“We shape our tools and then our tools shape us. The medium is the message. This is merely to say that the personal and social consequences of any medium - that is, of any extension of ourselves - result from the new scale that is introduced into our affairs”  

- Marshall McLuhan

McLuhan would be disappointed by how we are thinking about AI-enabled media. Our focus is on forging the tools and not wielding them. When we have the tools, we are engaged in acts of mimicry. Chatbot, make me a photoreal sunset shot on iPhone15. Once we master that we turn to remixing. Chatbot, fit Harry Potter in Balenciaga gear. 

We are not appropriately thinking about the scale this new technology allows for and the second order effects of that. Our brain is trapped in objects when it should be thinking in systems, worlds and simulations. 

The reason we are not thinking this way is because the answers we need to organize this newly emerging scale lives outside technology, where most AI thought happens today. We need to be drawing from biological and ecosystems thinking. We need to think economically about AI beyond productivity gains and job cuts, towards how we can form new models of value creation and distribution. In the realm of philosophy, we need to venture past AI ethics and think ontologically about networked relationships between things. We need to get out of a mindset focused on being and instead turn our thoughts to becoming. 

As a part of that process, we need to rethink how we tell and interact with stories. Stories alone are not enough to relate to the human experience anymore. Simple parables or heroic journeys fall short when relating to a world as complex as ours, this place that does not resolve neatly on the last page because there is no last page. The world and by extension, humanity, just goes on and on, always evolving, always changing and that’s the beauty of being alive. 

Now that we have the ability to do so, our media should more accurately reflect the evolving nature of the world. It should live and breathe as we do. This is how internet culture works today. Memes are alive. Long form storytelling should be too. We need to think in worlds that have arcs and scaffolds, but not fixed scripts.  We need to move away from passive consumption on our couch and into a shared public life of communal mythmaking.

When you think of virtual worlds today, you think of gaming. Zelda is a virtual world. It is open and exploratory, but it is also scripted and closed. There is a final destination, a last hill to climb, a final boss to beat. There’s also only one way to enter Zelda which is through your console and one way to interact with it which is through a joystick staring at a screen. 

All of our media is actually limited to the delivery interface it is designed for, but that’s not how we interact with our world. Our base reality is multimodal. We watch on TVs, we type on keyboards, we listen while walking with airpods in our ears. We move through space with our feet, but also using our eyes and ears, our sense of smell and our touch. Immersion is just that. It’s interacting with worlds through many modalities and touchpoints that each offer unique perspectives and orientations. 

Before, media was primarily limited to a specific medium because of technology. That changes with AI. The scale, the speed, and productivity gains give us an ability to transform a digital world into a multiplicity of experiences that is now open to us. 

What if you wanted to talk to a character in Hyrule while you walked through a park in your town? Or hear new music from that world after listening to the official soundtrack on Spotify for the millionth time? What if you wanted to create a new quest that puts yourself in the Hyrule and do it in a way that doesn’t require staring at the Switch, but rather in any number of fashions as you go about your day? Solve a puzzle on your phone at lunch. Co-create lore with an LLM while on a pointless Zoom meeting.

What if the events from that, the places you discover or objects you create get added to the world and the world adapts to what you do? What if your actions matter in creating the world, not just defeating it? This is what CAMS proposes. To free virtual worlds from the confines of gaming. To free long form narrative from the structure of books and movies. Open worlds that evolve and can be rendered up in many forms of media are possible now thanks to AI. We just need a way to organize and interact with them.

It is time to think like a fully networked species who now have the power to re-enchant our world by creating new digital worlds. Complex Adaptive Media Systems (CAMS) is a new design pattern that allows us to do just that.

With scale comes the requirement for new ways of thinking. The government of a village requires one way of modeling itself, the government of a nation another, and nations working together yet a third. One big difference as we scale up is that the relationships between people change. A village may simply gather everyone together to collectively decide on things. A national government may introduce a hierarchy of elected leaders who have to advocate for local interests while also considering national needs. At an international level, countries need to weigh protecting their own interests against the benefits of working together. At each level of scale additional complexity emerges that requires new models of organization.

