New+Draft


 * Self-Organizing Virtuality: a New Paradigm for Learning Ecosystems**

Self-organization is a foundational characteristic of nearly everything within our purview. The cosmos is self-organizing: it grew from a singular point to an estimated 150 billion light-years in diameter in the span of approximately 13.7 billion years. Life is self-organizing: in the presumed absence of divine intervention, the first living cells emerged 700 million years after the Earth formed. Life and the planet have co-evolved in the ensuing 3.8 billion years, giving rise to a third major self-organizing entity: mind. Mind has spawned a fourth major domain that is on the threshold of becoming self-organizing: technology.

"Whether we are talking about molecules cooperating to form cells or organisms cooperating to form ecosystems or buyers and sellers cooperating to form markets and economies, we will find grounds to believe that Darwinism is not enough, that natural selection cannot be the sole source of the order we see in the world. In crafting the living world, selection has always acted on systems that exhibit spontaneous order."

- Stuart Kauffman At Home in the Universe- The Search for the Laws of Self Organization and Complexity

Human technology cannot yet be characterized as self-organizing, although in many disciplines it is becoming so. Advances in software design and robotics have pointed toward a wisdom that the universe has always possessed: optimal existence results from evolution and self-organization. Evolutionary computation, machine learning, artificial life, complexity theory and swarm computing all borrow heavily from the foundational characteristics of the universe, life, and mind.

One of the more compelling paradigms to arise from the computing revolution is virtuality. This is not a new concept, as the arts and literature have a legacy of created worlds within the pages of a book, the proscenium arch and the frame of a canvas. Computational virtuality, however, takes the paradigm much further. Immersion is the hallmark of this arena. It can be all encompassing, as with the head-mounted displays of virtual reality, or more modest (yet, arguably, of near-equal effect) as with the widespread adaptation of MMORPGs and virtual worlds for common personal and mobile computing platforms. Computational virtuality is also open-ended: games can be scripted (to a degree) but virtual worlds are largely stochastic, with elements of free will and non-determinism that roughly parallel real life.

Virtuality can be thought of as a mode of being. We speak of being ‘in world’ when we are immersed. Actions, such as unassisted flying and teleporting, are ‘not possible in real life’ (NPIRL). An uncanny fusing of one’s identity with one’s avatar occurs – witness when our avatar suddenly loses its clothes in the company of others – it is almost impossible not to feel embarassment and turn away. We project our personalities into virtuality but we also undergo a degree of transformation. Social reticence may ease, allowing more free-flowing interaction and forthrightness. Phobias may lessen as much more is possible without fear of injury. Real life disabilities can be overcome in-world in ways not possible before the onset of virtuality. Some of us who have spent significant amounts of time in-world may have found ourselves invested in virtuality in ways that surprise us, and which have led to rich and meaningful experiences, albeit of an other-worldly nature.

Yet for many, virtuality falls short of offering a full blown mode of being, and comes to resemble a game without goals, or a chat room with a 3D engine bolted on. It may be the specific designs of virtual worlds that are responsible for this dystopic reaction, or then again it may run deeper. Perhaps it is not the specific designs that are the problem, but the very fact that these worlds are designed at all. A designer or a team of designers could never think of everything that would keep a vast diversity of minds engaged enough to keep returning, month after month, year after year. And so we see the exodus from virtuality that has occurred, and now begin to look for root causes.

As our knowledge-modeling tools evolved from myth, scripture, and philosophy to the scientific method, our models of the universe also evolved. Leading-edge theoretical physics takes us beyond the twin 20th century revolutions of Quantum Mechanics and Relativity to String Theory, M-Theory and other models that will inevitably arise in the search for a unified theory of all of creation. The trajectory of this evolution has brought the Newtonian 'background' universe into the foreground, relativizing and quantizing its myriad components into a seamless fabric of reality. Our bodies are made of the same quantum stuff as stars and planets; our minds and consciousness emerge from that self-same substrate.

Mapping the progress of virtual worlds to this trajectory, virtuality remains in a Newtonian universe. The virtual world is a passive background to we the 'players'. Worse, it is more stage set than city street, more painted backdrop than landscape, more prop than tool. Granted it is a social environment that dissolves geographic distance and has tremendous potential for teaching and learning - but the environment itself does not evolve; it does not surprise us with emergent wonders; it does not //live//.

What if the 'world' in virtual worlds could borrow from the playbook of our world? What if it could evolve, self-organize, and spawn life and mind? Algorithms that facilitate these principles (albeit imperfectly) exist and enjoy a level of relative maturity in the domain of artificial intelligence. And what are virtual worlds, if not code? Apart from the inputs of its human participants, a virtual world is code and data, and subject to the same manipulations and innovations as any other code base. If code can predict stock market trends, win at chess and jeopardy, drive a car or a Mars rover, model global weather and more, can it not be coaxed to apply existing self-organizational and evolutionary algorithms to the creation and ongoing development of a virtual world?

