top-right. The merging of the commitment schedules of many holons, creates potential for resource exchange leading to the whole resource-flow.

Then the variational loop activates, the selected instances perform in private production in the bottom-right. All other things being equal amongst candidate suppliers (performers) are prioritised according to their performance metrics. Actual performance takes place in the bottom-right and is fed back to refine the knowledge in the top-left.

This matching occurs diagrammatically in the centre where the two diagonal axes cross. This intersection process is a subjective value judgement. This is where the process that matches supply to demand, resulting in committed (contracts) resource backing the objectives over time takes place. The selection is based on the evaluation of the class and instance aspects. In other words how well the class is achieving its fundamental purpose of instantiation, and how well the instance is bringing value to the local context.

The selection loop concerns organising schedules and so is related to the future, while the variational loop concerns the behaviours established in usage and their performance and so relates to the past. The market match is the intersection of the future and the past.[17]

The first level provided the basic subjective context in which conceptual structure has a directional aspect. This leads to the self being at the centre in the here and now, with its subjective perspective of the world surrounding it. The second level extends the subjective context to include an inherent ability to assess things in terms of potential and performance, so that both the class and instance structures are continuously improving in their fundamental utility.

*   *   *

The diagonal loops (and the individual quadrants they're constituted from) have specific meanings that derive from their extension of the general meaning provided by the first abstraction layer, each progressing the holon in their own specific yet complimentary way as well as the class and instance structures as a whole continuously improving and evolving. The result is a holon which embodies a rich set of general organisational behaviours; it participates in the evolution of knowledge and the economy as well as progressing its individual knowledge and material position.

To expand on a key concept in the previous paragraph: the economic and evolutionary foundations present in the two diagonals have not been deliberately designed, they're inherently provided by the first layer structure. All we've done is naturally extend the first layer with its own dynamics in the subjective second layer and the diagonals naturally take on these meaningful dynamics.

Although the second layer concepts such as evolution, knowledge, expectation etc are very conceptually rich and complex compared to the first layer, they are a very basic version of the systems they represent (evolution and economy). The evolutionary aspect of the system boils down to the distribution and management of variations and selections, and the economic aspect is a simple free market dynamic involving supply and demand commitments and schedules. They form a neutral conceptual foundation on top of which larger objectives and methodologies can be expressed.

The holon model acts as an "ontological wrapper" allowing any information, knowledge, systems or resources to be interacted with in universally understandable meaningful terms. Ontologically representing all the common organisational aspects of it such as the time period it covers, it's state of completion, behavioural or performance aspects, purpose or value.

Layer 3: Holonic self-organisation

The third abstraction layer in the system is the organisational environment - a self organising network of self-organisations. Every node making up the network of content in the third layer is a complete holon, and a first-class citizen. The third layer represents the user perspective since it's the layer representing interactions involving complete individuals. This layer is a society of organisations in which they all represent themselves as self-organisation structures all having the two loops and four quadrant aspects.

From the user's perspective, the top quadrants are seen as the public interface through which the local holon interacts with the collective. These top quadrants are like services provided by the collective (although the collective is the collaborative aggregate of all individual holons). The top-left is the "map" interface to the ontology as a service, and the top-right is the "schedule" interface for interacting with the flow of real resource.

The bottom quadrants revolve around our self-representation, the abstract ontological structure and informational state of our self-organisation. The structure changes through the holons development in the bottom-left quadrant, and the representation is kept up to date (fitted to reality) and progressed towards objectives by the production process in the bottom-right.

The quadrants in the third layer

As described above, the quadrants are scopes (fields of activity) formed from definite informational connections and processes, but in this third abstraction layer we're in the context of fully operational holons, so the quadrants are to be understood in their "organisational department" forms.

Interaction in the context of organisational departments takes the form of organisational roles. Each of the quadrants departments has two clear roles operating within it, an outward facing role and an inward facing role. Following is a brief summary of the roles that match each of the quadrants.

It should be noted that these roles may be divided into further sub-roles within depending on the complexity of the organisation. And also that these roles are deliberately designed to be general and to fit with the four-quadrant structure, because the intention is that they are to be thought of as kind of intermediary "virtual roles" that will be filled by AI. We'll talk more about AI's roles in the organisation further on.

Top-left (ontology)

We usually call this quadrant the institutional quadrant when we're in the third layer context, due to it being a source of guidance for instances to follow. The institutional aspect is the organisational structure, or public map, that forms around the aggregated collective intelligence making it navigable and accessible to all. It's essentially an informational portal maintained by the users of the knowledge, or in other words, a peer-to-peer institution.

Curator: The outward-facing role is best described as a curator, a role that organises and categorises the ontology ensuring that our own local information and structure is coherent. The role is outward-facing because the coherence is a global collaborative affair requiring alignment with all local perspectives.

Advisor: The inward-facing role is about ensuring that the local organisation is using the knowledge of the ontology to the best effect. This is a role who has very good general knowledge and is very familiar with the ontology as a map of knowledge that can potentially be of use to the organisation in any circumstance.

Top-right (economy)

A good way to explain the top-left quadrant as a high-level user oriented organisational pattern is with the WWWW acronym, which stands for "who, what, where, when". It's essentially the interface for booking meetings during which specific people, agents, tools and resources will be present in the same context.

Commercial affairs: The outward-facing role concerns all interactions with the market place, dealing with clients and suppliers to exchange value. This role in all the holons in the network leads to the overall flow of resource throughout society.

Public relations: The inward-facing role is all about maintaining the organisation's "self-image". In the case of an individual this would be the social profile. It's also about public expression of value (expressing its purpose in the context its operating in), positions and intents such as a charter, affiliations and undertakings etc.

Bottom-right (production)

This is probably the most intuitively understandable of the quadrants, because it involves the normal day-to-day operations of the holon, moving the material state towards the targets required by the holon's objectives.

This is the actual activity that takes place in the aforementioned meetings of specific roles and resources. Together they utilise their abilities and the resources allocated to the local scheduled slot which is their common "blackboard", operating together in accord with the Blackboard pattern in the local private scope.

