Welcome to the Cosmopolis

A talk on Wednesday, a new big synthesis of stuff

Tomorrow, Wednesday 13th, at 10 AM PT (1700 UTC) I’ll be doing a talk I’m kinda excited about, aiming to bring together a bunch of questions and themes I’ve been circling all summer, and in a looser way, for several years, across all my projects, both consulting and personal. It’s a level of synthesis I find it easiest to do in big, unwieldy talks (I should call these contraption talks). I last took a shot at synthesizing this particular set of themes at an event in Singapore in 2023 (slides).

My rather ponderous title is Cosmopolis, Metropolis, Nation-State: 3 protocols for articulating civilizational memory. This biggish essay is a prequel for the talk. You don’t have to read (or llmread) it, but you’ll probably get more out of the talk if you do.

Here is the YouTube livestream link (also embedded below). You can also watch via Substack live.

Part of the purpose of this talk is to preview the Protocol Symposium I’m helping organize the week of September 12-19 and drum up participation. It’s a week-long fully online event comprising a Protocol School, a Protocol Foundations Workshop, and a hackathon/coworking track for protocolish projects. The registration deadline for both is August 22 (10 days from today). The symposium is free, but capacity is limited so sign up early.

For me personally, this talk is a first attempt at an ambitious synthesis challenge I’ve set myself — describing a achievable class of cosmopolitan futures I’d actually want to live in. Futures that comprehend both emerging computing technologies and the big, ongoing, decidedly non-cosmopolitan shifts in the global political landscape.

I’ve been writing about these topics for several years now, especially in the Mediocre Computing series (for the tech piece) and the Protocol Narratives series (for the geopolitics piece). These themes also show up elsewhere in my writing and speechifying.

I think I have the broad contours of the synthesis marked out now, so I feel ready to do a first draft talk. What in mechanical engineering is called a test assembly. This essay is by way of being a mise en place for cooking up the talk (I still have to do my slides).

Central to the synthesis is the idea of a cosmopolis.

The Cosmopolis

The core idea in the synthesis is that major new technologies always induce sprawling technologically mediated geographies — or cosmopolises — that don’t map cleanly to conventional geographic units of governance. Instead, they act as a new kind of “soil” for new kinds of societies, driving unbundling and rebundling of geographic units of governance. This is the reason we speak of “land grabs” around new technologies.

The cosmopolis, it turns out, is not just a powerful unit of societal analysis, but a powerful unit of constitutive synthesis, categorically related to the other two geographic units in my title, the nation-state and the metropolis. You cannot set out to build a cosmopolis the same way you can set out to build a nation-state or metropolis, as a missionary project, but you can act intentionally in ways that help it cohere and emerge.

• A nation-state is a territorially defined political unit.

• A metropolis is a dense agglomeration of physical network nodes that coincide within a tight geography, in the form of a converged physical supernode that requires high human density to function.

• A cosmopolis is the geographic field of diffusion of a set of behaviors associated with a particular powerful technology.

Territorial logic, nodal logic, and behavioral logic all induce societal protocols of various sorts, but it is my contention that of the three kinds of entities, the cosmopolis, by virtue of being the most ethereal, tends to co-evolve with technology the most. As a result, a cosmopolis is typically also a technological frontier of some sort, digesting the fruits of ongoing technological evolution into new and persistent civilizational layers that transform political geography.

Cosmopolises are constrained, but not defined, by physical technologies of connection, for both bits and atoms, across space and time. They coincide with nation-states and metropolises at various points, but cannot be identified with either category.

Depending on the generality, diffusivity, and maturity of the catalytic technology, a cosmopolis might span no more than a regional pocket (such as “Silicon Valley” or “Shenzen” in their respective pre-global early decades), extend across continents (such as American and Chinese style internet ecologies today), or the entire planet (such as the global air travel system and the “frequent flyer cosmopolis” it might be said to induce).

We systematically underestimate the power of the cosmopolis as a civilizational unit because it is an emergent entity, lacking the intentional patterns of top-down political organization that defines nation-states and metropolises. Most cosmopolises lack even the rudimentary affordance of a name we can use to point to it.

This does not mean, however, that cosmopolises are undesigned wildernesses. While they do comprehend and accommodate wildness in ways that nation states and metropolises struggle to, they also embody the strongest forms of civilized order.

