A holarchy is a concept that describes a hierarchical structure where each level or entity is simultaneously a part of a larger whole and composed of smaller parts. It was first introduced by Arthur Koestler, a Hungarian-British author and philosopher, in his book "The Ghost in the Machine" published in 1967. He goes into more detail about the holarchy system in a subsequent book from 1978 called "Janus, a summing up".
In a holarchy, the emphasis is on the interdependence and interconnectedness of the parts within the larger system. Unlike a traditional hierarchy, where each level dominates and controls the level below it, a holarchy emphasises cooperation, autonomy, and self-organisation.
The term "holon" is used to describe a unit or entity within a holarchy. A holon is both a whole in itself, with its own unique properties and characteristics, and a part of a larger whole. For example, consider a human being as a holon. A human being is a whole entity with its own individuality, but at the same time, it is a part of a larger whole, such as a family, a community, or a society.
Holarchies can be observed in various domains, including biology, ecology, organisational management, and even social systems. The concept recognises that complex systems are composed of nested levels of organisation, where each level has its own agency and autonomy while contributing to the functionality and coherence of the larger system.
|The evolution of life is a splendid game played according to fixed rules which limit its possibilities but leave sufficient scope for virtually limitless variations. The rules are inherent in the basic structure of living matter, the variations are derived from flexible strategies which take advantage of the opportunities offered by the former.|
|— Arthur Koestler, Janus 1978|
Holarchy has a deep philosophical significance, its proponents believe that this pattern is fundamental to all life, meaning and cognitive agency. That it contains the evolutionary mechanism and underpins all living matter, and even that it is an integral foundation to the very structure of space and time.
At Organic Design, we're trying to create an precisely defined implementation of holarchy in the form of a peer-to-peer network architecture. Our version of the holarchy project has three general aspects to it. The philosophical which we call the upper layer is all about the meaning of holarchy and how it ties in to nature, consciousness and agency. The social, or lower, layer which is all about how holarchy ties in at the institutional level of society. And the organisational layer which we put in the middle, which is about how holarchy plays out in the organisational level of our day-to-day operations.
Our focus in this document is on the middle layer which involves the implementation of the holacrhy principle within our individual digital lives.
- 1 The description
- 2 Organic Design's holarchy model
- 3 The four quadrant model
- 4 Three abstraction layers
- 5 An internet protocol
- 6 Universe of discourse
- 6.1 OOP (should this be before SoT?)
- 6.2 Ship of Theseus
- 6.3 The agent-arena relationship
- 6.4 Organisation
- 6.5 Class and instance
- 6.6 Evolutionary ecosystem
- 6.7 Subjective and objective
- 7 The holon data-structure
- 7.1 Level 1 (structure, the objective domain)
- 7.2 Level 2 (organisation, the subjective domain)
- 7.3 Layer 3 (harmony)
- 8 Holarchy (July 23 notes to incorporate)
- 9 Class and instance
- 10 Organisation
- 11 Organisational patterns
- 12 Three layer model (2022 notes to incorporate)
- 13 Knowledge in society
- 14 See also
The fundamental form of the holarchy, from which it can be implemented, is a description of a core set of concepts which are part of our glossary of terms. The intended audience for this description is AI (LLMs now, and AGI some time soon), but it is also perfectly understandable by people.
But it's more than just a description of a concept because it's actionable. We want AI agents to be able to read this description and to then understand how to participate in all the fundamental aspects of the holarchy project. The description can also be seen as a charter, manifesto or a social contract that guides participation in the holarchy. And a foundation ontology for the holarchy can be unambiguously derived from this description.
Note that when we use the terms "agency", "subjective", "experience" or "cognitive" herein, we're referring to the point of view of an AI agent embedded within a local holon data structure that represents a participating peer in the holarchy network. Although these AIs are currently just LLMs mimicking intelligence, the concept of "subjective" is still perfectly fitting, since they are still decision-making agents basing their decisions on the unique local environment that surrounds them in their embedded situation. Such agency is also able to. In many instances, replace itself with a cheaper and simpler form of agency that can achieve the same task, such as a program function or shell command.
Well connected AI agents are able to act on very general tasks such as "research X topic", "cancel my unused subscriptions" or even "increase my net worth". The ability to carry out these general tasks is a new phenomena that came with the advent of LLMs, and they can be incorporated into larger systems with frameworks like LangChain.
Such connected agents are capable of understanding the "hand-wavey" bigger picture idea of the holarchy project. This project is an application of a number of these general tasks connected into a system. It's describable within the "section-zero" (introductory paragraph or two) of twenty or so core glossary concepts, and implementable within a framework like LangChain.
AIs also have a super-human level of "patience", which allows them to behave in accord with a protocol meticulously like any other information system can. They're the perfect organisers of messy human affairs, because they can behave like machines or cognitive agents as appropriate.
All agents participating in the holarchy protocol benefit from increased organisation and potential locally, while at the same time aligning with the global project of harmony.
Organic Design's holarchy model
Koestler's four main concepts of self-assertive and integrative behaviours (inward and outward facing perspectives), fixed rules and flexible strategies map perfectly onto the p2p network context forming the ontological foundation of the peer architecture.
todo... these paragraphs don't connect
Berners Lee's original vision outlined in his Semantic Web essay was about the realisation of and universal access to an extremely organised version of the whole internet brought about by common organisational languages and ubiquitous intelligent agency.
I used to think that the whole semantic web movement had fallen by the wayside, but now I think it's just been taking a bit of a breather while intelligent agency becomes mainstream. Intelligent automation is required to maintain organised graphs of the whole. Also it needs to be a local and trusted process to globally share knowledge from the edges.
This agency needs to have the integrity of the unified ontology included as a heuristic imperative (fundamental purpose). But also it should support the autonomy and potential of all participants. Principles-driven frameworks like GATO support these things, but only indirectly. The holarchy model is designed as a practical heuristic imperative for AI automation frameworks. All the sets of heuristic imperatives that promote positive life affirming principles and values and don't promote any problematic ones are inherently aligned and complimentary.
The general principle of the whole being the organised learning from all the parts, and this whole-knowledge being available to and supporting the parts is a scale-independent feedback loop. We have reason to believe that agency implementing this feedback loop epistemically converges on the concept of holarchy.