“The British cybernetician W. Ross Ashby proposed what he called ‘the principle of self organization’. He noted that a dynamical system, independently of its type or composition, always tends to evolve towards a state of equilibrium, or what would now be called an attractor. This reduces the uncertainty we have about the system’s state, and therefore the system’s statistical entropy. This is equivalent to self-organization. The resulting equilibrium can be interpreted as a state where the different parts of the system are mutually adapted.” -Francis Heylighen The Science of Self-Organization and Adaptivity

In a multitude of disciplines over the last hundred years, trends in leading edge thought have gravitated toward the eclipse of centralized command and control in favor of distributed models that borrow heavily from natural systems. These a priori systems are all, without exception, self organizing. They tend to inhabit the interzone between stasis and chaos, seeking stable ‘attractors’ about which to cycle. Non-dynamic ecosystems, like deserts, yield little in the way of significant biological innovation. Dynamical systems, like coral reefs, attract and multiply combinations and possibilities, creating oases of complexity and fecundity where innovation can prosper.

Language, arguably the most important component of human consciousness and culture, is largely self-organizing. Defying efforts at central planning and control, language evolves and mutates, spawning new words and phrases as it obsoletes others, forms dialects, specializations and slang.

“Wild systems are highly complex, cannot be intellectually mastered—that is to say they're too complex to master simply in intellectual or mathematical terms—and they are self-managing and self-organizing. Language is a self-organizing phenomenon. Descriptive linguistics come after the fact, an effort to describe what has already happened. So if you define the wild as self-managing, self-organizing, and self-propagating, all natural human languages are wild systems. The imagination, we can say, for similar reasons, is wild.” -Gary Snyder (American Poet, winner of Pulitzer Prize for Poetry)

The feedback mechanisms of cybernetics, the distributed low-level intelligence of modern robotics, machine learning, evolutionary computing, swarm intelligence and artificial life algorithms are all technological outgrowths of this trend toward ‘wild systems’ that have relinquished determinism in favor of self-organization and evolutionary models.

Much as models and theories of the universe, life and mind have evolved from deterministic, top-down theistic or pantheistic scenarios to non-deterministic, bottom-up, self-organizing and evolutionary scenarios, pedagogical theories have evolved from teacher-centric, one-way transmission of knowledge to learner-centric, networked exchange and creation of knowledge.

Personal learning environments and personal learning networks are concepts that have gained traction in the wake of these trends in learning theory and the proliferation of digital networks and tools. These approaches to learning place the individual in the center, connected to an ever-expanding and changing cloud of tools, knowledge repositories and peers, much of which arise through serendipitous chains of association rather than as the result of a concerted, centrally designed initiative.

Just as the internet has proliferated as a kudzu-like web of nodes and connectors, encircling the planet via high-speed telecommunication networks, submarine cables and communication satellites, so have myriad networks of teachers and learners (many fulfilling both roles simultaneously), piggybacking on the capabilities and programs of the web, mobile computing and communications, formed an intellectual superstructure, communicating, publishing and subscribing to a burgeoning, dynamic body of knowledge.

Communities of practice and communities in the service of education and knowledge creation (often one and the same) form spontaneously in this ecosystem. Attempts at wholesale design and engineering of these entities routinely fail:

"Much like a living organism, they (Communities of Practice) are self-organizing, and cannot be designed prima facie. They grow, evolve, and change dynamically, transcending any particular member and outliving any particular task." //-// Sasha A. Barab, James G. MaKinster, and Rebecca Scheckler Designing for Virtual Communities in the Service of Learning.

Virtual worlds, like Second Life, provide fertile ground for teachers and learners to reach out and connect. Their presence embodied in avatars, participants can inhabit and traverse a symbolic landscape, interacting with other intelligences, both human and machine. Non-player characters (NPCs) are avatars with artificial intelligence that can interact with participant avatars, much like the classic artificial intelligence programs ELIZA and Racter, which carry on conversations by processing key words and phrases of input and constructing semi-intelligible sentences that appear to respond and converse. So far, NPCs provide the only measure of intelligence and interactivity at the level of software in virtual worlds today. NPCs are a breed of ‘Narrow AI’ designed for a specific purpose and fairly useless outside the confines of their initial design. This is in contrast to general purpose AI, an elusive achievement whose time has not yet arrived which posits an autonomous intelligence, human-equivalent in cognitive ability, though not necessarily in design or behavior.

