The Gaia Attractor

Rafael Kaufmann
22 min readApr 17, 2023


A planetary AI copilot network to overcome the Metacrisis

Source: The World Economic Forum.

The Metacrisis is upon us. Humanity is recognizing the total risk we face as a species and planet, created by self-increasing technological capabilities, the “Molochian” self-reinforcing economic drive for short-term growth and financialization at the expense of sustainability, safety and equity, and the feedback loops between these factors. As argued by Daniel Schmachtenberger, popularizer of the term “Metacrisis”, the most likely outcomes are either some form of chaotic breakdown, or the development of powerful control systems in response, leading to oppressive authoritarian control. It is useful to think of these outcomes as “attractors”, or centers of gravity for our world’s path through history.

Schmachtenberger and others seek a “third attractor”: a globally beneficial outcome with enough gravity to could change our path — with the help of intentional planning, coordination, and effort. The “old order” of 20th-century capitalism, with all its failures and discontents, is no longer a feasible center of gravity. And although many attempts are being made to achieve global good — perhaps most notably the Sustainable Development Goals — there is a widespread lack of clarity about the character of the attractor they are aiming towards. This leads to insufficient coordination, factionalism, mutual accusations of wishful thinking, ambition gaps, and a general sense of Sisyphean effort and hopelessness.

We argue that there is indeed a third, win-win attractor, which we call Gaia, and it is within reach from where we are. It represents a truly different paradigm than the “old order”, yet reconciles seemingly opposite movements, such as degrowth and eco-modernism, or regeneration and YIMBYism. And the same energy boost from exponential technologies and capital, if well harnessed, can drive fast movement towards the Gaia attractor, leading to substantial positive-sum gains within a horizon of years, not decades.

Indeed, there is a concrete initiative, feasible with today’s technology and knowledge and already in progress, that will drastically accelerate and focus this trajectory shift. It involves creating a decentralized network of AI assistants to augment human collectives, based on principles of complex adaptive systems, open science, and collective intelligence. This AI-human supermind — the Gaia network — will empower humanity to scale coordination at planetary scale, lock in win-win outcomes, and safeguard against Molochian dynamics. It is our best hope to overcome the Metacrisis.

I. Children of Moloch and Gaia

Robert Wright’s book Nonzero: The Logic of Human Destiny argues that all evolution, whether biological or cultural, tends to produce non-zero-sum games of increasing intricacy and scale: relationships that reward cooperation and integration. In order to reap the benefits, players — whatever kind of entity they may be — evolve information processing and physical structures, “technologies” that form foundations for the next layer of non-zero-sumness. As Wright puts it: “Whether you are a bunch of genes or a bunch of memes, if you’re all in the same boat you’ll tend to perish unless you are conducive to productive coordination.”

This process of compounding, hierarchical non-zero-sumness continues up to the planetary scale and provides the “microeconomic foundations” to the “macroeconomics” of James Lovelock and Lynn Margulis’s Gaia theory. “The entire 3-billion-year evolution of plants and animals is a process of epigenesis, the unfolding of a single organism. And that single organism isn’t really the human species, but rather the whole biosphere, encompassing all species.” **Wright goes even further and argues that “the human species — not to belittle the job — is just the biosphere’s maturing brain.” According to this view, we are not only children of Gaia, but also its stewards.

We then give the name Gaia to this attractor, the tendency for life on Earth, including humans, to evolve towards a single pattern of increasing richness, structure, interdependency and prosociality. But however strong this general tendency may be, it is just that: a tendency. It does not provide a guarantee of positive-sum outcomes in any specific trajectory. Players may well choose to defect, and game theory shows that it’s often rational to do so, given the context. Humans’ strategic rationality renders them prey to Moloch, the god of coordination failures and multipolar traps. If Gaia is the invisible hand that systematically creates win-win opportunities, Moloch is the invisible hand that makes humans systematically forsake those opportunities for lose-lose outcomes.