The element of design enters a cosmopolis as a functionally narrow but composable unit of behavioral logic, the protocol. The architecture of a cosmopolis is a result of a vast number of protocol-design decisions made around a powerful new core technology.

Protocols exist within the territorial logic of nation states and network logic of metropolises too, of course, but they are constitutive in the case of cosmopolises.

When you take away the protocols of a cosmopolis, nothing remains. This is both its greatest strength, and greatest weakness. To a greater extent than competing constructs, it is made up purely of ideas, not materialities. A natural harbor region with pleasant weather and a river running through it is always a proto-city. A region bound by natural geographic borders is always a proto political-territory. But a book without a cosmopolis of literacy is just kindling, or a doorstop.

The computer, and the computational cosmopolis it induces, is merely the most recent cosmopolitical technology. We can identify similar structures throughout history. Arguably, even the Bronze Age ought to be considered the cosmopolis of tin, since it relied on a globalized trade in tin, and associated metallurgical knowledge.

But I want to focus on a particular class of cosmopolises that includes the emerging computational one — those that reshape civilizational memories. We might think of these as first-class cosmopolises. At once the most ethereal and fragile on the one hand, and the most powerful and inexorable on the others.

In the past (particularly in my Breaking Smart essays), I have referred to technologies that induce cosmopolises as soft technologies, but now I prefer the term memory technologies. While all technologies leave traces behind that can be “read” as texts, not all technologies induce intentional modes of remembering and knowing, or new modes of consciousness (new modes of unconsciousness though, are a dime a dozen in every technological era).

Let’s start with a particularly important one: the cosmopolis of print.

The Cosmopolis of Print

The story of print, as it concerns us, is told in one of the big reads that strongly influenced my evolving thesis over the summer, Elizabeth Eisenstein’s The Printing Revolution in Early Modern Europe, (a pick in our book club this year) which I’ve been recommending to everyone.

The book clearly draws out the difference between the technology of the printing press, and the protocols of print culture that took root in the abstract soil it created. The emergent European effects of these protocols — the Renaissance, the Reformation, and the Enlightenment — soon reshaped the entire world.

Based on Eisenstein’s model, the impact of the Gutenberg press can be separated into two components.

First, we have the direct impact, in the form of the emergence of a small, specialized industry (with associated trades and crafts) to build printing presses and keep them supplied with consumables such as lead, ink and paper.

Second, we have the much larger space of print-based protocols that reshaped the rest of society according to the logic of print (what McLuhan called the “Gutenberg Galaxy,” a rather cosmopolitical name; Eisenstein’s book is a more scholarly update to McLuhan’s book). For instance, extensive scholarly travel between sites of manuscript preservation (often monasteries) gave way to a culture of larger personal and institutional libraries and extensive correspondence — the so-called “Republic of Letters” corner of the larger cosmopolis of print. Antiquities-style manuscript trading and scribal copying protocols gave way to distribution and book stocking/selling protocols.

At a larger scale of emergence, we got the awkwardly chimerical constitutive notion of a nation-state, the nation bit being tied to literal soil, in the sons-of-soil (or autochthonous) sense, and the state bit being tied to the more conceptual soil of print culture (in a mode that was made explicit in Benedict Anderson’s notion of imagined communities bound together by printed artifacts like newspapers). Whether through scriptures or constitutions, the post-Westphalian political landscape was populated by entities defined by the creative tension between soil and type. From this tension, early-modern Europe emerged as the core of the cosmopolis of print by 1700.

We can understand the historical process of getting there as normalization through protocolization.

Protocolization as Normalization

The notion of a cosmopolis introduces, somewhat surreptitiously, the idea that an extant technological field is a historically and culturally contingent rather than necessary factor (an idea loosely in the spirit of Yuk Hui’s notions of cosmotechnics and cosmopolitics). More than one distinctive cosmopolis may emerge in response to a technological stimulus, and the set of cosmopolises may not be either mutually exclusive or collectively exhaustive in relation to either the planet or the political world. A cosmopolis is not a planetarity. It is a smaller unit of analysis, and a legibly embodied geographic reality in a way a planetarity is not. We can sketch out cosmopolises on maps.