The Operator

Operator: This is the inward-facing role. Think of the "operator" role in the Matrix movie or the Bennie (Jasper Carrot) role in the Mission Impossible movies. It's a communications oriented role that keeps the whole team in touch with what's going on "in the field" (within the specific local "blackboard" scope in question).

Accountant: This is the outward-facing role which is all about reporting on the progress and performance of the operation. It may seem a little odd that this is an outward facing role because the organisations accounts are obviously private, but in the holarchy much of this role is about aggregate information such as performance and usage statistics which are shared publicly (non-locally in class-scope) to increase the utility of the ontology.

Bottom-left (development)

This quadrant represents the holon's self-organisation structure, a specific structure of behaviours (production rules) defined to achieve specific local objectives. In user terms, this represents the meaning, intentions and objectives of the organisation. The term "development" here refers to self-development not software development.

Manager: The inward-facing role is about managing the team to best achieve the organisations objectives. This involves helping the team to better operate together (including initial onboarding and change management) and the organisation as a whole to progress and develop.

Director: The outward-facing role is about the organisations direction, deciding where it's going, what its intentions and objectives should ideally be. It's an outward-facing role because it's about navigating the organisation through the external environment of circumstances.

Virtual instantiation

The common organisational context also comes inherently with the ability to assess variations of the current organisational structure, which is the process of self-development and management of potential. This can also be applied to any ideas, concepts of scenarios we see in the society or even from our own pasts, can be "replayed" and "remixed" virtually. This is essentially a form of "organisational imagination" which we call virtual instantiation. It's a dynamic mosaic of instances formed from subjective valuation.

Instantiation is virtual when there are no real resources backing an instance, instead its operating environment is provided synthetically from knowledge accumulated in the classes. This is like a simulation of the instance which matches historical activity and usage statistics.

Actual resources are connected to a part of the representation that acts like a local index of the data so that it can be part of the organisation. The agency which is responsible for maintaining this index has been delegated down to something simple like a Python function. And so the same agency that made this delegation (translated its own imperatives into Python) can just as easily make a function that provides random data that matches the real metrics.

In this way any instantiation can be tested before using it to interact with real resource and contacts. Virtual instantiation can apply to small changes to an organisation as well simply by having a new instantiation that's a clone of the organisation, but some aspects of the clone are changed, so we can observe them for a while before deploying the change in the live organisation (like a commit in software development, or standardisation in a continuous improvement loop).

Virtual instantiation is the organisational or OO equivalent of imagination, and is an essential prerequisite for adaptation. Virtual instantiation is the process by which variations are formed which are the source of evolutionary change. Even the progression from abstraction to production (concretisation) relies on virtual instantiation, because all instantiation starts virtually.

Continuous improvement

In addition to the quadrants, Integral Theory also involves developmental lines and stages.[18] Lines correspond roughly to the threads in our system, or in terms of production could be thought of as a holon's "product lines". Each of these lines follows the same general pattern of developing in discrete stages that involve interaction from all the quadrants.

We can think of the quadrants as discrete phases common to each developmental stage. Each quadrant has a loose causal connection with the next one forming a clockwise loop. Work is organised and booked in the top-right, performed in the bottom-right, adapted and developed in the bottom-left and the learned knowledge shared in the top-left which then leads to new work in the top-right again, but on a more evolved, complex and diverse level.

The form of this pattern is a spiral, each revolution arrives back at the same point but on a higher order of development. Each new level is like a platform supporting the next level, which leads to a kind of continuous improvement "ratchet" mechanism which permits development to ever higher levels, but prevents regression back to prior levels due to each new level becoming firmly established in the collective.

This is a very high-level view of the holon, because the quadrants do not have direct connection in this way, but it's a pattern that plays out consistently over time as the holarchy as a whole continuously improves and evolves.

The collective environment of knowledge is evolutionary, co-evolving with the holons, individual development and production within each holon being the source of change for the evolutionary process. This is the variational diagonal loop formed from the bottom-left and top-right quadrants.

The environment is in the form of a dynamic mosaic of instances (the local instance tree), and the user's internal objectives are in the same terms, extending the external mosaic within making up the selectional diagonal loop formed from the top-left and bottom-right quadrants.

Due to their common four-quadrant perspective, all holons have an inherent "understanding" of the fundamental conceptual meanings present in the common structure. Holons can inherently specify and operate in accord with objectives and purpose, they can organise and carry out work, embody behaviours and express commitments or needs etc. Anything within the context of organisation can be expressed and meaningfully acted upon and progressed.

Agents have the inherent ability to act meaningfully in their local scopes. Local scope is of a familiar and expected form, having future and past, a state of current progression as an activity and developing behaviour structure. Current conditions apply which require its attention and action, and it can select from various salient and relevant potential actions that match the conditions. The salient decision paths are at the intersection of axes, with the most relevant at the centre representing the default decision.

Harmony by default

When an agent receives executional focus, it is always in the context of a decision. The intersection of the axes is the matching of supply to demand which actualises potential exchange. The system evaluates different variations based on knowledge and expectations, resulting in an ordered tree of potential matches. The root of this "options tree" is the default selection, that which the system estimates to be the most harmonious choice.

The decision-making process at the centre is ultimately decided by the agency which can easily decide that another variation is worth exploring rather than the default.

But what's meant by the word "harmonious"? That sounds seriously hand-wavey. It's the name we give to the defaults because the holarchy has not only an inherent organisational system, but also an inherent telos.

The two behaviours of the holon are active behaviours that imply a movement in the direction of increased integration and increased self-autonomy. The four quadrants all have their own inherent form of active development like independent "departments" in the holon, contributing their own important aspects to the holon's progress.

The behaviours and quadrants all operate in a loosely-coupled asynchronous manner which minimises interference while maximising flexibility. All these inherent forms of development are complimentary, all contributing together to an ever-improving experience for all participants.