The same is true of the internet, but how it governs and organizes itself more closely resembles our natural world, which is one giant biological system made up of collections of subsystems, all of which nest inside each other. There is one Earth, but within it many different biomes, inside those biomes are forests, inside those forests tiny ecosystems exist on each tree. There is no governance in the forest. Leaf cutter ants don’t elect an official to speak up for their share of a tree’s yield. That tree doesn’t need to clamor for water rights amongst all the other trees in the forest. It just works as a decentralized process better known as a complex adaptive system.

Complex adaptive systems are one model we are going to apply to CAMS (which is why the names are so similar). Why, though? The answer is the scale that AI enables. We are no longer thinking in terms of a website or a collection of NFTs. We are thinking in terms of constellations of objects and entities interacting with each other over a network in ways that more closely resemble the natural world than the old model of a product team managing an ecommerce website via the agile development process.

Complex Adaptive Systems (CAS) are foundational constructs for understanding the operational dynamism in diverse fields, from ecology to economics, and now, significantly, in media systems. A complex adaptive system is a network of individual components or agents, each acting according to simple rules or strategies, without any centralized control. Through the interactions of these agents, the system as a whole exhibits behaviors that are not predictable from the properties of the individual parts—a phenomenon known as emergence.

At their core, CAS are characterized by several defining features:

• Diversity and Individuality of Components: Each agent within the system has unique properties and behaviors, adding to the system's overall complexity.

• Connectivity and Interactions: Agents within a CAS interact with each other and with their environment in ways that both affect and are affected by the system. The nature and intensity of these connections can vary widely.

• Adaptation and Learning: CAS have the capacity to change and learn from experience. Agents alter their behavior based on the outcomes of their interactions, leading to the evolution of the system over time.

• Non-Linearity: Inputs and interactions within a CAS can lead to disproportionate effects, making the system's behavior unpredictable and sensitive to initial conditions.

• Emergent Properties and Behaviors: The most striking feature of a CAS is emergence, where the collective behaviors of agents give rise to properties at the system level that are not present at the level of individual components.

• Self-Organization: A consequence of local interactions and feedback processes is the ability of the system to organize itself into patterns and structures without external guidance.

Complex adaptive systems provide a powerful framework for responding to change and for fostering innovation and resilience. Applied to media systems, these principles translate into a flexible, reactive, and continuously evolving landscape where narratives and user experiences are fluid and alive. As complex adaptive systems inform the structure and functioning of CAMS, they equip these media systems with the flexibility to not just weather the storms of change but to thrive within them, dynamically adjusting to the input from each user and the shifting trends of the greater media environment.

Climate change is the most pressing issue facing our continued existence. We only have one planet. We aren’t able to get to or colonize other planets yet. Therefore if we can’t fix the problem soon, we are metaphorically going to become toast. Yet, we aren’t doing a very good job of managing it. Why?

One big reason is the unapproachability of the problem. There is no one person or thing directly responsible for climate change. You can’t ask to schedule a meeting with the CEO of Global Warming. Another problem is the massiveness of it. One country can get their house in order and reduce emissions, but it doesn’t matter unless all of us do. Say everyone gets their act together tomorrow, that’s great, but guess what? We won’t see the impact of that for years. Climate change works on large timescales. The steam engine was invented in 1712, we dropped the atomic bomb in 1945. Those are two milestone events in climate change. One was three centuries ago and the other seven decades ago. The worst parts of climate change won’t be felt by us, but by our children and grandchildren. 

When do you think about climate change? It’s not on the normal days. It’s on the really hot ones when the asphalt is melting. It’s on the ones that should be cold, but aren’t. It’s when a tropical storm turns into a category 5 hurricane at a speed that’s never happened before. You couldn’t live if you spent every waking hour worrying about global warming, so instead your awareness phases in and out as its effects manifest themselves in the world around you.

What do you call something that exists and is real, but is massively distributed in space and time? Whose effects are present, but unapproachable and only experienced indirectly? It’s a weird thing to consider and no one really had until 2010 when a philosopher named Timothy Morton coined the term hyperobject.