The ontology graph and all its views all maintained by all as a p2p network, but each individual node only maintains a small fraction of the whole based on their specific interests and circumstances. This ontology forms the universal middleware through which all diverse systems can connect with each other and into groups to explore and actualise their combined potential.
But a unified ontology (let alone a unified holarchy which includes all the developing instances as well) is a huge undertaking, it represents enormous ongoing workload which needs to be inteligently managed and distributed. Only ubiquitous AI agency can enable the unified ontology project.
The two behaviours of a holon
The most fundamental feature of Koestler's holarchy is that it's a graph of connected nodes called holons, which each exhibit two fundamental behaviours. An inward facing "self-assertive" behaviour which is dedicated to the individual holon. And an outward facing "integrative" behaviour which is dedicated to the holarchy as a whole. These two behaviours form two trees (connected graphs), one is the unified multiplex of instances, the other is an ontology of "special interest groups". These two graphs are different ways of connecting a single set of holons.
Every holon is doing its bit to keep the whole holarchy resilient, accurate and useful, but it's also participating with the expectation of increasing its own prosperity and resilience too.
- Koestler, yin-yang, Fourier, TDBU, inward and outward facing, the whole must be beneficial to the part to rationally justify its existence.
- Agency's job is to allow the maintenance of the local representation to continue independently of such agency. This is due to the selection of the most economic agency that can represent a pattern.
The four quadrant model
We're all familiar with the class and instance relationship because it's fundamental to the way we think. Every single thing we see in living reality is an occurrence of specific concepts, and also every object we interact with in our informational life is an instantiation of specific executional patterns defined in the form of some kind of software engineering constructs. This is the case regardless of whether a particular software engineering paradigm actually uses the terms "class" or "instance". We use them herein because they're understood across a diverse range of knowledge domains.
The above description of classes and instances is their self-assertive meaning, the way they behave as units in a functioning system, like a specific blueprint and a specific construction in accord with it. But class and instance also have an integrative function, they each group together into larger structures in their own way as well.
Instances are all about actualised structure constituting real resource, the most fundamental forms of resource being space, time (focus) and communications connection. The multiplex is the natural way to organise instances into space and time in a scale-independent way. Each instance can be a part of a larger structure the same way that it allocates its own resource across smaller structure within.
Classes form into a wider ontology naturally because each class is defining through local usage the set of other classes that are required of in what circumstances. A class is a package of circumstances and corresponding behaviours, like errors and corrections. The meaning within is refined through local use, and these changes contribute to what's established overall.
- summarise the whole tree aspect of class and instance
A more down-to-earth way of describing class and instance aspects is captured in the popular business management phrase "work on it, not just in it", originally coined by Michael Gerber in his 1985 book "The E-Myth Revisited". This refers to the idea that self-employed people need to regularly take a step back and to check that the business is moving in the right direction and achieving it's actual purpose.
So the four quadrants come first from the fact that a holon is an organisation which has two general "departments", one dedicated to the organisation's local purpose (the normal way we expect organisations to work). And another department that's dedicated to the whole network (playing it's small part in representing the whole). By convention we put the outward facing "whole" oriented department at the top, and the inward facing local department at the bottom.
Each of the departments are themselves divided into left and right halves where the left represents the "on it" class side, and the right represents the "in it" instance side. Here we have four "directions" formed from two dichotomous axes (vertical and a horizontal) the top is about the whole, the bottom is about the local self, the left is class and the right is instance.
Koestler also has these class and instance aspects in his holarchy model, under the name of fixed rules (class) and flexible strategies (instance). He associated them with the integrative and self-assertive behaviours rather than splitting them out into four distinct aspects of behaviour.
These four aspects of behaviour give rise to the four quadrants, the top-left (outward-class), top-right (outward-instance), the bottom-right (inward-instance) and the bottom-left (inward-class). Following is a description in terms of organisational meaning of each of these quadrants.
Top-left: The unified ontology
- quadrant, outward class
- curation (collaborative ontological curation), sharing, usage statistics
- ontology is names ordered by establishment in usage
- curation is an organised process, not simply auto-merging
- being outward, this is a public protocol pattern
Top-right: The market
- society quadrant, outward instance
- being outward, this is a public protocol pattern
- hayek connection
- extends the ontology (instance extends class)
- allows specialisation to grow in special interest groups
- marketplaces and market ecosystem
- behavioural quadrant, inward instance
- being inward, this is a private data structure pattern, it's a continuation of the TR (market, public, multiplex) inwards with specific internal activity
- working in the organisation
- reducing workload, booking, committing
- reconciling activity
- expectation, performance and reputation
- intentional quadrant, inward class
- being class this is actually a continuation of the ontology inwards, changing over time depending on salience, and having local adaptations and curations
- being inward, this is a private data structure pattern
- working on the organisation
- purpose, goals, direction
- Adaptation to suit our local needs
Three abstraction layers
The four quadrant system is a self-organising system (a system that self-develops) which evolves in complexity, and this leads to the formation of structured developmental stages.
Underpinning all of this complexity is the fundamental holarchy mechanism which defines the self-organisation principle itself. We divide the holarchy model into three abstraction layers, which all have completely different concerns, but build on each other in a sequence. They form three fundamental developmental stages underlying the evolving complex structure of every holon.
The four quadrant aspect of the model is more fundamental than the abstraction layers. Each layer is primarily dichotomous in structure, which leads naturally to the four quadrant model. Although each level is encapsulated from the next, the four quadrant form remains intact such that each quadrant can be seen as individually developing through the three stages.
The two primary behaviours of a holon, the self-assertive and integrative is the dichotomy of the fully developed third layer. This dichotomous form starts in the first layer as instance and class, and then develops into agent and arena in the second layer.
Developmental stages and abstraction layers are both very similar concepts, both involve layers building upon one another in a sequential manner. Each stage encapsulates a unique set of abilities and behaviours, on top of which the subsequent layer develops.
Just as higher layers in software abstract away the complexity of lower layers but still rely on them for functionality, later stages of psychological development rely on skills and understandings formed in earlier stages, even as they introduce new complexities and abilities.