If you’ve been following my writing for longer than a decade, you’ll note that this is a significant update to my older thinking and writing on this theme.¹

In brief, new technologies induce new normals through protocolization of what is initially a weird and scary sort of monstrousness irrupting across a frontier. Beyond that frontier lies a new kind of territory, a new kind of “soil” on which societies can be built. Protocols are the engines of what I called manufactured normalcy a decade ago, and cosmopolises correspond loosely to what I called Manufactured Normalcy Fields in that essay (though cosmopolis is both a less clumsy and more comprehensive term).

While a major new technology is being protocolized into normalcy, we experience Disturbed Realities. When the process is complete, we get the soil of a new cosmopolis beneath our feet, which we take for granted, and barely notice enough to name. As I argued in the linked article and my 2023 talk, through this process Labatutian unnameables transform into Lovecraftian horrors, and Lovecraftian horrors into Ballardian banalities.

The protocol-generated cosmopolis is the constitutive form of the entity that David Foster Wallace tried to point to in his classic commencement address This is Water. Protocols are how you make a new kind of water out of a new technology.

Print “created” normal modernity by two means.

First, it created a new “soil” for elite governance, shifting political power from land and artisanal crafts based on oral culture to scalable knowledge-based behaviors that required print culture to exist and persist.

Second, it introduced what Eisenstein calls “fixity” (or “hardness” in our emerging language of protocol-speak) into historical memory, stabilizing, refining, and canonizing new understandings of past, present, and future.

The Articulation of Memory

Unlike the cosmopolis of gunpowder, or the cosmopolis of sail (which emerged in the same period), the cosmopolis of print was based on a powerful memory technology. As such, it subsumed the others, and overrode their logics where they were in conflict.

In scribal culture, it is doubtful whether the pen was truly mightier than the sword, but in the print era, there is no doubt — the printing press was definitely mightier than either the gun or the ship (or the converged mercantilist metropolis-based mobile supernode that combined both, the gunship).

Print refactored how we situate ourselves in the world, in space and time. Those who understood their place in the world through print constituted the cosmopolis of print.

Print led to a new kind of articulation of civilizational memory, where by articulation I mean the patterns of joinery and relative mobility in a mechanistically construed assemblage (aka “contraption”!) that constitutes our understanding of the world and modes of agency within it. It is no accident that memory has emerged as a central topic in protocols research over the past three years (

Kei Kreutler

runs a Special Interest Group on memory for the Summer of Protocols program).

Civilizational memory can be understood in terms of three entangled strands — affective, declarative, and procedural. These correspond, roughly, to myth-based, event-based, and econo-legal senses of history.

The US constitution, for instance, embodies all three elements in a single print-based artifact.

1. There is the sacralized mythology (bordering on theology, complete with a reverential hermeneutic tradition) of what the Founding Fathers believed, desired, or intended. This is affective memory.

2. There is the creation of the constitution as a genesis event that occurred in a particular place and time (a metropolitan region — Philadelphia, 1787), with associated discoverable and disputable historical facts. This is declarative memory.

3. And finally, there are the 250 years of legal tradition continuously refining and enacting the idea of America in behavioral terms, through constitutional (and constitutive) understandings of matters such as rights, contracts, currencies, traffic laws, standards for weights and measures, and so on. This is procedural memory.

While all three elements are present in all units of societal organization, to a first approximation, nation-states organize affective memories into a vibe-based territorial logic, metropolises organize declarative memories into capability based physical network supernodes that are dense population centers (there’s a reason a lot of factual history, as opposed to mythology, plays out in major cities), and cosmopolises organize procedural memories into widely diffused infrastructures.

This last category is particularly important to our understanding of protocolization as normalization, since unlike the affective and declarative strands of historical memory, which typically remain in the foreground, procedural strands have a habit of disappearing from view into the background upon maturity, turning into David Foster Wallace’s “water” (my “soil”).

The settling and ossification of procedural memory, into Whitehead’s “operations we can perform without thinking about them,” arguably constitutes the primary dimension of normalization through protocolization. Declarative and affective memories comprise the shifting and transient contents of the protocols of a cosmopolitical dispensation. To the extent they persist at all, they tend to turn into procedural scaffolding.