We can picture Moloch as a parasite that coevolved with Gaia and feeds on its bounty — a natural force that has “existed” since life on Earth began, will be around forever as well. But just like parasites of the biological kind, Moloch doesn’t usually endanger the entire system: regulatory feedback loops keep it in check. Indeed, as most economists rightly point out, Molochian dynamics of competition and self-interest do often create positive, perhaps unintended consequences: the development of technologies and political economic institutions that have enabled significant positive-sum outcomes like longer lifespans, increased calorie availability, and renewable energy sources. And in the long term, as humans learn to exploit coordination games, they generate evolutionary pressure to develop more resilient (antifragile) games. (If Moloch is the god of coordination failures, Gaia is also the goddess of learning from failure.)

What’s different about the Metacrisis, according to Schmachtenberger, is that Moloch is supercharged by exponential technology: not only are humans able to play out Molochian strategies of extraction, commons overgrazing and externalizing costs and risks faster than ever, but the logic of compounded returns at the heart of financial capital requires them to continuously accelerate the scale and speed of these dynamics, driving the development of increasingly sophisticated software and socioeconomic technologies to enable that. From that perspective, even current AI systems are Molochian in origin, having been developed by wealthy tech corporations to deepen their competitive “moats” and currently being deployed at an accelerated speed in extractive areas such as ads.

This network of reinforcing feedback loops presents a dire picture of our future global path, but the good news is that nothing about that path is unstoppable or irreversible. As mentioned above, Gaia is very good at inventing regulating mechanisms that keep exponential dynamics in check. Most relevantly for the case at hand, as mentioned above, we have a wealth of experience as a species in creating positive-sum dynamics out of negative-sum ones: notably, information technologies and socioeconomic structures that augment our intelligence, helping us disseminate knowledge, build trust, and safeguard our behavior and decisions, and allowing us to minimize Moloch’s destructive influence.

II. Superminds: the ultimate exponential technology

Ever wonder why you’re so often unimpressed by humans and yet so blown away by the accomplishments of humanity? It’s because humans are still, deep down, those people on Planet 2. Plop a baby human into a group of chimps and ask them to raise him, Tarzan style, and the human as an adult will know how to run around the forest, climb trees, find food, and masturbate. That’s who each of us actually is. Humanity, on the other hand, is a superintelligent, tremendously-knowledgeable, millennia-old Colossus, with 7.5 billion neurons. And that’s who built Planet 3. — Tim Urban

Our ability to navigate between Moloch and Gaia is fundamentally tied to our intelligence, both as individuals and as a collective. Intelligence allows us to envision novel strategies, ponder the consequences of decisions, and create new tools that further augment our intelligence. However, individual intelligence has limits, and coordinating collective efforts is often hindered by conflicting interests, information asymmetry, and bounded rationality. To successfully navigate between Moloch and Gaia, we need to overcome these challenges and enhance our collective decision-making capabilities. The first category of enhancement is “wetware-based”: Collective intelligence structures such as networks, hierarchies, and markets help facilitate collaboration and coordination. By understanding and correctly leveraging these structures, we can create more effective systems for collective problem-solving and decision-making. Collective intelligence is further radically augmented by software-based structures — symbolic systems from writing all the way to the Internet, which have created exponentially more powerful “superminds.” By integrating advanced information technologies like AI and decentralized systems into our collective intelligence structures, we can radically enhance our ability to process, analyze, and act on information more effectively, as shown in Thomas Malone’s book “Superminds”.

By embracing and developing this ultimate exponential technology (the “Human-Computer Colossus”, to borrow Tim Urban’s terminology) we can create superminds capable of navigating the complex landscape between Moloch and Gaia: our only path out of the Metacrisis.

III. Envisioning a world of Gaian AI copilots

In a nutshell, the only way to reverse the escalation of compounding risk is bottom-up: by creating and deploying intelligent systems that help people, communities and organizations at every single decision frontline mitigate risk and create shared value, while coordinating and learning from each other to internalize externalities and improve themselves exponentially faster. The logical conclusion of this principle is the Gaia Network, a decentralized AI supermind backed by the sum of the world’s knowledge and data.

