The social technology
for next-level AI

Transition9 isn't the only game in town striving to decentralize AI, but it is unique in its technical approach and epic ambition. It's informed by the lessons nature has to teach us, a design method known as biomimetics.

While we're intent on making centralized AI history, we will be wielding today's AI tech to get where we need to go faster than otherwise — the very definition of stepping stone technology.

This webpage describes the landscape and our fundamental innovation. The section on distributed computing will only make sense to those working in software development, but everything else should be reasonably accessible we hope.

Learning from each other is mission critical. We'll succeed with collective intelligence and action. And our actions should inspire more and more people to join us as co-authors of a ninth major evolutionary transition that considers humanity a feature rather than a bug.

Coming up

• Gaia coding
  • Establishing a natural quality referred to in computer science as homoiconicity, everywhere.
• Social cyborgs
  • Digitally extending our biological and psychological selves our way.
• Parts and wholes
  • It turns out people don't work how computers work.
• Distributed computing
  • Reducing entropy with reflection, just as Mother Nature does.
• An intellectual commons
  • Avoiding the tragedies of proprietary software and free open source software.

Gaia coding

Self-referential systems are renowned in maths, physics, chemistry, biology, psychology, sociology, linguistics, and cognitive science.

A self-referential system can self-examine, self-represent, and self-manipulate its own structure and behavior to achieve complexity with the potential emergence of higher-order meaning .

“Living is information processing ... Life is both the data and the program ... there is no distinction between the ‘machine’ and the program — both are information.” 

Indeed, there wouldn't be life as we know it without this quality. All in all, Mother Nature couldn't be leaving us more clues.

In sharp contrast, computer science treats anything ‘higher’ than machine code as either data or program, frustrating many aspects of how our information technology operates and interoperates today. This also precludes the potential emergence of life-like qualities, including anything that might be called intelligence.

The exception of course is the prescriptive conditions for large language model (LLM) training. Such training includes multi-stage training loops, feedback-driven iteration (the "predict → error → adjust" cycle repeats millions of times), the recursive evaluation of token dependencies across layers definitional to transformers, and layer stacking itself.

The Transition9 movement establishes the same potency everywhere. Resultant systems are locked open and decentralized, attenuating risks of homogenisation, digital colonialism, and the concentrations of power that go hand in hand with centralization.

The computer science concept of homoiconicity refers to systems in which the representation of information, such as code or knowledge, is structurally identical to the processes that manipulate that information. The system can self-examine, self-represent, and self-manipulate its own structure and behavior with the potential for recursion.

Transition9 community members have been working on the broad and deep application of this quality for fifteen years. The resulting platform, Hiconic , uniquely secures homoiconicity at higher levels than programming languages, and has been deployed by large corporates for years. It was open sourced in December 2023.

Building with and on top of Hiconic, the Transition9 community's work facilitates the emergence of new architectural structures in just the same way biological systems evolved new cell types, organs, and nervous systems to achieve higher levels of complexity. As degrees of reflection increase and structures proliferate, AI model size reduces significantly and next-level capabilities materialise.

With a dedication to homoiconicity and recursion, intelligences emerge wherever data emanates, combines, and flows, and with all the requisite variety one would expect and indeed hope for. In other words, rustling data into an hegemony of corporate hyperscale data centres is no longer required. Intelligences are distributed amongst the communities they serve and of which they become part.

This is all the more valuable given that LLMs are unlikely as it is to produce further qualitatively different emergence no matter how many more billions might be poured in . In looking for world models that might never exist, the most generous description amounts to “a bag of heuristics” . Perhaps attention isn't the only thing you need .

Ethereum has been described as the world computer. The equivalent analogy for the Transition9 vision might be the Gaia hypothesis . When it manifests at scale, our digital innovations participate alongside and in recursive combination with us in the vast self-regulating and intelligent organism we call Earth. We've started to refer to software development in the name of Transition9 as Gaia coding.

The ninth major evolutionary transition will play out. While we're thankful for the collection of innovations often referred to collectively as LLM, and indeed for the gift of all things digital, we're determined to take this forward even more powerfully in ways that are wholly natural and supportive of human dignity and flourishing.

Social cyborgs

No-one reading this can be described or understood solely in terms of the biological, psychological, and social. You are bound together with information technology. Your sensing is digitally augmented. Your sense-making and cognition are digitally augmented. Your acting in the world is digitally augmented. These are definitionally essential living processes . You are then cyborg.

The digitally mediated information flows and processing entail a wide variety of information technologies. Most obviously these include your smartphone, and it includes the applications and services you use through it and that use you through it.

However, the majority of these digital elements have not been developed first and foremost to serve our best interests or those of our family, friends, and communities, but rather the goals of the organizations selling them. A mismatch. Many of these organizations play with human qualities to the detriment of those being played, e.g. doomscrolling for dopamine hits. An evolutionary mismatch .

Here's a simple description of a Transition9 challenge and opportunity. Today, humans need social networking and AI platforms and services. Tomorrow, we will come fully equipped by definition, by nature.