In AI Meets the Metaverse: Teachable AI Agents Living in Virtual Worlds, Ben Goertzel posits a general purpose AI that inhabits a virtuality such as Second Life and learns from the human participants (avatars) via linguistic and behavioral interactions. The sheer number of avatars that could potentially be employed in teaching the AI would vastly augment the cognitive development and knowledge stores of the AI, rapidly accelerating its capabilities via distributed, non-deterministic means. Goertzel’s AI follows a human developmental model augmented by the multiplying effect of a virtual community of minds. This is one vision of the power of virtuality combined with artificial intelligence to foster emergent phenomena.

We suggest a different model, based not on human developmental lines but as the next stage in the progression of the great, natural, wild systems. The universe, life, mind, and technology are the systems that have brought us to the threshold on which we now stand. Computational virtuality, coupled with self-organizing and evolutionary algorithms, holds the possibility of humanity’s offspring, technology, kick-starting the cycle (universe, life, mind, technology, virtuality) again, at an arguably higher level than the natural antecedents.

Virtuality, a mode of being, is the domain of minds freed from the material constraints of biological bodies, geographical distance, national boundaries, economics and law. Yes, it can be argued that some or all of these play a role in how we approach virtuality, and yet, once there, these constraints fall away, and we are left with minds interacting with one another (usually in simulacra of familiar bodies and environments). It is precisely because of this freedom from constraint, coupled with the exponential multiplying effect of optimally interconnected minds, that we posit self-organizing virtuality as a higher level substrate.

Billions of years of self-organization and evolution have resulted in minds that are capable of virtualizing the substrate from which they emerged. Mind now becomes the new substrate, from which emerges virtuality and the limitless growth and development possible therein – but this virtuality will not engender such myriad possibilities without itself becoming a wild system. It’s not that no other model is possible – there are plenty of deterministic pathways that have created impressive cultural artifacts – but more that no other model is as powerful, efficient and capable, or else it would have become the dominant driver for the universe, life and mind. Self-organization, constrained by evolution, by definition will try every possible pathway until it settles on the optimal. If it didn’t try it, it wasn’t possible – possibility being contingent on the materials and forces at hand in any given domain.

What will self-organizing virtuality look like? How will it enhance our experience, our culture, our lives? If it is employed first and foremost in the service of learning, it cannot help but richly augment and rapidly expand the capabilities of mind interacting with mind. No longer a painted backdrop to our simulated real-world movements, self-organizing virtuality, we predict, will exhibit the following traits:


 * It will grow from seeds.** Initial parameters, combined with self-organizing algorithms and taught by participating minds ala Goertzel, will create an indeterminate, wild system that is completely unpredictable and open-ended. Such a system cannot arrive fully designed and built, like Athena sprung from the brow of Zeus, but must grow, as all its antecedent wild systems have.


 * It will be shaped by its participants.** Along with evolutionary algorithms, self-organized virtuality will receive continuous feedback from its participants. If its development takes it down pathways that are antithetical to the goals and sensibilities of its participants, they will reject it, voting ‘with their feet’. Conversely, productive, innovative pathways will be rewarded as participation grows virally, spread by word of mouth and the Internet. It is worth noting that this is exactly what has happened to commercial virtual worlds in recent years, as participants have fled the static simulacra once so popular in favor of the arguably more connective and responsive (yet radically narrower) domain of social networks.


 * It will be stable**. Self-organizing systems evolve towards equilibrium, seeking stable attractors on the boundary between order and chaos. Lacking the enforcements of direct design, our virtuality will instead embody a living, morphing pattern system wherein the principles of the universe prevail – in this sense it will feel very familiar to us, yet it may also exhibit radically non-human versions of order.


 * It will reproduce**. Virtualities will reproduce, spawning daughter worlds that retain some traits of their ancestors, yet deviate in ways that became impossible in their ancestors. For as virtual worlds grow from seeds, shaped by the feedback of their participants, they will naturally follow certain pathways while avoiding others and become constrained in their development. Reproduction will be a natural outgrowth of this, as continued innovation and evolution will only be possible in subsequent generations that retain favorable traits (ability to attract and retain human participants) but branch in unexpected ways, because ultimately this kind of endless exploration of permutation and innovation is the only thing that will continue to engage the ever increasing, restless, evolving minds that are its substrate.