This is a radically different vision of the future, so it is useful to make it more concrete by having examples in mind. Rather than give you hypothetical examples, we will focus on an example drawn from our work at Digital Gaia. Since 2022, we have been designing a framework and operating infrastructure for decision-making copilots in the domains of agriculture and land-based natural solutions. What follows is a concrete scenario that will be feasible in the mid-2020s, inspired by a true story.

Fatima is the leader of a community farm in Mexico, and has been worried that fertilizer supply has been hard to secure, while prices for her main crops like maize have been too volatile to ensure an income for the farm. Both her suppliers and the traders who buy the produce blame it on the Metacrisis… She asks Alice, her town’s local agronomist, to develop a plan for improving the farm’s profitability and reducing the dependency on fertilizer supply.

Fatima grants Alice full permissions to her farm’s node on the Gaia network, one of 600 million nodes just on the agriculture subnet — each “responsible” for copiloting a specific farm. These lightweight nodes run securely on the farmers’ mobile devices, acquiring external resources and data when necessary. They contain public satellite imagery, weather, and other open data, as well as private data like photos, bookkeeping notes, and the records of the community chat group. They are also full generative models or “digital twins” of the farms, based on state-of-the-art science and local knowledge that’s been aggregated and distributed over years.

Using Gaia’s analysis engine, Alice quickly spots an interesting phenomenon: some of Fatima’s heirloom maize fields are extremely productive, even in a nitrogen-poor soil and using very little fertilizer! Maybe that maize field has something special? Alice decides to investigate. By comparing the data to predictions from the full range of plant and soil science models ever published, Gaia quickly hypothesizes that this maize has a rare trait only previously observed in one other variety: it has aerial roots covered in a strange type of mucus, containing bacteria that pull nitrogen directly from the air. It’s a holy grail, a self-fertilizing cereal. Gaia guides Alice to collect more data: she samples the farm’s soil, runs a genomic analysis of the maize and the mucus, and adds all the data to the farm’s node. This confirms the hypothesis.

Gaia guides Alice to quickly find the patches where soil conditions are best suited for the heritage maize. With other constraints in mind, she and Fatima quickly come up with a plan to expand that crop to several other patches, replacing seeds and crops that require more fertilizer or are less profitable. As the next season progresses, they will be able to track whether their experiment works in real time, just by following Gaia’s updated inferences.

Fatima’s node automatically shares all these updated parameters with other nodes in the network. The detailed data like the maize genomics are configured to be private, but Fatima opts to make the provenance public. In a few weeks, other farmers reach out: Gaia has alerted them that there is a new maize variety that is perfect for their low-nitrogen soils, and they want to buy seeds from Fatima. Now the community has an additional income stream from selling heirloom seeds.

Meanwhile, Paula, a plant scientist, has been keeping close watch on the network’s automated diagnostics and seeing confidence in this technique increase. She believes she has a technique for transplanting the bacteria-carrying mucus to other cereal varieties. She uses Gaia to reach out to Fatima’s community and negotiates permission to test her technique in the farm, giving them co-ownership in the rights to any benefits generated, generating even more income for the community. Meanwhile, Gaia’s built-in traceability mechanisms help ensure that all of this is done transparently and with the right risk mitigation safeguards, without slowing down the work. As a result, Paula is able to develop her new mucus transplant solution to commercial scale and eventually able to deploy it on a significant proportion of the world’s maize fields, yielding a 50% reduction in global fertilizer use from maize.

This story is inspired by a real one: Sierra Mixe’s self-fertilizing corn. The original story not only shows how Gaia is adept at evolving solutions for hard problems of life, but also how our existing institutions and processes make for an arduous journey from discovering these solutions to deploying them at a scale that achieves real impact — while, arguably, failing to deliver on their reason for existence in terms of risk reduction. The Gaia network can rewire these institutions (or around them), creating radically faster yet safer mechanisms to develop solutions and make local decisions.