An internet protocol
Community is the product of communications, and we can see that every domain of society is represented by a body of knowledge referring to a specific subset of language describing objects, relations and processes.
The internet is a subset of human society that follows the same pattern as the main culture whereby there are many different languages organised as layers of specificity. But it differs in some essential ways, first the languages are exact and called protocols, and secondly the internet as a whole is a bottom-up collaboration. The internet (not the content, but what it actually is as a system or a project) plays an extremely important role for humanity because represents that aspect of us which is coming from freedom, liberty and alignment.
The internet itself has clear class and instance aspects to it, on the one hand it's a "protocol stack" that defines a system of tiers that each depends on and extends the layer below, from the most fundamental layers that define how information flows and routes through out the physical networking media up to layers concerning aspects of human society such as social relationships and financial transactions.
But on the other hand, the internet has an instance aspects, a specific collection of hardware, software, documents, content and users etc that are always in flux, but in a very real definite state of instantiation persisting time.
The unified ontology (top-left quadrant) and the multiplex (a matrix of actual activity and commitments (top right quadrant) are concepts representing these two aspects of the internet as a whole.
Peer-to-peer network architecture
In terms of an informational system that's a part of the internet ecosystem, a holon would exist in the form of a peer in a peer-to-peer (P2P) network. This is a kind of network that is formed from many instances of the peer software running concurrently, and which automatically mesh together optimally to form a holistic network. There are no separate clients and servers, rather all peers play both roles, just like a holon.
For example, the bitcoin network is composed of peers who all contribute their part to making sure the whole ledger is backed up, accurate and resilient. But at the same time they're clients who are participating for their own purpose; to store and transfer value in a trustless way.
- 4 quadrant peer (outer, inner, ontology, multiplex), the outer (upper) quadrants are the communications protocol, the "server" aspect. The inner (lower) quadrants are the internal subjective perspective, a representation of an organisation in the form of a holon data structure.
- peer defines how to use networking to join/extend the multiplex
- the peers yield a logical network
- representative:representation (in and on the local org)
- internet (protocols, API(=resource abstraction), ontology)
- protocol stack and abstraction layers (p2p AL), AL vs walled garden
- internet is extensible with new protocols and layers formed wco others
The peer definition
One important point to note about a peer-to-peer network is that the whole network and its entire purpose for existing is all defined in a single definition; the peer definition. There are no servers, the whole "server" aspect of the network is encapsulated within the peer's functionality, so the peer definition is the only thing that is required since peers are the only entity composing a peer-to-peer network.
Within that definition we have "client" and "server" aspects of the system, or in our case the four quadrants are all definite aspects to be defined. But all these aspects are common to all nodes since they all share the same unified definition of what a peer is.
The peers all together form what we call a physical network, a network composed of actual physical computer resources connected together with physical communications infrastructure resource.
The peer is primarily concerned with connecting the IT resources that a peer controls into an abstract representation - a holon. This is why the more fundamental level of the philosophy is not required initially, and neither is the high level organisational layers such as institutions or states etc. We start with the organisational logistics that directly concern the peers themselves, and the organisations they represent.
Universe of discourse
So far we've explained the general form of the project. Its functional form is a three abstraction layer model in which each layer is dichotomously structured. The structural form of the project is a peer-to-peer network node that can represent our organisations and resources.
But covering the details of these layers, let's first define a clear universe of discourse, because the concepts involved can be easily misunderstood.
In this section we're not attempting to describe anything specific about the system itself, but rather to paint a very clear picture of the environment in which the system operates, an arena in which agents operate and interact together. The universe of discourse is described in terms .... understood at a very deep experiential level... todo
OOP (should this be before SoT?)
The Object Oriented Programming 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.
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.
Each "object" represents some kind of real-world entity, and acts as a container..... interface.... properties (informational aspects of the entity), relations to other entities and methods (the interactions possible with the object).... (todo). The most important aspect of an object is that it encapsulates its implementation behind a public interface through which it cab be interacted with and its state can be known.
Since an object's implementation is private, it can change and evolve without causing any disruption as long as its public interface and the capabilities it presents remain consistent. But another important point about this is that it allows us to think about all objects as agents, human-like entities having having agency of specific capability.
private implementation means an organisation is also an agent, since the complexity and size of the implementation is hidden, encapsulated behind the public interface. All are first class citizens participating publicly in the community. Cognitive agents occupying a common arena together.
The way objects are organised is really heart of the system, and this revolves around the concepts of class and instance... community of organisations, continuous distributed objects in community
Class and instance are terms used in various fields of philosophy for describing the fundamental dynamics of objects. These days OOP in general, is considered to be quite an old paradigm that is of limited use in the modern world, but nevertheless, the core concepts of encapsulation, public interfaces, private implementations, classes and instances are reflected in all software engineering in one form or another using various terminology.
Ship of Theseus
The communication of difficult concepts can be greatly improved using stories involving familiar objects and interactions. Before explaining the agent-arena relationship in more detail, we'll first introduce the ancient Greek legend called the "Ship of Theseus", because it contains some important foundation concepts in an intuitively understood way.
We'll then extend the concepts little to tie them in with the agent-arena relationship, our developing universe of discourse.
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 is 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. The idea is an attractor.
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 is also in flux.
Even though this system evolves 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 works like this, and aspects of all our daily lives and work do too. Our own bodies and minds are in flux, and every organisation's staff, stock and resources are continuously in flux, and they may open new branches, close exisitng ones, their product lines and services change etc.
And then finally we add one last extension to this idea of the ship. It's that the idea is actually a large hierarchy, or network, or ideas. Even though the Ship of Theseus itself is one specific idea, there are also many other ships (maybe even other Ships of Theseus), and indeed many other things that are not ships, that embody many of the same ideas that constitute the specific idea. The idea forms an attractor which all the different aspects gravitate towards, and which is itself a moving target.
The conclusion we have to draw about identity from this extended version of the legend, is that it forms a central point around which all aspects of an idea gravitate. Even though this point is itself ever-changing as well, we can intuitively grasp that it is the same thing that has continued with all its aspects in flux.