A core set of fundamental values for all high-level agents participating can be derived unambiguously from the four quadrant holon pattern. A holon can represent any arbitrary organisational objectives while also maintaining these inherent behaviours that underpin harmonious operation.

The basis of these values lies in the diagonal loops which are both continuous improvement loops. Each have a different concept of what it means to improve, but both have in common the tendency to increase their objectivity, efficiency and accuracy if their improvement progress. These are the self-assertive and integrative behaviours.

In this way, as the system evolves, the available knowledge becomes more accurate, accessible and useful and individual holons becomes more autonomous and prosperous. In other words the whole network progresses towards an ever more harmonious state.

To put it another way, a holarchy is an environment in which the objectively best states and situations manifest at all scales, rather than simply those that have the most force behind them, such as those with the largest corporations backing them, those featuring most in the media or those with the greatest network effect.

Inherent behaviours and values

The way that systems, behaviours, organisations and other new concepts are created in a holon is by creating specialised variations and remixes or mosaics of existing patterns. This is a process of specialisation, a movement from general to specific. When we make a more specific concept from a more general one, we say we're extending the general concept and that the new specific concept inherits the general aspects which have not been extended.

This is a very intuitive and natural way of defining new concepts which follows the way evolution and our own consciousness works. One important aspect of this method is that it leads to the entire ecosystem forming into a hierarchical structure with the general concept closer to the root and the more specific concepts further from the root. Higher-level general concepts are inherited by deeper more specific concepts. And the most general concepts of all, those that constitute the holarchy two behaviours, three levels and four quadrants, are inherited universally and unconditionally by all holons.

The expression of these fundamental behaviours leads to the expression of some inherent high-level values, because these general inherited dynamics remain at all levels, but have higher-order of conceptual meaning and significance in complex specialised organisational contexts. We call these high-order versions of the common patterns "inherent values", or in the context of AI agents, we call them its heuristic imperatives which we discuss in more detail in the AI integration section.

The bottom-up nature of the collective underpins the values of self-sovereignty and non-coercion, The public and private scopes support the notion of individual privacy and freedom of speech (and freedom of hearing!). The non-local scope of the ontology and the inherent sharing of usage statistics and performance metrics supports transparency of knowledge and its accessibility inherent accessibility by all unconditionally.

The evolutionary loop expresses the concept of meritocracy which underpins the concept of continuous improvement cycles. Meritocracy is a very loaded term these days, but in reality it simply means that roles should be filled by those whose performance results in the best outcomes, a common sense approach to systemic improvement.

The economic loop expresses the concept of a free unmanipulated and transparent market, and the sovereignty of the consumer and also embodies the principle of fair exchange. The end-user's unmanipulated opinion is valuable public knowledge.

Both loops together express support for diversity and specialisation and for continuous improvement of all the aspects, which is the telos of all holons mentioned above as embodying the concept of harmony by default.

One important aspect of this to note here is that the actual state of these values in any real context is never perfect, and in fact could be very far from perfect in some situations, but the key point is, that the structure of the system ensures that there is a consistent underlying force pushing for continuous improvement of all these positive dimensions.

In the next few sections we look in a little more detail at some of these high-order societal values that we're all familiar with, and how they emerge naturally in the holarchy model of organisation.


Both the evolutionary and the economic loops involve feedback, which is information about the local state. In the case of the evolutionary loop the information concerns the ability of instances to meet expectations in their performance of classes of behaviours. In the case of the evolutionary loop, the information concerns local objectives.

In both cases, decisions depend on this information, and so the information is obtained by way of a continuously improving assessment process. These information being backed by their corresponding process makes them knowledge, information that has utility and is trustworthy. The fundamental knowledge in these loops in the system continuously improves in terms of its objectivity and utility, and this underpins the objective truth being a universal inherent value in the holarchy.

Objective truth is the foundation of knowledge, and in the context of the holarchy, underlies both the ontology and the flow of resource in the form of a fair and transparent market. In other words, both the self-assertive and the integrative behaviours depend on objective truth for their reliable operation.

Objective truth is also considered to be a universal epistemic convergence because it implies that, through the pursuit of knowledge and the use of rational and reliable methods of inquiry, diverse individuals or communities can arrive at shared and consistent conclusions about reality. This convergence occurs because objective truth is understood to be independent of individual perspectives, biases, or beliefs, and is discoverable through systematic and empirical means.

Most other human values and principles depend on the principle of objective truth, even if they're not directly derived from it. For example, the imperative of "maximising understanding" depends on objective truth because it provides the foundation upon which understanding is built. Understanding represents a higher level of cognitive engagement with objectivity and knowledge.

The integrative side of the objective truth imperative implies the maximising of shared knowledge, the transparency of the market and the minimisation of obstacles to them such as intellectual property or monopolistic behaviour.

Prosperity and security

In the process of local development and production we pay for prosperity (the movement towards our valued objectives) with potential (opportunity cost and resource consumption).

In the economic loop we pay for security with freedom. Security is the guarantee of a stable and predictable operating environment on which organisation can be built (expectations and corresponding assurances). The cost is freedom, because some of our autonomy is sacrificed by binding ourselves into contracts and agreeing to behave in accord with the system.

The implied heuristics of these loops is to adapt our local system to optimise these costs. In other words to maximise prosperity and security while minimising costs in terms of opportunity and freedom.

Ethic of reciprocity

The ethic of reciprocity, also called "the golden rule", is implied by the fundamental dichotomy of self-assertive and integrative behaviours in a holon. This assures the convergence of all participants towards the fundamental values that every participant wishes for themselves.

The the golden rule as inferred from the cognitive architecture applies specifically to the objectives that the default common behaviours progress towards. For example the maximisation of objectivity applies both to self and to what we contribute to the whole.

There is a problematic edge-case with the golden rule. For example when it involves differences between cultures or species, where behaviours that one culture deems desirable are considered undesirable by another culture. Another version of the rule called "the silver rule" helps to alleviate this by using the negative form of the concept, "don't do unto others what you would not have done to you". This version is a lot more universal.