Before we better define and describe hyperobjects, it’s important to note that they can be any number of things. To make this more digitally relevant, we can actually classify Facebook as one of the few web services in the world to be big enough to be a hyperobject. Think about the scope and scale of Facebook for a minute. It’s as small as wishing someone a happy birthday or liking a baby picture. It’s also large enough that it tilts elections and changes the course of nations. Unless you are in a small subset of people who personally know Zuck, it’s almost impossible to approach or influence Facebook, instead it acts on you directly by sending push notifications or indirectly by silently surveilling you across the entire internet.

We aren’t thinking big enough about the scale and scope of the networks we are building. With AI coming of age, we are about to have the ability to virtualize and miniaturize both hyperobjects and complex adaptive systems, and transform them from massive things in the world beyond our control to digital systems in virtual worlds that we can conjure up and have agency over.

Hyperobjects are a concept birthed from the realm of philosophy, primarily recognized through the work of Timothy Morton. They refer to entities so vastly distributed over space and time that they dwarf our usual perception of scale and duration. These entities are multidimensional in every sense, encompassing phenomena such as climate change and the internet, which are so pervasive and all-encompassing that they seem to be 'everywhere' and 'everywhen' yet nowhere specific at the same time.

The scale and complexity of hyperobjects are not constrained by traditional geographical or temporal boundaries. They operate on such grand, extensive scales that our engagement with them becomes intermittent and diffuse, witnessing only fragments of their totality at any given point. The reach of hyperobjects extends far beyond the lifespan of any individual and has effects that are only recognizable through the accumulated actions and reactions across large swathes of time.

In the landscape of contemporary digital culture and media, hyperobjects present themselves as ever more relevant. The Internet itself is a prime example, comprising a vast web of connections that extend across the globe, involving countless nodes of engagement that contribute to its massive, yet intangible, form. Hyperobjects influence how we conceptualize everything from storytelling to user experience, demanding that media systems accommodate not just the immense size and scope of the data they deal with but also the intricate and interwoven relations they foster amidst audiences and content creators alike.

Understanding hyperobjects thus offers profound insights into creating media structures capable of not merely containing information but engaging with the expansive, interconnected fabric of 21st-century life. As we traverse further into the digital age, it becomes clear that media systems must not only recognize the existence of such colossal entities but also adapt their organizational principles to effectively mirror, map, and navigate the hyperobject-infused media topography.

Hyperobjects wield distinct properties that unfold across multiple dimensions, challenging our standard conceptions of physicality and affecting profound implications for our interaction with the world. These properties—viscosity, nonlocality, temporal undulation, phasing, and interobjectivity—represent cornerstones in understanding the overarching impact of hyperobjects.

Viscosity, in the context of hyperobjects, refers to their persistent tenacious influence, which clings to all they come into contact with. This characteristic means that hyperobjects leave lasting, tangible traces on individuals, societies, and various forms of life, long after direct interactions have ceased. The effects of a hyperobject are not lightly shed, permeating through time and space to impact future states in unforeseen ways.

Nonlocality is the aspect of hyperobjects that denotes their omnipresence and the impossibility of pinpointing them to a specific location or moment in time. They are at once everywhere and elusive, challenging the concept of locality, by existing diffusely across vast expanses of space and time, often beyond the human scale of perception.

Temporal undulation characterizes the ebb and flow of hyperobjects across time, as they do not maintain a consistent pattern of influence. Their impact intensifies and diminishes in cycles that do not necessarily correspond to human-centric timelines, eluding our traditional narrative frameworks built on straightforward temporal progressions.

Phasing expresses the intermittent perceptibility of hyperobjects, as they oscillate in and out of our field of awareness. Depending on various factors—such as our technological capacity to detect them, their interactions with other objects, or natural cycles—hyperobjects can appear transient, despite their enduring existence.