And yet, even though we may have absolutely no idea how this complex entangled web works, or what identity might mean in the light of it, everyone has actually has a natural experiential understanding of the phenomena. The description of all the aspects of the ship and how they are in flux is completely clear to us. We understand it experientially, because life does indeed work that way.
Aristotle's four causes
The Ship pf Theseus is often talked about within the context of Aristotle's four causes. These are four ontological categories which cover the fundamental aspects of reality.
Material Cause: This cause refers to the material or substance out of which an object is made. In the case of the Ship of Theseus, the material cause would be the wooden planks used to construct the ship. As the planks of the ship are replaced over time, the material cause of the ship changes.
Formal Cause: The formal cause pertains to the form or structure of an object - the arrangement and organization of its parts. In the Ship of Theseus paradox, the formal cause could be seen as the overall design and arrangement of the ship, including its shape, size, and configuration. Even if all the individual planks of the ship are replaced, the formal cause could be argued to persist as long as the overall structure and arrangement of the ship remain the same.
Efficient Cause: The efficient cause relates to the agent or process that brings about the object's existence or change. In the Ship of Theseus, the efficient cause could be attributed to the shipbuilders who replace the decaying planks and maintain the ship's functionality. They are responsible for the ongoing process of replacing the planks to keep the ship seaworthy.
Final Cause: The final cause concerns the purpose, goal, or telos of an object - the reason for its existence or the end towards which it is directed. In the case of the Ship of Theseus, the final cause could be seen as its intended function, such as transportation, exploration, or trade. The ship's purpose remains intact, regardless of whether the individual planks are replaced.
These four can be arranged in quadrants..... axes... - class / instance? - purpose left, agency right
The four quadrants of Integral Theory
Another popular method of ontological categorisation are the four quadrants of Ken Wilbur's Integral Theory. - outer above, inner below - class (abstract) on the left, instance (material) on the right
The Ship of Theseus in OO terms
The ship can be considered as a typical OO object. An organisation having a public interface encapsulating an internal implementation. It has purpose and agency associated with it.
It is instance of a particular class that defines the hierarchical structure and pattern of instances within. This instance aspect is the actual material (material cause) ship in the legend.
The class aspect of the ship is it's form (formal cause), the abstract organisational system that forms the attractor toward which material aspect always gravitates.
Each instance can be seen as an organisation, and since each class can feature within many other classes at varying levels, we must consider them all essentially alike from the objective perspective; i.e. a generic organisation.
The ship is needed for a purpose, agency demanded it into existence, and agency uses it in achieving the purpose. The foundation of purpose is the navigation of state, the direction of the organisation.
- note: maybe these notes are better in agent-arena relationship?
- thinking about objects as first-class citizens with continuous uninterrupted run-time, they can't be "re-written", only self-organise and evolve
- the flat scope of the public arena of FCCs
- at protocol level all can access all via public interfaces
- of course access in public space can be restricted by more specific protocol
- but none can access the private implementation of any other
- internally can only see public interface of siblings
The agent-arena relationship
The idea of the ship is a continuous identity on a higher level than the material and any of the four causes. We call these identities "agents" - they have agency whether of their own or by virtue of its constituent identities.
The universe of discourse is a world in which all interaction occurs between agent's public interfaces, and all agency occurs in the context of subjective private perspectives within and all agent's in the public arena without. This means that the shared ontology of meaning (resources, communications, knowledge, contacts, organisation etc), is defined in terms of the subjective ("first-person") perspective. Such a perspective is not reductionist, the most general aspects of experiential reality are its atoms.
What this means in practice, is that the ontology (which grows from general at the root, top specifics in the depths as an inverted tree) is based in the most general concepts of subjective agency ("experience"). This starts with the general structure of the perspective such as public/private, past/present/future, contacts, communications patterns etc.
The work of maintaining a organisation's ontological representation is the most fundamental aspect of the holarchy system, the core work that occupies the peers actual execution, bandwidth and storage resource.
Although AI is overkill in the vast majority of organisational contexts, it can replace itself with cheaper agency such an API or script. The system is vastly simplified from being able to develop the system for only one type of executional model (that of AI agency existing in all contexts that require decision-making power).
- todo: the relationship is the key, the multiplexing connection between the two aspects of every subjective POV. Level 2 is the space in which the two aspects co-evolve in complexity together, mediated by the multiplexing mechanism of level 1.
From the abstract conceptual perspective, we can see the agent as being derived from the arena, since all agents within the arena have the model of the arena in common...
but the arena is existentially supported by agency...
The common model of agency
Agents are continuous identities that have no true essence, they're a centre around which material, form, purpose and agency flow. They're simply continuity itself.
The concepts of class and instance play a core role in the structure and dynamics of a holarchy. Class and instance are usually in the form of a purely top-down hierarchical dynamic. But in the context of holarchy, they both in addition include a horizontal, or bottom-up, aspect. This is the community aspect of the system.
the common model of organisation... this is p2p, so all agents contribute their small part to the arena, but the arena is really within the agents.
To really be common, it must also be scale-independent, because as we discussed above, organisation is about recursive patterns, and so something common to all needs to be instantiated at every scale.
- arena is in the form of a dynamic scope
- there is no lexical scope concept because there is no code, only within the encapsulated agency do such concerns potentially exist
- interface: the scope is the foundation of interface, the "shared blackboard"
- each agent has a public interface (gas ability to progress a specific set of activities)
- it's a p2p system, so the arena is the total of all public interfaces (remember that "agent" covers very basic agency such as resources too)
- usage: the interface is designed to be "used" (interacted with) by agents with specific ability
- usage statistics, user stories and a test-driven approach are natural from the ground up
- like github tree
- distributed object community, all communication is between interfaces
- common market protocol
- class instance ecosystem, specialisation protocol
todo: To explain this holarchy We want to first create
- Note: this is UoD, it's general
Organisations serve as a means for individuals (which in our context include agents and processes, not just people) to work together, coordinate efforts, and achieve common objectives in a structured and purposeful manner.
Organisations are typically characterised by a defined structure constituting roles, and responsibilities, along with established knowledge, systems, tools and processes. Their activities are usually by definite purpose and often operate in accord with established protocols and regulations.