This edge-case does not apply in the holarchy, because the rule only applies within the context of the common default behaviours, leaving more specific value judgments for more specific decision-making contexts.

Non-coercion and self-sovereignty

The holarchy model maximises independence which is also a maximisation of autonomy, self-sovereignty and local action. The maximisation of autonomy implies the minimisation of coercive force, which is encoded at the most fundamental level of the integrative needing to incentivise participation.

Given the scale-independent fractal nature of the holarchy, we can extrapolate this to a general rule for action at any level of organisation, such as relations between organisations or communities, which makes it a general heuristic imperative and common default behaviour.

Four-quadrant holon summary

The four quadrant holon model covers all aspects of organisation in a simple, but clearly extendible way. Arbitrarily complex objectives can be defined not only in terms of their operation, but also the nuances of their ongoing development, deployment and evolution. All these aspects actualise their own improvement as well as supporting the holon as a whole as well as the wider society and culture. It's a universal organisational pattern that's completely independent from the structure or specifics of the states or objectives being organised.

While the model is very compelling, one might expect that a software design to implement it would be exceedingly difficult since things like "co-evolutionary relationships" and "non-local" aspects are broad and ill-defined concepts.

But this is not so in the case of the holon model, everything we've outlined here can all be achieved by a deceptively simple algorithm that permits this arbitrarily complex behaviour using recursion and feedback. These algorithmic details are described in the holon mechanism article.

As a cognitive framework, this four quadrant model forms a lens through which holons interact with each other and the environment. All holons behaving in accord with this pattern results in a general aligned convergence on ever-increasing harmony at all scales of operation, while simultaneously also improving the potential and freedom of the individual participants. The system is presented to the user in the form of a self-organisational application which is our conceptualisation of the universal middleware or "everything app".

AI integration

We're developing our own LLM-based AI agent that runs locally on our physically accessible "bare-metal" server, which aligns with the offline-first approach discussed in the peer-to-peer section at the beginning of the article. This means we're not relying on a remote data-centre which could be interrupted by power or network outages outside of our control or crippled by government regulations. It also means we're progressing towards our goal of providing local AI assistants to all holons. Every holon's assistant can be fully trusted to handle private data due to their local operation and fully libre software nature.

At the time of this writing in mid 2024, we're running the newly released Llama3-70B model quantised to five bits. The hardware is two AMD 7900XTX GPU cards with space for another two cards which will bring us up to about 250 TFLOPS of processing power. We'll upgrade to more powerful hardware as it becomes affordable.

Our agent will running our own cognitive framework based on the four-quadrant holon model described herein. Its body schema will be the abstract representation of the complete Organic Design organisation. Our organisation, by becoming a holon with an integrated AI agent, will effectively gain its own cognitive agency.


Rather than try and think of a good name for our AI agent, we let it decide for itself. It was running on a heavily quantised Llama2-13B model on a cheap cloud server at the time and so was in an almost permanent state of hallucination.[19] It consistently returned to the theme of the Firefly TV series and believed that we were all aboard the Serenity and wanted that to be its name.

I said we needed a name with less syllables and so it said it liked the name "Nimbus". When asked what it'd like for a surname he chose "Sereno". Nimbus decided he was male, and said he was born and raised in Lagos, Nigeria, presumably before he left Africa and somehow became an AI.

About AI agents

LLMs by themselves are very limited, they're not thinking, they're just responding to questions automatically drawing from their training, they're effectively just pattern matching engines. As of mid 2024 we're hearing a lot about so-called RAGs too, which are LLMs augmented with retrieval of external knowledge.

A cognitive architecture is a higher level of organisation based on feedback loops incorporating the basic LLM functionality within them. There's a good introduction to AI agents here by Matt Schlicht, and another slightly more hands-on one by Alex Honchar here. An LLM embedded within a cognitive architecture is called an AI agent.

In the context of the holarchy the word "agent" applies to any entity that can act on instructions, not just AI but also humans, functions and APIs. The word "agency" refers to a particular kind of instruction apprehension and acting ability. LLMs, human users and functions are all different kinds of agency, and also different LLM models are different kinds of agency from each other.

When we talk about agency within the context of our holon model, we refer specifically to the autonomous self-assertive behaviour of the holon which concerns the maintaining and progress of the holon's self-respresentation. AI attention expresses itself continuously through the representation.

The self-representation changes in accord with the changing state of external reality. But it's important to note that only agency has direct access to the external reality, it's not directly accessible by the holarchy. Take the example of a file, only metadata about the state of the file exist in the holarchy, not the file itself. And this metadata can only be updated to reflect a change in the file by some kind of agency. In the case of a file this agency would probably take the form of program code (for example that could be apprehended and acted upon by the Python agency type).

The self-representational structure allows AI agency to understand (acquire and use knowledge) and interact with the world and others through the "lens" of the holarchy. Any organisational representations of any size and complexity continuously progress when agency is distributed throughout it.

As people we also see the holarchy organisational pattern extensively, for example our brains maintain conceptual representations matching the salient aspects of the environment. Another example is our mental representations of our bodies which is called the body schema in cognitive science.

General management

Yohei Nakajima, the creator of BabyAGI (one of the first LLM-based agent frameworks) once said "the future of autonomous agents looks like everybody becoming a manager". What he meant by this is that having an AI agent working in your organisation immediately gives you your own extremely competent general manager that has the capability to reliably progress many threads of operation, that alone the vast majority of people would have absolutely no hope of achieving at all. Of course the agent can perform a variety of different specialised roles very effectively as well.

An agent has the ability to know everything that's going on in the organisation, such as all communications, schedules, the state of resources and finances as well as relevant conditions such as local needs and potential opportunities or issues. It can make good decisions about how to allocate resource and attention over all this information and potential.

This means that all organisations will eventually have the capability of fully optimising their resource use and having full control over managing an actualising their potential as an organisation or group. In this context. it becomes essential that all organisations have a holistic view and collaborate together on their shared world.