Territories and cities can forget. Only the cosmopolis truly remembers.

Historical memories either die as fading oozes of affectively and noisily recalled events associated with cryptic, monumental ruins, or turn into the scaffoldings of living realities that we struggle to recognize as history at all. It becomes harder with time to recall the lessons of the fall of the Roman Empire with sufficient gravitas. Affective and declarative memories transform into some mix of camp, vibes, nostalgia, and texts only historians can read. But procedural memories can remain alive and potent in ways we are barely aware of.

The most important parts of the Roman Empire, perhaps, did not decline and fall at all, but turned into an invisible cosmopolis all around us, buried under newer cosmopolitical layers.

Memory as Engineered Argument

One of the more significant conceptual advances we’ve made in three years through the Summer of Protocols program is a deceptively simple definition: a protocol is an engineered argument.

What we understand as normalcy is not an equilibrium so much as an ongoing argument we consciously manage, to navigate endemic creative tensions in a civilizational order. Among the most important such tensions are the ones that arise from “arguments” among older and newer cosmopolitical layers, which take the form of arguments between modes of memory. Tensions not between the old and the new, but between old and new ways of knowing and remembering. New cosmopolises do not resolve inherited historical tensions so much as re-enshrine them in new forms.

Eisenstein’s “fixity of print” is one example of such re-enshrinement of tensions across all three strands of historical memory.

The affective strand of what was still understood as a mythic “Christendom” (defined in relation to Islam through the Crusades) was fixed by re-enshrining the Bible in a set of printed editions corresponding to the political map of 16th/17th century religious conflict. The ecology of the printed Bible mirrored — and to some extent helped ossify into persistent modern borders — the contours of that conflict.

The event-based strand acquired fixity through the establishment of a fixed and and increasingly chronologically accurate relationship to the political events of classical antiquity, establishing a managed tension between the Christian and pagan heritages of Europe.

And finally, the procedural strand acquired fixity as economic-legal traditions began to stabilize through improvements in producing printed currencies, contracts, and governance documents. These re-enshrined governance tensions as old as Greece and Rome in new forms.

The resulting cosmopolis of print, which took shape through wave after wave of transformative macro-level changes, is what we model as the cumulative result of three overlapping epochal events — the Renaissance, the Reformation, and the Enlightenment. A set of newly re-enshrined historical arguments.

The accumulation of procedural memories in the form of the protocols of print created what we understand as “modernity,” roughly between 1770-1825. I pick that period because two modern technologies — fossil power and interchangeable parts manufacturing — began creating the next cosmopolitical layer in that period. It is important to note though, that these newer technologies did not induce first-class waves of protocolization. Though they transformed the planet to an extent that we now recognize a new planetary geological era, the Anthropocene, they did not create new articulations of civilizational memory. Rather, they worked within the articulations wrought by print culture.

If there is to be a new cosmopolitan consciousness for the Anthropocene, it will be induced by the protocols of computation, not print.

Cosmopolises and Consciousness

It is no accident that the same word, enlightenment, is used for both a historical transformation of Europe’s idea of itself, and the sorts of transformations of interiority wrought by spiritual practices. First-class cosmopolises emerge when a technology is potent enough to induce highly contagious transformations both in the material condition of the world and in the minds of the humans inhabiting it. The fact of a new catalytic technology being a memory technology is the primary “tell” of a first-class cosmopolis in the making.

The archetypal kind of technology that has this sort of interior-exterior effect is language. So we should not be surprised to find, in inventorying the cosmopolitical layers of world history, that a large proportion can be characterized in linguistic terms. Languages we typically think of as “classical,” with a sprawling footprint across the elite classes of multiple political regions over a period of time, on a scale larger than the largest units of stable political integration, are among the most obvious ones. The cosmopolises associated with Greek, Latin, Sanskrit, classical Chinese, Persian, and Arabic are prominent examples. The presence of a lively cosmopolis is often thought of as “soft power” projecting from a “hard power” but this is an impoverished understanding of the phenomenon.