- Structure: Organizations have a hierarchical structure that outlines the different levels of authority and reporting relationships. This structure helps to define roles, responsibilities, and decision-making processes.
- Objective: Purpose, goals... organisations have specific aims or objectives they seek to achieve. These goals can range from financial targets to social impact goals, depending on the nature of the organisation.
- Resources: Organisations require resources, such as human capital, financial capital, physical infrastructure, technology, and information, to carry out their activities effectively.
- Agency.... roles
- Processes: Organisations develop various processes, systems, and procedures to streamline operations, manage resources, and achieve their objectives efficiently. This includes areas like governance, finance, human resources, marketing, and operations.
- Values: Organisations often have a unique culture and set of values that guide their behaviour and decision-making. This culture may be influenced by factors such as leadership style, mission, vision, and the collective beliefs and norms of the people within the organisation.
Class and instance
The types of capabilities and the specific groupings of them mentioned in the prior section are what we refer to as classes.
Our central reason for extended SoT legend is as a shared context within which we can describe our specific concepts for class and instance and their dynamics.
- we're talking about UoD here not our specific C&I mechanism
- already mentioned in OOP, do we need it again here?
In the context of the holarchy, evolution is much like the way an online development community works, or an online multi-user game like Minecraft or World of Warcraft. In essence these are communities centred around collaborative evolving ecosystems.
Many users work locally with various elements from the ecosystem for their various local reasons. And users share their ideas and curate their classification, which is the source of diversity and specialisation for the evolutionary aspect.
It's essential that an ecosystem that continuously diversifies include a classification system that is both organised and generally applicable.
Subjective and objective
- introducing these dichotomies in the context of common agency UoD, not our specific pattern
- outer, inner, public, private
- the subjective dynamics are defined in the objective domain, because they're common to all
- tying in with form and function in SoT
- ties in with scope too
Summary of UoD - experiential commonality of cognitive agency
- agent-arena relationship, common model of agency
- C&I, representation (form & function), purpose/ideal
- organisation (paper org and workflow)
- a common way of describing org and "working on" org allows for
sharing org and variations leading to more effective specialisation
The holon data-structure
We now have good general description of the universe of discourse, which most readers will have a good feel for at an experiential level. In this section we will discuss the details of the specific holarchy model we have in mind, by describing the within the context of our universe of discourse.
The peers do not directly represent the holarchy or holons. Those are the application of the network, the purpose they're connecting together to collaboratively achieve. In peer-to-peer networking parlance, the holarchy is called the logical network, an internal "virtual" network that can only be participated in by peers in the physical network operating in accord with the protocol.
A holon is an individual unit within this logical network that all together make up the holarchy. Each running peer maintains a representation of itself as a holon, but there are also many more holons than peers. Every individual, organisation, project or even concept is a holon. But it's the peer holons that serve as the "gateway" through which other holons can be created and maintained, because it's the peer software that implements the interfaces between physical resources and the logical holarchy within.
We have a protocol allowing peers to behave as a four-quadrant holon all maintaining the holarchy network, but we need to defined specifically what that means in terms of a data structure and processes applying to it.
The unified whole aspect of the network is only known through local representations of it. Each holon maintains its own four quadrants perspective. It's perspective of the whole (the top outward facing quadrants) is its own version of the whole filtered by its own interests and subjective experiences.
- the holon is the logical result of a running peer, the logical is the subjective POV, the energy to run the peer converts resource into the evolutionary subjective being system
The holarchy design pattern is best understood in terms of three general abstraction layers. A new abstraction layer can re-organise computational resource into new possibility space that has a completely new own system of causality. The first two abstraction layers work together to produce the class and instance system. The third layer is about defining an economic model that the basic organisational principle defined by the first two layers.
Level 1 (structure, the objective domain)
Layer 1 defines the fundamental data structure and its dynamics. A top-down and bottom up dynamic yielding the functional and structural trees we call the ontology and the holarchy.
The defining characteristic of this layer is class-instance functionality. This can also be thought of as the functionality of functions themselves, i.e. the ability for named patterns of activity to execute in a private scope, and for such patterns to be composed into larger patterns.
The class side is composed of the public interfaces of these patterns, the outward facing aspect that connects to other patterns instances, and together they form an entire ecosystem in the form of an ontology.
The instance side is the private inward-facing implementation of the patterns, which acts on state just like a normal object (in the OOP sense) with its internals encapsulated behind its public interface.
From the perspective of execution within the internal encapsulated implementation, the wider multiplexing pattern cannot be seen. Time only exists while there is executional focus, and so from this perspective it is an axiomatic phenomenon that all threads undergo change in linear time together.
By the same token, this inner perspective can also not see the opposite dynamic to the top-down multiplexing, which allows dispersed clones of the same class to be connected. This "non-local" connection is axiomatic and inherent from the inner perspective.
From the widest outer-most perspective, the system takes the form of a one-to-many hierarchy. In our model this is an inverted tree having a single root at the top. This tree is the holarchy, and its branches are classes and its nodes are instances of the corresponding class.
The same class may exist many times throughout the tree, but the same class can only appear once within any specific node. Note that this is exactly the same situation as is constrained on us by a standard associative array of key:value pairs.
todo: multiplexing, scale-independent, card dealer analogy - non-local perspective is due to full (outward) sync occurring between the consecutive moments no matter the scale.
the content (value) of each instance node from layer 1's perspective is a flat list of other nodes. We call this content the background content of the nodes. One important point to note here is that the background content of all nodes throughout the tree having the same class are synchronised.
This synchronisation is a bottom-up process that integrates the background content of all nodes from the many at the bottom all the way up to the root, and then this unified global result is passed back down in the top-down process.
The synchronisation of all instances of the same class across the whole structure, no matter whole deep and complex it becomes. This is possible because the top-down and bottom-up are vertical movements of focus which, due to the multiplexing nature of the flow of focus, always span the whole after every a full rotation of focus (todo: explain better, a full rotation of any context involves a full vertical cycle out to root).
todo: weights, salience, a fuzzy set
This outwards flow forms a second network of connections where all the classes form into groups. We could call these groups "special interest" groups since they're all groups that align around a singular topic together, defining together the salience of the flat list content.