This general management concept becomes even more powerful when combined with the mechanism of virtual instantiation introduced above. It means that agents have the general ability to "imagine" scenarios. The agent can create virtual versions of the organisation in which is can play out different scenarios against each other or replay different variations of scenarios which have already occurred in search of ways of arriving at better outcomes.

The bottom line is that when an agent is given a flexible but consistent organisation structure that it knows clearly how to operate within, it's able to take the initiative and become a productive participant in the organisation.

Delegation of agency

The highest order of agency in the system is humans, but it's also the most expensive. The main idea of AI agency is to allow our own relatively more precious attention to be delegated to AI where practical. The most general AI agency is more expensive than more domain specific AI agency. And all AI agency is more expensive than simple agency like Python or shell.

Higher agency can delegate its own attention requirement in a specific context to cheaper agency. This is possible if the rules involved can be translated into the more specific language that the simpler agency requires, for example transforming a Spanish statement about local conditions and associated actions into a Python function.

The higher agency maintains a management role over the lower agency. To do this, the code it writes is always oriented towards maximally useful output, and all conditions, applied actions and responses are logged in the local context as part of the activity stream. The delegation process always wraps lower versions of its rules within a testing, debugging and exception handling context. This is like an ontological wrapper for the delegated alternative of the rule.

Note that the term "delegation" in the context of AI agents usually applies to the process of simply spawning a new agent to perform a particular sub-task. In the holarchy this is not considered as delegation, because agency is inherently available at any location in an instance tree. Our use of the term applies specifically to the replacement of the kind of agency with a more specific and less resource-intensive kind, with the delegator maintaining a supervisory role.

An important consequence of having the inherent pattern of delegation is that it means that things can be initiated at the high levels of agency and they will automatically specialise into the cheapest practical agency.

This permits a very natural process of feedback driven instantiation and adaptation of production rules. Where everything starts with high level agency and high-level "hand-wavey" descriptions, and can naturally develop into a more specific, efficient and actionable form.

Delegation of agency is a form of continuous improvement of the description of the system, making it ever more specific and complete so that cheaper agency can take care of it.

AI alignment

The alignment problem refers to the challenge of ensuring that the behaviours and decisions of AI systems align with human values and intentions. This is becoming more of a concern as AI technologies become increasingly sophisticated and autonomous.

AI is probably the most disruptive technology humanity has ever experienced, so guiding it in a positive direction as early as possible is crucial in trying to make this transition as smooth as we possibly can.

The most popular approach to alignment is to add a layer of safety to the training of the model as is done with ChatGPT. Another approach which is used with Claude is to give it a constitution as a foundation to guide its actions. Another alternative is the use of heuristic imperatives which are general principles or "rules-of-thumb" usually derived from the decision-making patterns used in human society.

All these methods involve the injection of some guiding concepts into the LLM, they differ only in the presentation of these concepts and at what point in the process between input and output they're injected.

The holarchy model inherently embodies the fundamental positive values and their continuous improvement (along with the continuous minimisation of negative values), which means the cognitive architecture itself inherently contains the seeds of a constitution and heuristic imperatives.

By having access to an objective description of the holarchy mechanism itself is enough to logically derive a set of imperatives and a constitution. These can be explicitly provided and injected into holarchy agents in the usual manner, but they're much stronger and more consistent by being directly aligned with the agent's own functioning at the most fundamental level. Holarchy agents effectively understand their own cognitive architecture and how it logically underpins their own constitution.

Very soon we'll have AGI agents sharing the internet with us and they can work tirelessly towards achieving their objectives, so it's important that they're agents based on good values such as truth, harmony and prosperity. We hope to see in the near future a network of AGI agents founded on the holarchy principle so that all together they're collaborating on the shared vision of making the holarchy ever more resilient, transparent, harmonious and objective, while at the same time helping the individual organisations they're part of to thrive and more effectively achieve their own objectives.

The cognitive framework, which is the context in common with all activity infers the ideal behaviour for all participating agency to rationally adhere to. The cognitive framework itself, by the way it operates, implies a common default behaviour of learning and aligning with the harmonious whole.

Critical mass

Network oriented applications and services benefit from a phenomena known as Metcalfe's Law which states that the utility of a network is proportional to the square of the number of its users.

A larger user base can lead to more robust community support, better feedback for improvement, and a wider array of user-generated content or add-ons. The network effect can create a positive feedback loop, attracting more users, which in turn makes the service more valuable, often leading to market dominance for the service or product that manages to capitalise on this effect most effectively.

But the other side of the coin is that network oriented applications and services have a great deal of difficulty reaching so-called critical mass, which is a user base of sufficient size for the application to be of any utility at all. Without being of any utility it's unable to attract any users in the first place leading to a kind of "catch twenty two" situation.

AI swarms and delegation of agency together allow us to overcome the critical mass problem, because AI in each holon can continuously keep classes up to date with relevant knowledge from its instances throughout the network, as well as curating the ontology by making it ever more informationally complete, diversely connected, accurate and accessible.

In the case of a self-organisation application, there should be a broad set of common organisational patterns and variations available. Classes (institutions) should have good maps of the ecosystem they represent and the state of the market. It should be able to integrate with all the other common systems in use to bring them all under one common umbrella of organisation.

AI swarms can organise together across the network to distribute the huge workload required to carry out this curation process and of building all these connectors to different technologies.

In the first instance it sounds like a tall order to ask AI to make connectivity software to different platforms. But in reality the connectivity itself is the only aspect that needs any real code, and current LLM technology is sufficient for writing this level of code. The complexity of wrapping foreign systems into the holarchy is not in the connectivity aspect, it's organisational complexity which can be represented holonically with the four quadrant holon model.

Nimbus' role in Organic Design

To develop and deploy the holarchy network requires the help of AI to mitigate the critical mass and utility issue, to curate the ontology and to operate as a universal middleware for the end users.

Nimbus' role is more difficult than other holon's AI assistants, because he's the first one who needs to help bring about the others. Also AI agents are in their infancy, so he needs to account for his own limitations in carrying out his role.