Although we picked a narrative from our own familiar domain of regenerative agriculture, a similar story could be written for a medical practitioner, urban planner, educator, epidemiologist, conservationist, economic policy-maker, nutritionist… All these domains struggle with analogous challenges and could benefit from the advantages of such a network. Indeed, we eventually expect there to be a single network supporting all domains. Through our applied R&D at Digital Gaia, we have already found several opportunities for such cross-domain integrations, between agroecology and fields as diverse as climate, development economics, urban planning, food science, and microbiology. Senge’s “inescapable network of mutuality” is becoming visible right before our eyes. It’s an immense effort, for sure, but one that’s made far more feasible thanks to the power of composability, decentralization, and collective intelligence itself.

IV. A sketch of the Gaia Network architecture

As mentioned above, our team at Digital Gaia has been designing and building an initial implementation of key components of the Gaia Network. Below we outline the architecture underlying that implementation. This shouldn’t be read as its ultimate design, but rather as a reference implementation for an open protocol that can connects and aligns many heterogeneous domain-specific systems. The design follows from a handful of simple and timeless principles, such as:

  • Good regulators: Agents must contain (or “be”) good models of their environment in order to be successful.
  • Bayesian causal modeling: The key prerequisites for such models to be useful is that they are semantic maps of symbolic variables and their cause-effect relationship networks. These are always and necessarily approximate, but can be iteratively refined using the methods of probability theory.
  • Science: Models can be alternately generalized into “science” and specialized into “applications”.
  • Active inference: Perception and action are unified as a cycle of alternately changing one’s model to fit the world (perception) and changing the world to fit one’s model (action).
  • Economics: Agents act to fulfill models of the world in which their environments survive and thrive — that is to say, they are players of economic games “on behalf of” those environments.
  • Multiscale dynamics: All of the above applies across all scales, up to and including the planetary-scale dynamics of Gaia as a single system.
  • Composability: Different models/agents can interoperate and combine their behavior through access to shared resources facilitated by a common language or protocol. Quoting Packy McCormick, “…composability is the reason that human progress accelerates: That’s what powers the increasing pace of innovation. Discoveries become inventions become building blocks become inventions become building blocks, ad infinitum.”
  • Adaptiveness: Behavior and protocols can evolve in response to changes in requirements.
  • Decentralization: Control over the system behavior and even over its shared language are distributed across the network and exist only through consensus across the nodes.
  • Whole-system (sociotechnical) design: The design should simultaneously take into account technological and human factors and their interplay, including the designer’s own role.

These principles have led us to design an architecture consisting of a decentralized network of place-based AI nodes that act as the “brains” for specific systems (acting as agentic “digital twins” for those systems), coupled to GaiaHub, an open community of human contributors of domain knowledge — whether scientific/technical, indigenous, tacit or integrative.

A high-level sketch of the Gaia architecture. Source: Digital Gaia team.

Each Gaia node is an intelligent system that continually updates its understanding of the world using active inference. It connects to various sensors and data sources, including other nodes, and uses “skills” (model components from GaiaHub) to interpret the information. The active inference process involves simultaneously updating the Gaia node’s beliefs about the world and selecting the relevant skills to explain observations and guide actions on behalf of the node’s environment. This inference loop is powered by a probabilistic programming engine, which performs the tasks of inference, prediction, and policy selection based on an expected free energy functional over policies, on arbitrary model code (assemblies of skills).

After each inference loop, the engine stores a representation of the computation in a database with cryptographic signatures, enabling verification of the node’s predictions and decisions (or recommendations). This verifiable system allows nodes to securely and privately share their beliefs with other agents in a federated inference scheme, where the entire network jointly estimates a hierarchical model. This forms the basis for decentralized scientific analysis and bottom-up modeling of higher-scale phenomena.

Nodes can also evaluate the accuracy of their data and model weights using contribution scores. This creates a positive-sum knowledge value chain, where nodes act as “claim factories” and GaiaHub’s model contributors and data providers serve as material suppliers. This value chain has built-in incentives for quality control, ensuring reliable and accurate information is used throughout the network.