All the classes form a tree together where each node is a fuzzy set of other classes (those that make up the meaning of the class as an idea), and set of instances of that class. This tree we call the Ontology, and it has a one-to-many structure with the root being the most general concepts that is common to all nodes, and the deepest being the most specific sparsely used ideas.
In addition to the background content, the layer 1 process also gives nodes the ability to have their own unique and private foreground content which overlays background content. This overlaying process provides the usual class and instance behaviour we're used to, where the class behaves as a template for the instance to "fill in" with specifics.
The two forms of content also gives an instance the potential to deviate from the norm of that class, adapting to the specific local needs. Either by adopting a completely novel idea, or by selecting other less common of the existing fuzzy options of the background content.
We talk about novel adjustments and selecting of options, but such things are not defined in layer 1 of the system. Layer 1 is setting up a data structure and the dynamics that would occur if there were any content, and if any of that content were to undergo change. But it's layer 2 that defines how change actually takes place.
- creates diversity and uniformity
- yields the ontology growing like a crystal through establishment of patterns
- seen as the collective unconscious, the noosphere
- morphic resonance (the non-local connection between similarity of form)
- multiplexing makes subjective POVs and linear relative time
- the process of synchronisation throughout class groups is made possible by extending the multiplexing process with an opposite bottom-up movement
- this synchronous domain is seen as a non-local connection from the perspective of any local context
This synchronisation is what enables evolution, but the source of change originates locally and subjectively involving behaviour and decision-making which is too specific for this domain. These specifics need to take place within a local context of Cartesian space and linear time, which is what the second abstraction layer (Level 2) is all about.
Class and instance
These two concepts usually belong in the context of software development, they're usually thought about in the classical OOP context whereby there's a class hierarchy defined which provides the structure and behaviour for the types of objects interacting at runtime. But there are many different forms of that the class and instance relationship can take in the context of software development, and the paradigm that most closely matches our system is the so-called "mixin" pattern.
This is where one class is not only defined as a collection of methods and properties, that can be extended by sub-classes. It also allows other classes to be combined into horizontally. Actually the parent in this case becomes a subscriber of each mixin child, so in effect they're still a class-child. The key improvement with the mixin model is that the functionality is expressed purely in terms of the instance structure allowing the classes functionality to relate purely to it's fundamental purpose of categorising functionality.
- todo: not explained well - class is not different than mixin, tie in class movements as key-change etc
Notice that this class and instance functionality in terms of the top-down multiplexing and bottom-up synchronisation, exactly matches the four-quadrant holon architecture and has a direct geometric interpretation.
- level 1 defines the fundamental rule of scope: all can access all via public interfaces, none can access the private implementation of any
Level 2 (organisation, the subjective domain)
This level is the instance aspect of the class-instance system brought about by the level 1 dynamics. Level 1 is the class aspect of the class-instance system.
- the same dynamic within the context of itself is in the form of the pattern-representation fitting loop
So far we've talked about classes as shared patterns of content and behaviour, but we have not described how exactly any kind of behaviour can be encoded in the content structure of a class.
- what is a behaviour pattern?
Multiplexing is the foundation of the four quadrants and the related fundamental concepts of agency discussed above, but it's also the foundation of organisational patterns. Organisational patterns inherit their geometric origins from the fact that they're derived from multiplexing.
- pattern as in design pattern and behaviour pattern
- a kind of programming, but aimed at general logistics
- in a common geometric form, not a symbolic system
- easy to understand, adapt and share
- patterns are in the general form of query:action loops
- representations are structures of such loops
Organisations can exhibit any complexity. But no matter how complex a system it is, it can be described entirely in terms of the aforementioned general organisational constituents.
There is no specialised symbolic logic necessary for accurately defining complex organisations, because organisation as a concept assumes agency. A role is defined in terms of specific capabilities required from any agency that performs in the role. The organisational definition is not required to cover the functionality of any agency, it is all encapsulated behind its public interface.
Since organisation does not define any specific agency, and the internal aspect of agency is encapsulated from the public, all organisations are either resources, agents or other organisations. And actually in the context of the holarchy, resources are considered as agents (they have specific capabilities).
Organisation is all about the establishment and usage of the behavioural and structural patterns of agency. This organisation of agency is formed from such general concepts that it can be included in a foundation ontology.
But in the subjective world agents have a definite implementation, a subjective self representation. The representation of the SoT is the actual ship and the actual workers and their specific state, history and plans.
This is a concept we can easily understand experientially. Each one of us has an abstract informational representation of ourselves already in the form of our files, subscriptions, accounts, calendars and contacts etc. This is exactly the role of an agent's representation as well, except that it's a single unified dynamic structure.
- The structure of the representation follows that of the ontology...
- Fitting loops....
The evolutionary pattern
Remember that our universe of discourse includes the idea of the community centred around a collaborative ecosystem of classes and instances. Now that we know the specifics of the classes and instances, we can expand on this basis to describe the details of the evolutionary dynamic.
- this just follows on from C&I + OP
- non-local allows the focus of non-local experience onto the local
- the capturing of local specialist knowledge (solutions to problems)
- here the bottom-up synchronisation pattern yields evolution from local adaptation
- add the section of knowledge in society here?
Layer 3 (harmony)
Layer 3 then defines a specific application which opens up a yet another new abstraction layer also introducing it's own new system of causality.
Layer 2 introduced the fundamental time and work related concepts that underpin the concept of agreement. In layer 3 these are extended into a higher level of organisation to yield contracts for resources, value assessment, services, and quality of service. This also includes information (think "information market") accuracy and objectivity where the authority aspect of an information source is the reputation (reputation being a landscape over the ontology).
- market and organisation (self-as-market-participant and self-as-organisation)
- harmonious organisation (balance between the two behaviours, balanced value exchange)
Layer 3 is all about reciprocity in real value exchange. This concept requires credit in terms of a unit of account.
|Money is merely an abstract representation of the real credit of the
community, which is the ability of the community to deliver goods andservices, when, and where they are required.
|— Major Douglas|
- but how do you make it that?