The concept of interobjectivity is intrinsic to the nature of hyperobjects. It reveals that hyperobjects arise from the symbiosis and interactions between multiple entities, manifesting as a cumulation of influences rather than as direct, observable phenomena. For instance, global warming, a quintessential hyperobject, emerges through the interplay of solar radiation, fossil fuel combustion, and carbon emissions. While we can observe its manifestations via indicators such as greenhouse gas concentrations, temperature fluctuations, and rising sea levels, these are merely the 'footprints' left by the hyperobject, not the hyperobject itself. The interobjective quality of hyperobjects teaches us that they are not merely results of scientific measurement or detection but are complex constellations of interaction that predate and exceed their quantification.

Together, these multifaceted properties prompt a significant shift in our narrative approach. Hyperobjects require storytelling that embraces complex, intertwined presences—narratives that account for traces rather than objects, impressions rather than clear-cut entities, systems rather than isolated elements. In the framework of Complex Adaptive Media Systems (CAMS), appreciating and incorporating the properties of hyperobjects is essential to evolving and enriching our media landscapes, as we aim to craft narratives as boundless as the hyperobjects themselves.

The final organizing model we want to introduce is the concept of digital physics. This term has several meanings, so we want to be clear from the outset that we are not talking about Wolfram’s theory that the universe is mathematically designed and knowable in computational terms. Instead, we are talking about the digital physics that govern how objects can behave and interact with each other in the virtual environments they live in. 

Let’s start with some easy examples of digital physics. Social media algorithms decide who sees what content and when. I’m not guaranteed to have all my followers see a tweet I made. If a tweet I make gets traction, an algorithm decides if more people should see it or not. That’s an example of digital physics. So is how a character moves through a video game. How high can it jump? How fast can it move? These are mathematical functions in code that dictate and mimic the physics of our physical world. 

Digital physics are the code, processes and systems which govern the movement of goods, content, interactions and economies through networks and virtual worlds. They are virtual representations of things that exist within our societies and IRL environments, that allow us to order and bring meaning to virtual spaces which are conceivably boundless and infinite. While many of them are designed to mimic patterns we live with offline, they are also not constrained by the real world. An easy example of that is messaging a friend. That happens near instantly. There’s no logical reason to delay it just because the postal system takes a day or more to deliver mail.

When we design networked software and deploy code on the internet, we have the arbitrary power to define the physics which govern everything that happens within the system. We could make a game where walking takes just as long as in real life, and then ask the characters to traverse one thousand miles on foot. It’s stupid and no one can play it, but it’s possible. The point is that digital physics are arbitrary at the time of design, but bind virtual entities who come to inhabit these spaces.

All sorts of things are possible with digital physics because the laws that constrain us in the physical world do not apply to networked space and computer systems. Time in virtual space can be governed by computation power and the speed at which you can simulate events. Time can be fabricated. We can move through time if we want to simply by deciding what objects to return and present to a user. It’s all virtual, so it can all be manipulated. Space is the same. We represent space virtually simply as points that sit in relationship to each with a scale then applied to them. The cost to store data representing one square inch of virtual land is the same as the cost to store 1 million square miles. 

Understanding digital physics allows us to think big enough to understand the capabilities of AI-enabled media and entertainment. We can teleport and time travel thanks to digital physics. The Inuit people recognized this ability, and when updating their language named the internet “ikiaqqijjut” to reflect these properties:

Liminality and liminal spaces take on unique abilities when transferred into the realm of digital physics. It’s here that the Intuit term ikiaqqiijjut (the tool to travel through layers) starts to feel like a much better description than our English word, internet.

A liminal space refers to a physical or metaphorical threshold, situated between two different existential planes, phases of existence, or social statuses. Originating from the Latin word "limen," meaning "threshold," liminal spaces are characterized by their transitional nature. They are places or periods of ambiguity, disorientation, and potential transformation, where the usual constraints of time and social structure are minimized or absent.

AI is the first tool we’ve had with the scale, intelligence, and speed necessary to transform large amounts of information in a way that is reactive, fast and fluid. It gives us the ability to turn liminality into a presentation layer or form of user interface.

Imagine cutting across and then refashioning huge swaths of information into representations that unmoor us from place and time, and places us into worlds between worlds. It's the ability to travel between layers, but unlike traditional shamanic practices (i.e. psychedelics) we can now do this in a controlled and directed way. Now extrapolate that outwards so that you have several liminal experiences working in concert with each other, and a user moving between them. Phasing in and out using the language of hyperobjects. That's a net new media type native to AI and will be essential in the CAMS system for allowing people to interact with virtual worlds in a multitude of modalities and environments. Hyrule no longer has to live in a Switch cartridge.