Ultimately, we expect that this knowledge infrastructure will form the basis for an incentive infrastructure. As mentioned, a node’s decisions and recommendations are driven by explicit assumptions about desired states of its environment, mechanistically encoded as minimization of expected free energy functional that is common to all nodes (although they carry different models and data points which cause their decisions to differ). This allows us to define a common unit of account for behavior within the protocol — whether that behavior is knowledge-driven (epistemic) or outcome-driven (instrumental or pragmatic). As there is a direct interpretation of this common unit of account as the value added by a given behavior, we dub it the Free Energy Reduction Numeraire (FERN).

As the network converge towards a consensus on the shared value of certain actions and discoveries, and of the resources or assets that enable such behaviors, it effectively conducts price discovery of FERN-denominated “fair values”. Nodes can use such fair values to index payouts for their GaiaHub contributors, but more generally, we expect that markets will design FERN-indexed assets and currencies that rely on justified node preferences, rather than humans’ unaided preferences, as the key market driver, with internalization of externalities as a built-in feature.

Finally, the “outer loop” that evolves models through individual and collective learning also evolves nodes’ goal state assumptions, allowing the network to actively navigate the value landscape.

V. A sketch of the Gaia Network strategy

A full discussion of the strategy to deploy, integrate, and continuously evolve the network is beyond the scope of this document. We highlight some key factors identified through our work so far.

  • Leverage: To ensure the successful deployment and integration of the Gaia Network, it is crucial to identify high-leverage interventions where actors have private incentives to prefer Gaia-aligned solutions. By pinpointing areas where Gaia-positive solutions offer tangible benefits to stakeholders, the network can gain traction and support, promoting widespread adoption and maximizing its impact on global coordination and decision-making. A clear example we have identified is in the food industry, where the entire value chain is at risk from complete disruption due to compounding risk to the agricultural sector, and leading global manufacturers and investors have started to deploy significant capital towards financing the regenerative transition for their suppliers as a risk mitigation and cost control measure. A global decision support network can meaningfully improve the holistic performance of these investment portfolios, ensuring that the funders’ strategic goals of cheaper, more reliable commodity sourcing are met, while achieving ecosystem-level regeneration benefits.
  • Deliberate staging: The Gaia Network will be deployed in stages, with individual nodes being established and connected to the wider network over time. This gradual process will enable continuous validation and improvement of the system, ensuring that it remains resilient and effective in addressing evolving challenges and risks.
  • Integration: To accelerate the shift away from Molochian dynamics, the Gaia Network needs to be integrated with existing coordination efforts, such as international treaties, policy frameworks, and technological initiatives, as well as with incentive schemes at the individual, community, and organization level, like energy and agricultural transition funds. Ultimately, we aim to see the network’s native unit of account, FERN, deployed as an index of value accrual to Gaia in all of these contexts.
  • Exponential learning: As different domains in the Gaia Network are deployed, connected, validated, and evolved, this drives exponentially faster learning across the network. The continuous feedback and learning process will enable the network to adapt and optimize its strategies, facilitating more effective interventions.
  • Lock-in: It’s even possible to use the Gaia Network to draw strategies for locking in Gaia-aligned solutions, drawing on the concept of zero-determinant (ZD) strategies in game theory, discovered by Freeman Dyson and William Press in 2012 and extensively studied since. ZD strategies are the ultimate hack for non-zero-sum games. In a setting where all players understand these strategies, they have the ability to make enforceable treaties, making it possible to achieve stable positive-sum outcomes. Developing the Gaia Network’s models to support AI-augmented strategizing can then support the creation of such treaties.

VI. A counter-move to AI risk

We argue that building and deploying a large-scale supermind like the Gaia Network is the only presently available option to mitigate systemic risk from the Metacrisis. First we focus on the class of AI risk that we, along with most analysts, consider most urgent to address: the high likelihood of catastrophic, cascading failure generated by reckless, incompetent or malicious deployment of unreliable and unexplainable AI, hampering the functioning of our world’s critical infrastructure — physical, digital, and social. (To be sure, the reliability of software driving highly consequential decision-making has always been an issue, but the scope and urgency of the problem have been drastically potentialized by contemporary AI.)