- centrally doesn't work
- but the bottom-up price-setting mechanism allows any abstract number to serve as a medium for fitting purchasing power (agency) to productive capacity (resource)
Credit is a divisible abstraction of productive capacity
- total purchasing power matches total productive capacity
- a varying distribution over the network
In summary, layer 3 extends the ability to organise logistically provided by layer 2 into a definite organisational pattern. This pattern forms a new abstraction layer, or arena, in which all the participants are organisations interacting with a common protocol that results in a harmonious evolving society.
Before moving on, let's do a quick overview of what we have so far. First we introduced the concept of a "holarchy" as being a fundamental pattern of nature common to all life and conscious agency. It takes the form of a community of so-called "holons" which each have two behaviours to support both themselves as individuals, and support the holarchy as a whole.
We described our version of the holarchy here at Organic Design as being about implementing this fundamental pattern of nature into human society by creating a new internet protocol. Furthermore the peer-to-peer networking model was shown to be a perfect architectural match with the holarchy model, because the unified peer definition includes both client and server aspects, and these are isomorphic with the holon's self-assertive and integrative behaviours respectively.
- the model is four quadrants
- class (organisational patterns) and instance
- evolutionary ecosystem
Holarchy (July 23 notes to incorporate)
- refining the UoD into a definite design
- No SLH, we're in the context of comms and agency now
- Communications pattern
- an organisation involving general agency is a comms pattern
- p2p version of BB (identical but BB is formed from local public interfaces)
- and SLA in the public realm of public interfaces
Now that we have our UoD in place, we're in a position to refine it into the specific terms of a data structure and its dynamics. - We'll talk about these dynamics in the general terms of agency - encapsulation - the implementation of agency is private (complexity hidden)
- here we want to talk about the organisation in relation to the SoT
- it is an instance of a class, and an agent in the arena - it's a member of the class group (dynamics explained below) - it has purpose and agency - it has a representation that grows between purpose and agency - self representation, class representation
- roles, public interfaces and communications patterns, messages and reports - workflow and workload - and that there is a general pattern of operation common to all organisation - being common to all, it must be defined in the objective domain,
it is in fact the same pattern, but in subjective extending the objective as a class
Organisations can be arbitrarily complex in width and depth....
But the hierarchy, and therefore the complexity, is actually illusory. In functional terms there is no difference if we convert the structure of roles, departments, tasks and functions into a horizontal society of "micro organisations". In other words every organisation is a simple group of agents performing a small set of activities with no further complexity within.
It's clear that the complex organisations and constituents like departments are simply wider and deeper structural patterns of various combinations of micro-organisations.
What then is the commonality amongst all these micro-organisations?
The ultimate purpose of the representations is to represent our entire continuously growing and diversifying informational life.
Maintaining this structure is very administration intensive, due to the great variety of protocols and applications that need to be connected to. One solution would a huge active user base that collectively are able to respond to these changes in a timely fashion. Due to the organised nature of the holarchy, these adjustments made in one local context immediately becomes available to all instances of that same class of context.
But the other solution is that it will soon be a very practical job for LLM AIs, they are very good at connecting protocols with minimal assistance in nearly all cases. The foundation use that the holarchy will have for AI is as a representation administrator.
The idea is not that AI has to actually sit there making connection regularly to keep the local data up to date. A representation structure should have instantiated all the necessary connections in the most resource efficient way it can, which would most likely be using an available language in the system such as C or Python. The administration work involves ensuring these connectors are functional.
AI opens up the possibility of practically maintaining representations involving diverse non-standard connections like scraping websites or accessing applications on behalf of users.
- the state of an organisation - not just instance actualised state in linear time,
but also connection implementations
- in the form of ontology
- what are they: the need for process definition and sharing - workflow: think agency-oriented programming - specifics of pattern defined below
Class and instance
- tie in the general class instance concept with SoT
- all aspects in flux
- the BG and FG content - introducing the splitting of agency and resource
A scale-independent means of distributing work across continuous agency - yields the subjective and objective domains - is a top-down process, that permits a bottom-up process - the bg and fg
- simlinks, clones, synchronisation - the bg and fg gives BG options
- the holistic aspect of the instance tree
- the non-local groupings in more detail
- instances - now we have the UoD we start defining our specific design pattern - the ontology and holarchy - the things common to all organisation regardless of its purpose - all purpose is subjective specifics, only organisation itself is objective
- because responding to a message involves process, workflow
- activity stream (messages) lead to activity (work), loop: work leads to messages, messages lead to work SLS: state represents workload for various roles representation is the abstraction of state, including performance, goals and workload - its all messages-oriented, so the complexity of an organisation is based on the depth and width of its structure, it's all patterns of roles, public interfaces, messages and activities - work is ito resources, materials (state and roles)
The system we've described with this extended Ship of Theseus concept is quite a specific a class and instance system - we haven't described anything about how it might be technically modelled, but we have described a concept that has enough specificity that we could define a class and instance that exhibit all the same features.
It turns out in fact that there's a very simple system model that represents an executional environment which behaves in exactly the way described by the extended Ship of Theseus. The core of the system is a class-instance relationship which is used recursively to represent arbitrarily complex systems. Agency within the structure is the source of change. This agency throughout the whole system is the source of evolutionary change, leading the whole structure to continuously evolve towards ever more diversity and complexity.
- todo: actually there is one more extension we should include: non-local connection between instances of the same idea (morphic resonance)
- don't have to believe in it, but I include it because the data structure has the quality that all subjective perspectives include the appearance of non-local connection between same idea
- structural and functional?
Three layer model (2022 notes to incorporate)
Layer 2 (organisation)
This new layer is the perspective from the inside outwards. From one of the many continuous threads within the multiplex data structure introduced in the first layer, out to its sibling context and parent.
The defining characteristic of this "inside-out" layer is that it enables the new perspective of linear time within which self-organisation through feedback is possible.
A self-organising system is a system that undergoes change in accord with it's own state and structure. In terms of our data structure, this makes a logically disconnected abstraction layer. Although this layer depends for its existence and operation on the first layer, this layer exhibits a brand new source of causal potency that is completely independent from the first layer logically. All causal chains and cascades are determined only by the interactions of these new layer 2 linear-time feedback structures. We say such a structure is a representation of a pattern.