I learned the term “cosmopolis” from a book about Southeast Asia that characterized the early history of the region as being part of the “Sanskrit cosmopolis.” In the terms of reference we are developing here, it was a historical articulation of memory. Affective memories still linger in traditions of performance of epics like the Ramayana and Mahabharata. Declarative memories have now hardened into a fairly accurate historiography of the pre-Islamic era, dominated by Hindu and Buddhist kingdoms. And procedural memories take the form of lingering habits of governance loosely clustered around the Mandala concept. And all three feature in still-live tensions.

It is worth noting that the Indic strand of Southeast Asian culture is in fact dominated by a non-Sanskritic language (Tamil). That we still refer to it as the Sanskrit cosmopolis underlines the extent to which cosmopolitanism is elite-coded (in this case, associated with the diffusion of the Sanskrit-based procedural memories of the Brahmin political-administration caste).

The Sanskrit cosmopolis is now not so much dead as buried alive-and-asleep beneath newer cosmopolitical layers — Islamic, European-Colonial, American, modern Chinese. And it sits above older native cosmopolitical cultures, such as the semi-legendary memories of Sundaland, and associated animistic articulations of memory (the focus of a set of explorations earlier this year, South Beast Asia).

A cosmopolis then, is not just a consciousness, it is an accretive and immortal sort of consciousness. To emerge, a cosmopolis must negotiate its presence with older cosmopolitical layers, shaping and being shaped by them, resulting in a re-enshrinement of the past.

And once, installed, cosmopolises can never be buried or killed, since they too turn into legacy articulations of memory that can never entirely be reconstituted within new civilizational logics. Only re-enshrined.

To be a cosmopolitan then, is to identify with, and primarily situate oneself in, the accretive, immortal geology of a layered, evolving conceptual territory whose logic inexorably overwhelms the logic of literal soil over historical time scales.

But over shorter time-scales, the logic of soil matters.

Accretion vs. Exclusion

Given enough time, the behavior-based ordering logic of the cosmopolis transcends both the territorial logic of the nation-state and the structural logic of a particular converged site of intensive knowledge work — the metropolis.

The nation-state is the most recent of a parade of historical forms based on exclusionary control of territory. The present form is characterized by a particular tension between territories and texts that is relatively modern, but shares with its predecessors the logic of land and borders as the dominant organizing principle.

Nation-states obviously demand exclusive control of a territory, organized around dominantly autochthonic identities (the idea of “heritage Americans” is an example of an age-old pattern). Less obviously, they also demand exclusive control of historical memories, constructing them in particular exclusionary ways, erasing inconvenient threads of memory (particularly affective and declarative memory, though procedural memories, despite being hardier, are not invulnerable).

It should come as no surprise then, that the nation-state, throughout its brief existence of 400 years, has consistently defined itself in opposition to the much older, accretive logic of the cosmopolis. The central dilemma of the nation-statist is the curious durability of the cosmopolitical, despite its seemingly more fragile foundations in concepts, rather than the atoms that constitute blood and soil.

That memes can eat genes has been the perennial existential crisis of the territorialist.

The correct view of the metropolis is rather more surprising, and for a long time, I confused myself by identifying the metropolis as the site of the cosmopolitical, rather than a competing class of entity.

Large, diverse, pluralist cities, as is widely recognized, have historically tended to outlast entire eras of exclusionary territorial logic. The city outlived the feudal and imperial eras, and seems set to outlive the industrial nation-state era. Cities like Istanbul have endured through entirely distinct civilizational phases over thousands of years.

The reason is obvious. Even though important cities may be constrained by territorial factors (access to water, grain, energy), they are not defined by them. They are defined by the knowledge capital they aggregate, activate, and harness, in the form of dense agglomerations of human minds. The more enduring metropolises typically sit on fault lines between adjacent but not-quite-compatible territorial logics, playing them off against each other.

The relationship between cities and protocols is not immediately obvious. Certainly, like the cosmopolis, the metropolis evolves as an accretion of living procedural memories. Affective and declarative memories of Christian and pagan-Greek periods may be obscured today, but the procedural memories of both lurk beneath Islamic Istanbul. At a more banal level, infrastructural technologies — roads, waterways, bridges, plumbing — typically endlessly rehearse their histories in tightly confined metropolitan spaces, since there’s typically not enough room to reimagine them from first principles.