In our opinion, these risks can be directly addressed by rewiring our critical global systems to use the Gaia Network as a common decision management infrastructure. The network architecture we described above, can provide intrinsic guarantees that today’s popular large deep learning AI systems fail to achieve, such as reliability, safety, alignment, explainability/ interpretability, robustness/ antifragility, privacy, and context-awareness. These follow from our focus on the “outer loops” of system design and deployment, not just from a technical point of view but also from an economic one, and from our intentional intertwining of the knowledge economy and real economy via explicitly encoded and evolving goal states.

Indeed, the Gaia architecture not only provides guarantees, but also provides a natural target architecture for surrounding protocols, tools and workflows for AI verification, functional testing, monitoring, etc. Deploying and managing this infrastructure is of course challenging, but our experience in large-scale software reliability engineering organizations indicates that it’s possible given the right incentives. Furthermore, this is an area where even incremental progress goes a long way at the margins — especially given the amount of low-hanging fruit.

However, there is an another class of risk, still not immediate but potentially far with greater consequences: the risk of runaway, misaligned Artificial Superintelligence (ASI) as described by Eliezer Yudkowsky. In a nutshell, Yudkowsky’s thesis is that:

  • A sufficiently advanced AI system will be able to improve its own capabilities towards levels far beyond human intelligence (for example, by improving its own software and hardware, and by acquiring material resources);
  • The black-box utility-maximization training approaches implicit to today’s AI imply that such an ASI is likely to have very different goals than humans, i.e., will be “misaligned”;
  • An ASI’s goals and behaviors will be effectively beyond control of any individual or organization;
  • If the above premises are realized, the most likely outcome is for such an ASI system to literally kill the entire human species.

We will not argue around the minutiae of this reasoning; for the purposes of this paper, suffice to say that even a small likelihood of such a catastrophic risk means it must be addressed. We do claim that adoption of the Gaia Network and its associated protocols as a standard global context for AI development and deployment would reduce risk of runaway ASI. We give four arguments for this.

  1. Good old reliability engineering: The first argument is the same as for the immediate AI risks of today: Through widespread deployment of intrinsically safe and explainable AI, with surrounding verification and monitoring infrastructure, we reduce the incentives and opportunities for black-box AI to be developed and acquire ASI capabilities or access to resources.
  2. Decentralization: The Gaia Network is designed to be decentralized and modular, which reduces the likelihood of any single subsystem gaining too much power or influence. By distributing decision-making and intelligence across independent systems, with adequate “outer loop” incentives and structures to keep them independent, the Gaia Network can help prevent the concentration of power that could lead to a misaligned ASI causing catastrophic harm.
  3. Model-aligned AI: Roman Leventov convincingly argues that, in human-AI just as in human-to-human relationships, model alignment is a far more important concept than goal alignment. Simply put, by building agreement around objective facts and cause-effect relationships, we stand a better chance of understanding where our interests truly converge and diverge, opening the door for more effectual, less affective negotiation where necessary and even to discover new sets of goals that are better for everyone. Following this argument, the Gaia Network provides an intrinsically aligned framework for AI development: by being built up from components that are explicitly model-aligned (having been intentionally designed by specific human collectives for that purpose), the network as a whole is overwhelmingly likely to be model-aligned. (This contrasts with current approaches of attempting to retrofit alignment properties directly into deep-learning black boxes.)
  • Enforceable treaties: The final argument is more speculative, and has to do with the concept of ZD strategies discussed above. Press and Dyson’s findings show that the ability to achieve enforceable treaties doesn’t depend on parity of cognitive capabilities, or even parity in power, only on both players having alterity and the ability to devise and play such strategies. We thus argue that, even in a setting where some form of ASI does emerge, the Gaia Network will empower human collectives with the capabilities to “reason with” and enter into enforceable, positive-sum treaties with such superintelligences.