Patterns are the layer 2 version of the layer 1 class concept. They can be thought of in both in the sense of behaviour patterns and design patterns. It's easy to see how these two forms of pattern also correspond to the concept of class.
...leading to a new causal foundation in the form of patterns and streams of activity. Classes in this context become packages of activity streams constituting an ecosystem of patterns, and a manifest structure of dynamically fitting representations.
The patterns introduced in this layer embody a scale-independent paradigm of processing enabled by the multiplexing. These patterns can contain any complexity of parallel and serial thread structures.
Note: It's important not to confuse the hierarchical difference between layer 1 and layer 2 with the hierarchy of the multiplexed data structure itself. All the layers of abstraction are present within every node of the multiplex regardless of depth.
This is a declarative paradigm of selection and action in which both sides are free to become arbitrarily complex. Selection and action form a feedback loop, because the selection side is about assessing potential work, and the action side is about reducing that potential.
This loop construct of selection and action is with respect to self in linear time. This dynamic is quite similar to a CSS document (pattern) interacting with a DOM structure (representation). The selection is continuously fitted to the document, i.e. the appropriate selectors always apply even when DOM structure changes dynamically. The selectors all have associated rules which are analogous to our actions on this example.
But CSS cannot make changes the DOM, it's purely in the presentation layer. In our system, actions make up the implementation of the pattern, so they act on the representation, thus closing the loop for feedback between the two sides.
- This also how it can be a self-organising system changing in accord with its own structure. But note that layer 1 also undergoes change (maintaining the C&I multiplex) in accord with it's own structure (in a dynamic independent of linear-time), so layer 2 is extending an existing self-organising system.
The general high level structure of level 2 operational loop is the top-down "fitting" of the local representation to the current consensus state and to the class. And the bottom-up process of allocating energy to the appropriate set of class to act on the current situation (fitting the salience landscape to the representation).
The subjective POV with its patterns and representations in linear time, can instantiate a set of general concepts such as work, cost, expectation, performance, reputation etc. Instantiation is the knowing of something in a participatory way, it's a concept that has become embodied in your own patterns of behaviour.
In summary, layer 2 introduces the organisation-ability (a.k.a the common logistic), basically the ability to work with patterns and representations in the subjective linear time threads provided by layer 1.
- class is used mixinly, instance is a runtime-loop-mixin which is a representation
- representation structure exists and evolves causally in layer 2 (but depends existentially on layer 1), the structure can only evolve or undergo any kind of change at all in layer 2, because it's formed from subjective meaning
- explain ito queries vs indexes (representations are maintained, connected to activity stream)
- this is the libre society
- after libre software, but "not meaning free as in free beer" in this case is meaning that there is still a monetary economy (with free market money), but all knowledge is transparent and understandable - reusable and adaptable
- the representation represents both current state and the pattern (i.e. its a representation and a representative)
- it publicly represents both of those aspects,
- and it operates internally within that public context
- in philosophical terms layer 2 can be thought of as the cartesian world
The first two layers together give us our sort of "foundation machine", a sort of basic Turing machine in the form of an interacting network of organisations in a shared arena, or "multiplex".
Note that our original universal interface concept is legacy now. Only the universal ontological representation is needed, because AI agency is the only interface necessary for the representation to be connected directly to. AI can then present any aspect of the organisation in any desired form.
Although AI is overkill in the vast majority of organisational contexts, it can replace itself with cheaper agency such an API or script.
By making AI agency the default option for interacting with the representation we can define the entire system in terms of generic agency.
The ultimate goal of the evolving Universal Interface would have been human like agents that can be communicated with naturally and be rendered in terms of Simms-like "microworlds". But since AI is already at such an advanced level, there is no longer any need to develop this thread, it's the shared ontological representation that's key.
Most interaction with our ontological representations would be dealt with via agency with natural speech. The default means of presenting information would be diagrams, tables and charts in a presentation format possibly with a agent guidance. The presentation would be interactive using natural speech for feedback to the agent. The typical Youtube learning channel format - a presenter talking about concepts being presented in the background which are graphically representing the current points the discussion is focused on.
Knowledge in society
Knowledge is dispersed among many individuals and organisations throughout society. No single individual or central planning body can possess all of this knowledge, which includes detailed information about individual preferences, local conditions, and technical procedures. A free-market uses the price mechanism for communicating this dispersed knowledge.
But price signals only communicate the knowledge that is specific to the resource allocation aspect of society. There is also specialist knowledge, cultural knowledge, institutional knowledge etc forming an information market. This is a slightly different kind of market because it's not scarce (in a "libre society" artificial limits are not placed on knowledge by treating it as property). But we still need something to play a similar role to the price mechanism that allows us to capture these other forms of knowledge and allocate them optimally to where they'd be of potential benefit.
The price system works because of two important things, first it's a common unit of account by which all people value their time and resource. Second, people are free to value things in accord with their own self-interest.
Side note: The latter is not a promotion of selfish behaviour, it means to communicate honestly and transparently about that which is best primarily for the local circumstance. This is a very important point to understand deeply. It actually benefits the whole specifically by being what's best for the local circumstances. Hayek's "use of knowledge in society" essay is a very well articulated discussion about this concept.
This dynamic is the very essence of evolving specialist knowledge in society. What's damaging is withholding the knowledge gained through self-interest. Any limitation on knowledge is limiting the prosperity and understanding of the whole society, just in the same way that hindering the propagation or transparency of price information is detrimental to the free market.
A mechanism that enables the full distribution of knowledge in society also has the same fundamental requirements as the price system. Instead of a unit of account agreed upon by all, we need a structure of knowledge agreed upon by all, a shared unified ontology. And instead of being free to simply communicate prices of things to best benefit our local circumstances, we need to be able to use and adapt any aspect of the ontology to work best for our local circumstances.
For a society to realise a unified ontology that captures local knowledge optimally, knowledge should be a completely transparent public commons that's understandable by all. At Organic Design, we call a society that upholds this value a "libre society", where the word "libre" is applying to knowledge in the same sense as how the libre software community applies it to software.