Let’s show you an early example of liminality achieved in generative AI, a  post-photography NFT collection called Life in West America by Roope Rainisto:

LIWA feels deeply familiar due to the setting, era and symbolism that it draws upon. It evokes the spirit of the 1950s American West, that post-WWII period of optimism and expansion. It does this with a scale and depth of 500 images that create a suspension of disbelief which challenges and reorients the viewer. How is this possible? Because the AI-rendered photographs are a very close approximation of real photographs in terms of their quality, yet the content of them are out of line with our expectations. Body parts are rendered strangely, houses are collapsed in ways that are not possible, the cars are strange, the angles of everything are all wrong. This is a world that obeys digital physics, not our own. This disconnect between media type and content unmoors our brains which cannot pattern match it to our own memories. It is half real, half a dream and that combination of liminality creates a powerful effect on the viewer.

Planet Hollywood - Life In West America by Roope Rainisto (2023)

Digital physics are not bound by the same rules as in real life, however we closely model them on similar principles and properties to have familiar mental patterns with which to relate to them. Up until now, the production of digital physics has been a largely human-led endeavor limiting the scope of what has been produced by time and cost. The limited scope of digital physics has meant that we could also use analog methods of thinking and management in their governance and operations. While there are cultural differences and nuances to approach, running Google is not radically different from running JP Morgan. 

This paper posits that the recent advances in AI across the board have tipped the scales of production and management to a point where they will soon be unbounded by the time and cost factors that have traditionally limited the output of digital physics. Faster, cheaper, better will result in a scale that tips the design and operating requirements past the point where traditional management approaches are effective. Different ways of thinking are required to deal with the scope of opportunity in front of us. Those ways of thinking exist already, they just have never been widely applied to this field before. 

While managing Google is not radically different from managing JP Morgan, managing the global economy (if it were actually possible) is. That’s because the global economy is a complex adaptive system, and it is impossible for us to systematically get our hands around the entire thing. Instead, we have to settle for influencing it by acting on the defining features of it as best we can. Interest rates and money supply are nonlinear inputs which we can adjust. Businesses self-organize and adapt their practices depending on conditions (layoffs in bad times, hiring in good times). We see unexpected emergent behavior (the rise of influencers) change consumer marketing, forcing a rethinking of how communications happens across subsystems within the global economy. We’ve never thought about intentionally creating complex adaptive systems before because we’ve never had the ability. Now that the ability is here, it’s worth contemplating how.

CAMS are envisioned as a new form of media which means their primary purposes are entertaining and educating human beings. So we now need to introduce a way of thinking about how humans interact with and relate to massive complex adaptive systems which, thanks to digital physics, have the ability to transcend virtual space and time. This is where hyperobjects come into play. It gives us a framework and methodology for how massive distributed systems appear to and act on humans.

An interesting development is that scripted dramas across novels (World War Z), TV (Lost) and movies (Inception) have started taking on more sophisticated forms of storytelling which resemble the properties of hyperobjects. World War Z tells the story of the zombie apocalypse as a series of interviews with survivors, never directly recounting the event in a linear fashion, but rather providing a context into the event itself through the impact it had on people experiencing it. Lost disrupted time, using not just flashbacks but flashforwards to dislocate the viewer and hint at things to come while also providing present motivations by backfilling our knowledge of the characters’ history. Inception with its conceit of controlled shared dreaming opened up space and time in a way that markedly resembles what is possible with digital physics in networked environments. Each of these were heralded for their innovations, hinting at the effect the properties of hyperobject can have on pulling audiences in through more expansive and immersive patterns of storytelling.