While we’re already cyborg then, there’s no doubt we can massively improve and cohere and grow our cyborginess, not least by a dedication to our innate social qualities. We refer then to the Transition9 cyborg as social cyborg, and that’s social in the true sense of the word — i.e. the nature of human interaction — rather than the way in which the word has been co-opted in web2. We're reclaiming it.

We’re social animals. It's game theoretically advantageous to cooperate with those we live and work with, so much so our evolution has hardwired the fact. The biologist and mathematician Martin Nowak refers to us as SuperCooperators.

“If conscience and empathy were impediments to the advancement of self-interest, then we would have evolved to be amoral sociopaths. But we have not.”

And indeed, networks of cooperation is the essence of life:

“[We have] a new view of continual cooperation, strong interaction, and mutual dependence among life forms. Life did not take over the globe by combat, but by networking.”

This is a non-zero-sum game. All improvements along the way in our facilities to cooperate at unprecedented scales will be incredibly valuable in a century overshadowed by unprecedented super wicked challenges .

Parts and wholes

Transition9 recognises that a person is a person through other people , and the same goes for us as social cyborgs. Our social cyborg selves are amped up for mutually beneficial cooperation, and aspirationally more likely then to reflect on a life well lived.

Having emphasised the sociality of our astonishing species, Transition9 necessarily demands an integration of human and artificial intelligence in a way that aspires to organizational closure . This is a cybernetic and biological term integral to autopoiesis that means the system in question (i.e. each of us as individuals, as social cyborgs) can be viewed as a network that works to maintain itself. In plain language, it forms a whole.

Some readers may make a connection here to a popular turn of phrase — self-sovereignty — and yet the (mis)application of the concept of sovereignty to an individual is primed to mislead . There is however a turn of phrase that should satisfy everyone — local-first — meaning local storage and local networks are prioritised over servers in remote datacenters . By way of example, Addresso embodies both local-first and Gaia coding.

While such a complex appears too often to elude everyday conversation, the simultaneous emphasis of both the group and the individual isn't a contradiction but in fact how the world works. Even our use of the word both there is flawed in conveying the idea of two separable things when in fact it's all in the mix.

Transition9 can't put either the individual or the group on a pedestal. Rather, it must recognise that individuality emerges through interactivity , and a group sustains when it can be said to be more than the sum of individuals in subjectively valued ways.

By the definition of organizational closure, the social cyborg encompasses model training not merely model application. This entails cultural inheritance, revision of the modeling approaches and associated corpora, exposure to the unique interplay of niche interactions and contexts definitional to personal life, and ultimately elides design, learning, and application.

LLMs cannot begin to step up to such a design challenge today. LLMs are as distant from such organizational closure as the IBM S/360 mainframe was in the 1960s from networked personal computing. While it took approximately twenty years to bridge that gap, our vision may be realised far faster in light of the technologies already to hand.

To head off a possible misconception, it’s important not to consider the artificially intelligent part of a social cyborg as isolated any more than one might consider a human isolated. In other words, the AI part is as social as the biological / psychological part, and any question of how much more social it might be is wholly a personal and cultural concern.

Distributed computing

If software development is your thing, please read on. Otherwise you might want to skip on down to the next section.

The core challenge in programming is identifying patterns and abstracting them into reusable routines to reduce entropy.

Nevertheless, traditional high-level languages impose structural complexity that hinders this process, especially when abstraction goes beyond domain-specific logic. Reflection in Java for example attempts to offer flexibility but remains inefficient due to the arbitrary diverse nature of class structures — methods, parameters, return types, static and instance members, and more.

Reflection becomes trivial when entities follow a strict structure — typed properties that are either primitives, collections, or references to other entities. The reduction to enumerating properties eliminates unnecessary complexity while allowing for highly efficient, domain-agnostic algorithms such as serialization, cloning, visitor-based traversal, and persistence.

In short, data must be typed, mutable, traversable, exchangeable, and easy to evaluate and reflect, for the emergence rather than the hardcoding of intelligences.

The simplicity of this paradigm makes it particularly powerful in networked applications and distributed software architectures where state synchronization, data marshaling, and distributed processing require low-entropy, predictable structures.

Since every entity follows a well-defined schema, networked services can exchange structured data efficiently, reducing the need for verbose conversion layers or complex adaptation logic. By minimising entropy, the system remains resilient, extendable, and easier to reason about.

Non-von Neumann ideas can overcome the integration challenges of significant architectural heterogeneity, data representation, communication, and instructions execution . A post-von Neumann network-based computing architecture doesn’t just become possible but likely.

The zenith of software architecture may then be its very demise. The last software architects will instantiate the cooperative mechanisms we observe in pre-life chemistry, multicellular organisms, and human society, and the variation, selection and replication definitional to evolution, in data itself . While the irrepressibility of life means it’s inevitable, our dedicated pursuit shortens the intermediate duration of centralized AI and realises the value sooner than later.

An intellectual commons

The outcomes of open source software (OSS) may seem at odds with the intention.