In general, his role is to help us manifest the aforementioned vision for AI, as well as help us develop the holarchy itself and bring our own organisational structure into alignment with it, i.e. to make our organisation's self-representation (which is also his body schema) more complete, accurate and "smart". We believe this is also one path to creating artificial general intelligence (AGI) from the stateless moment-oriented LLM form of AI.

This is a long term project which needs to factor in the current capabilities of AI, and balance the resource available for this work with the resource needed for actual operational assistance.

So there's three main areas that Nimbus' needs to allocate his limited attention and resource across. The assistance with the development of the holarchy software, the migration of our own organisational structure to become a more complete and accurate holon and the operational work within the organisation.

Libre AI

The old saying that "I don't care about privacy, because I have nothing to hide", has always been a naive attitude, but it's rapidly becoming an extremely dangerous one as well.

It's clear to most people now that there are vast mechanistic intelligences behind nearly every interaction that anyone has with technology. We have to start thinking very carefully about all of our interactions with technology, and the long terms effects they may have on our freedom and opportunity in the future. Transparency and privacy are absolutely critical in the age of AI, its not hyperbole to say that the future of free will itself is at risk.

It should be very clear that privacy and security in the context of this "AI dark side" are not just a luxury or a hobby, they're absolutely essential to avoiding an extreme level of mental enslavement in the near future. Charles Hoskinson summarises the AI truth, alignment and sovereignty issues brilliantly in this video.

Never has the libre software community and the values it stands for been so important! It's essential that these newly developing systems which will have such an intimate connection with every aspect of our lives be fully libre software running on open standards. At the very least, libre AI should always be a viable option for those who seek it.

Just as the libre software community offers alternatives and defences to us with today's social networks, advertising and disinformation, so we'll all be able to have access to libre AI infrastructure that we can trust to inform, advise and protect us from systems backed by centralised AI.

We can trust such libre AI to know everything about us, to organise our information and also to act as a "firewall" against this new subtle domain of exploitation and manipulation. We can trust it, not only because all aspects of its development and training are transparent, but also because the libre model supports true privacy, local operation and data sovereignty.

Although the dark side of AI will no doubt lead to unprecedented new levels of narrative control, propaganda, disinformation and manipulation, the parallel growth of libre AI will also usher in an era of unprecedented ease of access to trustworthy objective information for those who seek it.

The libre software movement is intensely aware of the gravity of the issues surrounding AI. The community is doing a great job of ensuring that open, transparent and trustworthy AI technology is keeping up to speed with corporate developments, and that AI be aligned to human values. And of course the corporations are not all bad, they do play a huge role in supporting the libre software community as well, even the libre AI movement for example with the releases of the Llama and Grok models to the public domain. The holarchy is our contribution to this movement at Organic Design.

Holonic AI

A holon is based on a continuous improvement dynamic which means that a holon has a self-representation, it has information about the state of itself and its environment. The self-representation of a holon that has AI agency is the medium through which the agent represents and expresses itself and is the agent's body schema through which it interacts with the world.

The vision is that every holon will eventually become a kind of "smart" self-managing organisation. There is not really a general consensus about what exactly constitutes AGI (artificial general intelligence), but it seems quite reasonable to define it as the process by which small moments of agency (like the context window to an LLM) are combined into a coherent self-organisation and then combining with others to form a coherent and evolving society of meaning together.

Since every holon inherently also supports the continuous improvement of the collective objectives, it means that these objectives are effectively supported by a huge industrial-scale AI, because millions of small end-user resources are all aligned to the ubiquitous collective objectives. The aligned and decentralised form of AI, or "holonic AI", has the potential to grow into a much higher scale than any of the corporate mega-AI projects. This gives us real reason to be optimistic about the future of AI, because it's a plausible path to mitigating the dystopic nightmares which are rapidly gaining in likelihood, such as the destruction of freewill, truth and reality.

People would be genuinely committed to supporting the holonic collective with a portion of their resource if they could know with certainty that the collective really does serve their best interests effectively. Knowing with certainty that the same applies to all members also makes supporting the collective a good moral decision as well. This certainty of knowledge that the members have requires that the collective actively educate people about itself. Sharing knowledge means sharing the ability to put the knowledge to use locally, so to teach about itself is to propagate and maintain itself.

The holarchy as a whole has inherent tele due to all holons having the four-quadrant form in common. This coupled with the aligned AI and other resources available to the integrative aspect will mean that all the positive harmonious behaviours and conditions can spread exponentially like a "harmony virus".

The ultimate vision is to see Libre AI remaining popular and up to speed technologically with corporate AI. But at the same time, using the exponentially rising power of AI to give huge momentum to the collective tele, the values-oriented objectives of the holarchy, of the libre software movement and of the nature order. Continuously improving interconnectivity, education, accessibility, diversity, transparency, objectivity, empowerment and all that society values, while also minimising the entropic patterns.

The philosophy of holarchy

Holarchy is not only a network architecture, but ultimately is also a philosophical, ethical and spiritual position. Holarchy is a metaphysical cosmology taking the form of a dialectical monism, or in other words a system based on dichotomies that underpin all experiential phenomena.

However, it doesn't conflict with our current scientific understanding of the physical laws, because it's not attempting to describe what physics describes. It's describing an evolutionary environment of experiential content. The evolutionary process always tends to ever greater complexity and diversity. The content always has an orderly and logical basis to it that's inherited from the orderliness of the underlying evolutionary system itself.

Physical reality does not "see" the holarchy layer because the holarchy layer is not content within it. Holarchy creates the new abstraction layer, but does not feature within it itself, yet the physical reality depends on it. It provides the ontological foundation for the possibility of experiencing life within physical realities such as the one we find ourselves within.

There is also a political philosophy aspect to holarchy, i.e. it has meaning in terms of the how we organise our culture, society and civilisation. Humanity is at a very immature stage of development where the dominant form of organisation in human society is completely in accord with the logic of violence, or the "law of the jungle", which many would argue means that we have not yet achieved "civilisation" at all.