 * It will enlighten its participants.** Virtuality, where minds can embody anew, freed from the constraints of biology, geography, economics and law will prove irresistible to ever increasing quantities of participants. For if it remains new, generating combinations and forms inconceivable to planners and designers, constantly incorporating the feedback of participants and changing and reproducing in order to survive, it will maximize connectivity and communication among its participants. This is why humanity has followed an exodus from virtual worlds to social networks. Social networks are designed to maximize interaction and engagement at the expense of everything else (i.e. privacy, originality and, in some cases, personal safety) and have reaped the benefits of their design. Self-organizing virtualities will quickly learn this lesson based on the feedback of their participants and will ‘sweeten the pot’ of stickiness with an ever accelerating phantasmagoria of immersive wonders unavailable to social networks with their flat, stripped-down interfaces. As participation in self-organizing virtualities grows, and connections among participants grow with virtualities quickly appropriating all the good parts of social networks, the scope of the ‘adjacent possible’ (to use another concept of Stuart Kauffman’s, via Stephen Johnson’s Where Good Ideas Come From) increases, and innovation and knowledge growth proliferate as a natural outcome.

In order to begin to bridge the divide between pure speculative theory and a practical, working self-organizing virtuality, we propose the concept of a rules engine that will help to bootstrap the nascent virtuality into existence.

The rules engine will not offer a set of hard and fast laws by which the virtuality will grow and function. Rather, it will present a scaffolding upon which the virtuality can accrete heuristic structure, much as crystalline growth can proceed from a seed crystal. The rules engine will be subject to the same evolutionary algorithms as the virtuality itself, its output being constantly challenged for fitness in a test harness consisting of model virtualities which are engendered for this purpose alone. Much as instructional scaffolding provides an ecosystem and armature for knowledge acquisition in children, the rules engine will present an analogous scaffolding for the virtuality to grow and learn.

An input to the rules engine could be a ‘teaching world’- a legacy virtuality, possibly based on OpenSim or some other open source virtual world platform, that would feed all of its interactions with avatars into a repository. The rules engine could learn from the repository (machine learning) about avatar behaviors, interactions and pathways and track key metrics about membership, stickiness, chat content, and more over time.

There is perhaps room for human observers to influence the development of the rules engine, with the caveat that therein lies potential corruption of the model (self-organizing, evolutionary: universe, life, mind) but with the additional caveat that intervention may be necessary on a purely practical level for the purpose of, if nothing else, setting boundaries of 'reasonableness' beyond which the rules engine attractors should not be permitted to cycle. Also, as the engine gathers its library of behaviors, interactions and pathways from the teaching world, these human observers can act as counselors, guides and teachers, imparting the benefit of 'soft' logic about human culture, language, history and the like so that the engine and its pupil, the nascent virtuality, do not develop absurd patterns or behaviors that make sense to the machine but violate human conventions, mores and cultural practices.

The next step in the training of the rules engine could be to gather empirical data from the internet, and by extension, from the greater culture and world of human affairs. Kevin Kelly in What Technology Wants characterizes the Internet as a “planetary electronic membrane” with “three billion artificial eyes (phone and webcams)”. This membrane ‘sees’ the world, recording more and more of what occurs ‘out there’, and also absorbs increasing swaths of the culture as movies, books, essays, debates, emails (2 million per second) and more flow through its networks and servers. What better conduit could there be for training a nascent virtuality about all aspects of history, culture, commerce, communication and all other domains that together constitute an existing world that has already self-organized? Fed on the Internet and guided by human counselors, the rules engine could gain a first class education in how a world works, and translate this into baseline inputs and rules for the virtuality it instructs.

With these training and educational inputs, the rules engine will be optimally prepared to evolve an heuristic scaffolding upon which the self-organizing virtuality can base its bootstrap and development. As it self-organizes and evolves, the virtuality will feed its acquired knowledge back into the rules engine, the pupil instructing and optimizing the teacher in a virtuous circle that will enable increasingly intelligent virtualities to spawn and flourish.

All of the elements are in place for self-organizing virtualities to begin. The conceptual framework, as outlined above, rests on sound philosophic, artistic, and scientific thought propagated by some of the most innovative minds this planet has produced. The technological capabilities exist, albeit in somewhat siloed form, to begin to build the computational infrastructure. There is no lack of curious, intelligent, forward-thinking people to populate and help to guide and teach each other and the nascent self-organizing virtualities. Once a critical mass has formed there will be no stopping the influx of minds that will grow and shape this novel domain which holds the potential for a new, untethered mode of existence that incorporates all the good parts of its ancestor domains while branching off in an infinite variety of fresh directions and possibilities. Having emerged from the same principles that engendered universe, life and mind -- the same principles that engendered humanity -- self-organizing virtuality will, in a very real sense, become our peer, our partner in exploration and knowledge-seeking. As we teach it (and will it still be something we refer to as ‘it’ ?) and it teaches us, it is entirely possible that a new order of being will begin to emerge.