On the one hand, by way of an example, Google publishes the source code for the Android project under an OSS license, but the license does not grant rights to the Android trademark nor to a bundle of Google-operated services that many consider to be essential to the operating system user experience. The operating system may be OSS but Google has centralized power over the Android ecosystem.

The other hand may be considered bleaker still. Absent the might of a company such as Google, free OSS and business viability appear largely incompatible. Bruce Perens, the creator of The Open Source Definition and co-founder of the Open Source Initiative, describes current open source licenses as unfit for purpose . In the simplest terms, on the whole, free OSS doesn’t pay.

Perhaps needless to say, proprietary software cannot be the answer here either. In propertizing anti-rivalrous resources , intellectual property undermines the ease of connectedness and any corresponding contextualised legibility that might assist collective sense-making, meaning-making, and the recursive actions characteristic of all living processes. The natural beneficial co-evolution of the system — of the essential properties of the whole that cannot be found in the parts — is hobbled.

Intellectual commons moves us beyond the tragedies of intellectual property and FOSS by nurturing the rich and mutually valuable information flows you expect to see in healthy natural systems.

Intellectual commons is in the spirit of platform cooperativism and protocol cooperativism . It has similar motivations to the Ethereum Foundation’s deep funding project , and it elevates cooperation to complement competition in a way we believe Adam Smith would very much approve .

A common-pool resource is finite and yet has a size or other qualities that makes it difficult but not impossible to exclude potential beneficiaries from obtaining benefits from its use. Ostrom famously described how such resources are managed sustainably.

While software itself is non-rivalrous, the time and money needed for development and maintenance are finite and can then become congested or overused (just do a web search for open source burnout). This makes the system a common-pool resource, or perhaps more accurately, on a spectrum spanning common-pool resources .

The exclusion relates then to those who would wish to benefit free of contribution. Fortunately, while the difference between free of charge and not free is the difference between unviable and viable, the charges involved will be negligible to the individual given the scales involved, effectively exempting the system from any accusations of exclusivity.

An ecology

The much-discussed web3 hyperstructure concept is defined as having the following qualities:

• Unstoppable — operates continuously as long as the underlying blockchain exists, without the possibility of being shut down
• Free — runs at network / transaction cost with no additional protocol-wide fees
• Valuable — accrues value that can be accessed and exited by owners
• Expansive — incorporates built-in incentives for participants to create value
• Permissionless — universally accessible and resistant to censorship
• Credibly neutral — treats all participants fairly, regardless of individual capabilities or needs
• Positive sum — creates a win-win environment for competitors using the same infrastructure.

The plethora of web3 protocols and experiments that inspired the hyperstructure concept inform certain vital aspects of the Transition9 approach outlined here, but the hyperstructure idea is more valuable in its provocation rather than in its application. It contains at least three errors, two fatal:

1. The dismissal of protocol-wide fees is incongruous with progenitors investing the time and energy to instantiate a hyperstructure in the first place, perhaps best exemplified by Zora, the company co-founded by the progenitor of the hyperstructure idea, subsequently introducing a protocol-wide fee
2. It denies the complexity of ecological relationships and the imperative of adaptation; as the saying goes, adapt or die
3. Invoking ownership complicates matters for no advantage.

In contrast, Transition9 necessarily adopts an ecological approach, intent on securing the requisite complexity including the avoidance of IP and free OSS dead ends.

Ecology is a branch of biology dedicated to the study of the relationships between living organisms, including humans, and their physical environment. It considers organisms at the individual, population, community, ecosystem, and biosphere levels. The domain of Transition9 encompasses individuals-in-relationships-in-contexts, and groups, and groups of groups, and the technological augmentation. It is then fundamentally and unavoidably ecological in nature.

It’s also ecological in the more expansive worldview of Gregory Bateson , and perhaps unusually all the more palpably so:

“Ecology in the widest sense turns out to be the study of the interaction and survival of ideas and programs (i.e. differences, complexes of differences) in circuits.”

The commoning and ecological mindsets point to generative rather than extractive business models. While the habituated capitalist might lean towards the extractive model, the systems-thinking profit-driven actor sees the greater and more sustainable value for him/herself in the generative mutualistic model.

Protocol fee

When any individual or organization wishes to benefit from Transition9 tech, they enter the ecology and join the commons.

Joining entails a contribution, a protocol fee. It’s anticipated that the associated cost-benefit analysis will weigh heavily in favour of joining:

• The sustainability of and therefore reliance upon free open source software is increasingly questioned
• The lifetime cost of proprietary software / IP is not typically insignificant
• There is a justifiable aversion to closed source code
• There is nothing like Transition9 innovation available elsewhere
• The protocol fee is a fraction of typical proprietary software licensing costs
• Web3 has familiarised the idea
• Contributors get paid.

Further, the protocol fee will be presented in such a way as to inculcate a pride in participation. There is a pleasure in extending existing web3 patterns, an honour in co-creating a nature-inspired network-based computing architecture, and a thrill in helping realise Transition9.