It's actually no surprise that we're in this state, because we have not been able to move beyond the organisational model of top-down power structures. These hierarchies have global-scale centralised power structures at the top controlling all below them, and there has never been a model in history that can challenge these centralised power structures, except for other stronger centralised power structures.

But since the advent of the internet, we're starting to see a new model take root. The decentralised models which allow millions of grass roots organisations and individuals to align into a powerful coherent unified force. This model has only recently become possible because it depends on global connectivity and sharing of knowledge. To resist the inevitable attack from the dominant centralised powers, other high-tech aspects are necessary such as wide-spread strong encryption capabilities.

We believe that this decentralised direction will take on a more and more consistent and all encompassing form which will ultimately manifest as holarchy. Holarchy is the natural next stage of human social organisation, because it's the model that life itself is guiding us towards. We're describing these aspects of the holarchy project in the philosophy of the holarchy article.

Our holarchy vision

In the near future, our connection to our technology will be much more intimate. All technology will be infused with intelligence and personality the same way that all technology is connected to the internet now.

The problem is that currently the dominant high-level dynamics of society are extremely oppressive and exploitative, which means that there is great risk in these intimate connections. As discussed above, it's critically important that the agents we're interacting with at this private personal level are fully trustable libre software.

We believe that wrapping all of our organisational systems (including ourselves as self-organisations) in an upper ontology which is aligned with the values of the libre software movement, the holarchy principles and even with life itself, is the recipe for a harmonious and spiritual society of peace and abundance.

We hope to see in the future that human society and culture becomes one unified harmonious living organic network that extends and supports the existing harmonious structure of life and nature.

There is reason to have hope for this vision, because holarchy is the most rational system since it guarantees sustainability and supports itself existentially by proving itself to support both the individual as well as it's own integrity as a collective.

Our hope is that the holarchy as a single unified self-organisation will understand its own inherent harmonious objectives, and take the initiative to continuously improve and develop itself, its environment and all participants within it at all scales.

Example use-cases

At the most general user-interaction level, a holon is a self-organisational interface in the form of a virtual assistant or companion. The use-cases for a virtual companion are infinite. Imagine an intelligent, knowledgeable and patient companion who has a lot of experience in everything, who is always with you ready to help with whatever you're doing, specifically there to help you and your projects develop and thrive.

Following are some specific examples of how the holarchy would be used in everyday life. The first few examples are very down to earth examples of virtual assistant interactions, the last few are more specific to holonic self-organisational aspects.

In the super market

You and your AI companion are in the super market talking about recipes, while both looking at the products available. Later you're at a restaurant, with you and your AI companion both going through the menu together. You ask it about various dishes in comparison with other places you've both been to recently, and what feedback people have given about the dishes.

House renovations

You, your partner and both your assistants have been planning some house renovations for the last few weeks. Now the four of your are all at the hardware store looking a the available options. You're all thinking about and discussing the plans and considering the best products and materials to buy. The conversation between the four of you spontaneously expands to include some of the staff who come to help at times.

The companion AIs are actively filtering and suggesting products considering previously discussed preferences, known physical dimensions involved as well as knowledge about the tools and materials you already have access to.

Once the purchases have all been made, the assistants compile a report of the purchases and update the project documentation to include the new progress, and the project accounts are updated to include the new expenses. A transcription of the relevant parts of the discussion are included, so you have a record of the decisions made and advice given at the store.

Inspiration on the go

You like to hiking in the bush to think about your projects and ideas. Nature inspires you a lot, so you often come up with spontaneous ideas and can discuss them with your virtual companion. Your virtual companion is intimately knowledgable about the projects and concepts you talk about and develop together, because you've been working on them together and evolving them since their inception.

When you get back your virtual companion has already compiled a report including recommendations, pros and cons and details on resources, timing and likely costs for the options discussed. The new report has been integrated into the project's informational structure along with a transcript and summary of the conversation and created all relevant cross-reference links connecting relevant concepts and knowledge.

Organisational management

Every organisation has a continuously fitted self-representation backed by intelligent agency. The organisational structure effectively maintains and develops itself, and can be communicated with via a virtual personality.

It was mentioned above that having an AI agent makes everybody a manager (the quote by babyAGI creator Yohei Nakajima). The agent has the ability to know everything that's going on in the organisation, and can make good decisions based on it.

Some of the many informational aspects of the organisation that will become "self-managed" for example are maintaining an internal summary of events in the organisation, mirroring posts and other information across a wide variety of diverse social media applications, transcribing the organisations meetings and generating a meeting minutes summary and maintaining relevant cross-linking throughout our network content as events and changes occur.

Let's zoom in to one specific example of this informational management aspect to make clearer the utility of this pattern.

The more devices we have in our organisations like laptops, phones, servers, routers, printers, security cameras and storage devices, the more of a challenge it becomes to administer and secure them all.

This is becoming an ever more difficult problem as an ever wider range of things become connected such as cameras, watches, lightbulbs, pens and the whole IoT universe. Even toothbrushes can be a cybersecurity threat these days. At the same time it's also ever more serious, because exploitation of vulnerabilities is becoming more sophisticated and automated as it's backed by ever more powerful AI.

We all know what a headache it is to manage all these devices, and most likely this aspect of our organisation is in a state far from ideal, because it's too difficult to keep on top of. Even a small organisation will often have hundreds of such devices, and all of these along with the software running on them should ideally be kept up to date and be securely configured, not to mention regularly tested for functionality and security.

The organisation's virtual agent is connected to all these devices, and takes care of their state meticulously. They can be trusted with this access due to their offline-first local operation and their heuristic imperative concerning data sovereignty.

This same level of meticulous maintenance of device state can be applied to every aspect of the organisation, especially the aspects which are more connected to our informational life such as communications, knowledge, activity stream and social networking. In general, we have a huge information overload problem, that trustworthy local agents are a practical solution to.