In fact many of the properties of hyperobjects adapt themselves well to narrative form on a grand scale. Phasing, non-locality, and temporal undulation are at work with fantasy epics such as Lord of the Rings and Game of Thrones where we jump around between different people and places, where one group of characters are central to driving the action at a key junction only to recede into the background as the storyline travels beyond them. Interobjectivity or the relationship between people and objects is there as well, the ring of power is ever present in LOTR and forms a nexus around which all plot action orbits. Another great plot device is to create interlocking narratives which pull characters closer into conflict with each other despite shared personal history and affinity for each other. The recent Netflix adaptation of the Last Kingdom achieved this to great effect. Finally viscosity, let’s just say that Trekkies and other forms of rabid fandom such as cosplay conventions are an enduring example of media taking a tenacious hold over people who interact with it, impacting their futures in ways that could not be anticipated. 

One could say that the largest media franchises are accidental hyperobjects already, the only difference is that each installment is scripted and the unscripted action plays out both in the cultural sphere and in the creator economies which drive the continuance of the story world. There’s untapped potential for even greater engagement with media if it had the production pacing necessary to break free of scripted endings and let these worlds play out as stories lived in real time by the audience who also serve as the actors. This is the potential promise of AI. Thinking in digital physics, complex adaptive systems and hyperobjects is the way to model this approach on the backend, while presenting it through layered liminality and multimodal user interfaces is how the audience can immerse themselves in these worlds without being overwhelmed by their scope and depth.

Now is the time to start thinking about what comes next. Networked media is about to scale exponentially, and that scaling is going to coincide with the ability to access and transform media at a scope and speed we’ve never experienced before. We are not going to be able to read faster or watch longer, we are still gated in the most human of ways, by our time and attention. The change in behavior then needs to come not by asking more of us, but by creating exploratory environments in which we can exert directed control over what we interact with and by doing so shape the media world we are immersed in. 

We are changing from a passive broadcast media which needs to resolve neatly at a creator’s determined cut off time, to a participatory, ongoing model of networked media in which the ending is not known ahead of time, but rather found collectively through communal action. This approach requires a new media model, a framework which we’ve outlined in this paper. Complex Adaptive Media Systems are not scripted. They are lived experiences which evolve in vector space, adding to their past, present and future as users travel through layers within them. 

We’ll close by returning to our scolding McLuhan at the start of the article. If generative AI and the economic incentives of blockchain L2s simply result in more media without changing how we interact and experience it, then we’ll drown in a sea of meaningless spam that debases all work whether human or machine made. 

That will not happen however, because humans are intrinsically wired to seek meaning. We are on the cusp of creating virtual, networked worlds which we can all exert agency over. Worlds however are not and cannot be scripted. Their complexity operates at a scope beyond what a creator is capable of conceiving and delivering as a neatly wrapped bundle of consumption for you to enjoy on your couch. 

This complexity should not be feared, we live in and amongst it every day. We’ve just been trained since the Enlightenment to think of media as a one-way broadcast between a creator and an audience. That barrier is breaking down as we upgrade to a fully networked society. In many ways, we are returning to the oral tradition that pre-dated it, one where we lived our stories, changing them to suit our local environments and transmitting them not as IP rights held by corporations, but as collective communal myths.

How this plays out and is implemented remains to be seen. The tools to worldbuild, simulate and travel through networked layers are in their infancy, but they are coming faster than our thinking about how to use them to their fullest potential. Complex Adaptive Media Systems introduces a framework for how to think about this transition by taking organizing models that exist in our natural (complex adaptive systems) and philosophical (hyperobjects) worlds and then applying that to digital physics where scale is no longer bounded by production time or cost. 

A new way of interacting with media awaits. It will be immersive, provide agency and ownership, and exist on orders of magnitude larger than we’ve ever considered before. In many ways, it will resemble life just rendered as liminal digital experiences. The boundaries between lived experiences and stories will blur as we move into a hyperreal future. 

In our past, we’ve viewed this convergence with dystopian fear (see The Matrix), but that’s because we’ve also been in a mindset of scripted entertainment produced from singular viewpoints. CAMS provides an alternate viewpoint that more closely models how we actually think and interact when confronted with enormous scale. It’s both the way the natural world organizes and how we as humans seek meaning from the massive. Now that the tools are here, it’s time for our media to upgrade and adapt.