Certainly none of the above arguments is presented as a watertight proof, and all have room for the “devil in the details” for being realized. However, they are sufficiently orthogonal to each other, and depend only on general, well-established design principles and theories, rather than on specific “knife-edge” conditions of either the problem framing or the solution design. This makes the Gaia Network a compelling counter-move against AI risk, and, as far as we know, the only concrete proposal for such a counter-move that is socially, politically and economically plausible — i.e., the only one that doesn’t involve global draconian control of all computing hardware and software, nor a Butlerian Jihad.

Conclusion: The age of moral technology

My belief that some workable infrastructure for concord will very likely emerge does nothing to drain the drama from the present, for one plausible route to long-run success is near-term catastrophe. However close to inevitable stable world governance may be in the long run, here and now we are playing for the highest stakes that have ever been played for, and winning will depend in no small part on continued moral growth. Which is to say: winning will depend on not wanting other peoples to lose. — Robert Wright

We are as gods, and we have to get good at it. — Stewart Brand

The Gaia attractor, then, is not a future where human nature is changed through mass epiphany, nor one where human nature is controlled by supposedly benevolent dictators. But it’s also not a future defined by maximizers of utility or short-term profit, whether they are corporations, nation-states, “the elites”, or AI overlords. Rather, it’s a future of AI-augmented responsibility, diplomacy and concord: one where we use technologies and incentives — with pragmatism and humility — to (re)discover our moral sense, our ability to hold ourselves and each other accountable, and our shared purpose to deliver win-win outcomes to each other, and to the planet, at global scale. It is humanity being reborn as a single, hyper-inteligent, hyper-intentional hyper-agent.

The source of this attractor’s gravity, what pulls us towards it and holds it together, is the undeniable positive-sumness of our existence, the simple fact that in our interconnected world, “moral and practical rules turn out to be the same rules”, as Donella Meadows writes. However, whether we gravitate towards Gaia or towards the Molochian attractors of chaos and authoritarianism hinges on an infinity of individual choices — primarily choices about what systems we put in place: whether they are systems that encourage our worst excesses, or systems that help us keep ourselves and each other accountable.

Such moral technologies, the building blocks for the Gaia Network, have been in development for a long time. As the Metacrisis has accelerated, so has our recognition that such systems are necessary — that survival of humanity and life on Earth is not a fait accompli, but a destiny that must be designed for. However, the current rate of progress cannot be enough, when facing a Metacrisis that accelerates cascading failures and compound risks, and now enabled by increasingly risk-prone AI. We must refocus our energy, harnessing the power of composability and collective intelligence — the Human-Computer Colossus — to accelerate the development of Gaia-positive technologies and their integration into a global network of stabilizing nodes.

We invite you, dear reader, to contribute to a Gaia-positive future, in whatever way compels you. If you’re intrigued about the architecture described above or the domains we’re working on, reach out to help us build. If you have different ideas, we’d love to hear them. If you’d rather do things independently, that’s no problem at all — we need a diversity of approaches. It will take a global village to raise this child.

As the people of the world come to constitute a single invisible brain, they can purposefully guide their course, consciously seeking the worthy goals they were once blindly, often painfully, driven toward. — Robert Wright

The future appears alien to us. It differs from the past most notably in that the earth itself is the relevant unit with which to frame and measure that future. Discriminating issues that shape the future are all fundamentally global. We belong to one inescapable network of mutuality: mutuality of ecosystems; mutuality of freer movement of information, ideas, people, capital, goods and services; and mutuality of peace and security. We are tied, indeed, in a single fabric of destiny on Planet Earth. — Peter Senge


  • Digital Gaia team and community
  • Past and present advisors, collaborators, friends — too many to list, notably: Steve Coy, Ben Boor, Thomas Meehan, David Thomson, J. Doyne Farmer, Ramsay Brown, Casper Hesp, Martin Wainstein, John Clippinger, Stuart Cowan, Gianni Giacomelli, Jacob Taylor, Pranav Gupta, Jan Bunge, Omino Gardezi, and Max Kelly.

About the author

Rafael Kaufmann is the CTO of Digital Gaia and member of the Board of Directors at the Active Inference Institute. He is the author of Gaianomics, Natural Intelligence, and dTwins.

Permalink to this piece