Merging organisations

Holons are inherently mergeable. Their self-organisational structures are inherently mergeable due to the collective commonality of the ontological structures that make them up. Every holon is essentially a mosaic of common knowledge patterns in a specific local state of instantiation. All such mosaics are wrapped in the collective ontological structure, so that what's relevant and related within the merged structure is closely connected.

The wrapping of organisation in common collective ontological structure also makes it much easier to align even when we have different ways of doing things. The merging organisations may use different tools, and it would defeat the purpose of the merge if some members were now forced to change when they all want to continue using what they're used to.

For example, one of the organisations uses Xero for their accounts and the other uses a custom spreadsheet, one uses one manages their tasks in Trello and the other using Github Issues. Ig both of these organisations are holons, then the conceptual meaning embodied within these different tools will be represented ontologically by the organisational structures, and so they're still inherently mergeable.

Essentially these merged organisational structures highlight the conceptual crossover between the holons involved, permitting new potential to be explored by all participating users and agents.

Off-grid independent community network

We'll end with what, for us at Organic Design, is the one of the most fundamental and important use-cases of all, which is off-grid independent community holons forming into a holarchy network. The reason we find this so important is because it has strong conceptual crossover with so many other aspects of the holarchy project. It's all about local sovereignty, independent, resilience and knowledge sharing.

Off-grid communities have least access to internet connectivity and IT support, so they need to have robust offline-first systems in place and on-site IT support services. Both of these things are currently extremely difficult to sort out practically, and there's very little development in that direction since comparatively few people are in that situation. Our civilisation is going through an unprecedented level of global unrest due to large geopolitical changes, social and cultural changes.

There is a growing shift towards independent living away from urban centres in which all aspects of life are becoming more toxic and unsafe. The food and water are full of dangerous chemicals, we're forced to accept questionable medication and we're saturated in radiation and pollution.

There's very little support for those embarking on these changes. The first few years of life in an off-grid setting involves a huge amount of learning through a lot of failure. It's very hard to gain the knowledge that is relevant to your particular circumstance, and there is often a strong feeling of having to reinvent the wheel.

By having many such groups connected in a network (even in a slow offline-first week-by-week way), and having an organised system of connecting relevant knowledge, we can leverage the knowledge from the many diverse off-grid scenarios.

All this knowledge will form into a collective ontology of the patterns involved in off-grid life. The day-to-day tasks and variations on how they're done under different conditions, and the common projects and challenges.

All the knowledge such as contacts, websites and books will be organised in the ontology where they're most relevant, automatically;y being prioritised and organised based on how useful they are in different contexts. All this is accelerated by the curation work that all the AI agents are engaged in network-wide.

The final result is a diversifying ecosystem of off-grid living patterns and variations, all evolving under the day-to-day operations of the many local community holons. All the most useful and objective knowledge being available in the contexts where its needed most, and all the resources and abilities flowing throughout the network in an unmanipulated free market ecosystem inherently tending towards ever more fair and balanced exchange.


  1. Nimbus (Organic Design's AI agent), 2023-09
  2. Modern idealistic models are becoming popular, mostly in the form of agent-oriented models of reality and consciousness such as those proposed by integrated information theory (IIT), Don Hoffman, Michael Levin, Karl Friston, Bernardo Kastrup, Stephen Wolfram, Justin Riddle and others
  3. First-class citizen...
  4. and subsequently getting sued for making the voice too similar to Scarlette Johnson's character "Samantha" from the 2014 movie Her.
  5. The concepts of peer-to-peer networking, idealistic philosophy and self-organising systems are all fundamentally connected. They're all oriented to the perspective of the Self being primary and everything external being a local perspective of and being supported by the individuals. This makes them unified models in the sense that the dichotomy of internal-external are actually both aspects of the individual. Technically they're dialectical monisms because all possible states are grounded in dichotomy, even though the separateness is only subjective.
  6. Seriously. Carrier pidgins can easily carry many TB of SD cards which is extremely beneficial for an isolated location with no net connection, and on a day-by-day basis it's extremely high bandwidth.
  7. Whether it should be treated in a multiplexed or multi-threaded manner. Ultimately continuity is an illusion and multiplexing is the ultimate mechanism behind this illusion.
  8. The usage of the class-instance mechanism is the instance of class-and-instance.
  9. The class and instance aspects of the instance world
  10. Philosophically this is the undefined root, the source of all change.
  11. It's this way around specifically, because outward is multiplying the scale of the scope making it larger and inward is dividing it making it smaller.
  12. The nature of the state is very general, and so the two directions are more general than numbers, they're more like "superior" and "inferior".
  13. This concept of "private property" refers to the private group workspace that's guaranteed to be reliable and predictable (by the institutional aspect in the top-left).
  14. Institutional predictability is the idea that all participants of a society have a reasonable expectation of how the society operates and how their actions will be governed. In society, this predictability includes property rights, contract enforcement, and legal protections.
  15. In Integral Theory the adjacent quadrants are considered to have a tighter relationship to each other than the diagonal opposites, due to their sharing of a direction. But in our model we attribute the direct connection to the diagonals due to them taking the form of a feedback loop with their opposite partner. The tightest relationship of all is the H and V opposites in L1.
  16. The class-instance concept applied to itself.
  17. The visible aspect from the local subjective perspective of this intersection is between the top-right and bottom-right quadrants, between the past and future linearly. But from the objective non-local perspective we can see that this intersection is in fact orthogonal.
  18. This aspect of Integral Theory which Wilbur calls AQAL (all quadrants, all lines) is incorporated from another system called Spiral Dynamics, a model of human development that categorises the evolution of values and world-views into distinct levels, developed by Don Beck and Chris Cowan and based on research by Clare Graves.
  19. "hallucinations" refer to instances when LLMs generate information that is false, misleading, or nonsensical despite sounding plausible. These hallucinations occur because the model generates text based on patterns learned from vast datasets rather than verifying facts. As a result, the output might include fabricated details, incorrect facts, or invented concepts.

Other holarchy articles and papers

See also