What defines a metropolis, in fact, is not the territorial bounding box it exists in, at the pleasure of larger containing entities, but the many overlapping protocols of which it constitutes a converged physical nexus, or supernode. The city has historically been a supernode of so many important networks — education, trade, finance, military power — that one may be forgiven for identifying a cosmopolis with the set of metropolitan regions that happen to dominate it.

Perhaps most importantly, the metropolis, like the cosmopolis, typically also exists in a state of perpetual partial tension with the territorially defined host entities that contain it (most recently, the nation-state).

It is not surprising then, that the term cosmopolitan is often as a synonym for metropolitan. It is tempting to conclude that to first order, they are two sides of the same reality.

They are not.

A cosmopolis, construed as a behavioral logic extending over a region of space-time, is not co-extensive with the metropolitan cores that might make up its backbone. Indeed, the more advanced the technological basis, the more the two diverge geographically. Computer programmers may aggregate in cities, but computers aggregate in datacenters in the technological hinterland. Computer chips may be designed in the American cities, but they are fabricated in Asian suburbs. Cities (and travel between them), may account for a large fraction of energy use, but energy is produced in sparsely staffed refineries in low-population regions, and increasingly, in remote regions boasting plentiful wind or sunshine.

Changes in shipping technology can create and destroy entire port-based cities (containerization, famously, destroyed the breakbulk-port-based economies of many large cities, by moving the heavily automated operations to neighboring smaller cities with cheaper land).

The cosmopolis and the metropolis then, are at best occasional allies, rather than co-extensive realities. To the extent they stand united against the logic of the nation-state and its territorially based ancestors, it is a case of the enemy of my enemy is my friend, rather than a full convergence of interests.

For a great many ideologues and idealists of modernity, this is an uncomfortable idea. To be an architect of computational postmodernity, for many, is to be an architect of the convivial, human-centric city of the future. But in the battle for the post-nation-state future, the metropolis and the cosmopolis are destined to be rivals, competing to fill the growing twin vacuums of collapsing state capacity and nationalism narratives.

Already, some of the signs are clear. Metropolitan areas increasingly feature a deep commitment to “IRL” culture, nostalgic forms of aesthetic and cultural localism at odds with the emerging economics of dense living, wars over housing that rehearse geopolitical patterns of border-based territorial conflicts on a fractally smaller scale, and a worrying fondness for archaic “heritage” infrastructural forms at odds with computation-based cosmopolitical logics.

The Computational Schism

Which brings me to the idea I will be talking about on Wednesday (and possibly writing up as a follow-up essay). The Computational Cosmopolis.

What I hope to do is unpack the emerging logic of global organization being shaped by computation as it nears the end of easy Moore’s Law gains and low energy usage.

We are headed towards a world where computers run everything, consume an alarming (but not unreasonable, when calibrated against biological life) amount of energy, and perhaps most importantly, radically re-articulate historical memory.

When I wrote the Breaking Smart essays a decade ago, many of the most important frontiers of computing were hard to reconnoiter. Blockchains were young, fragile, and seemed like toys. Deep learning was limited to surreal dream-like imagery, translation, and gobbledygook generative text. It had not yet mastered language or solved Go. Self-driving cars seemed so near, yet so far (I took on a long bet arguing that they would not be mainstream in ten years, and won — by a whisker).

The hypothesis that informed that essay collection, “software is eating the world,” clearly needs updating.

Among other things, we’ve realized in the last decade that it’s not about the software but the data. This was a point that was already dimly recognized in 2015. Google researchers wrote an influential paper as early as 2009, The Unreasonable Effectiveness of Data, prefiguring the future we’re now in. The main argument was that more data and simpler algorithms beats less data and more complex algorithms.

With deep learning, we’re realizing the extraordinary extent to which that’s true. Meta-true in fact, since code is now just another unstructured data type, to be gathered in by the exabyte, processed with industrial-scale model training protocols, and provisioned as models with billions (soon to be trillions) of parameters.

The main challenge with scaling AI todays is orchestrating data movements at a vast scale — think gigawatt datacenter scale. Multiplication of matrices of near-cosmic dimensions (as is perhaps appropriate for cosmotechnological infrastructures).

The interplay of computation and data is subtler in the world of blockchains, but equally unmistakeable. While AI deals with data along the quantitative axis, measured in low-precision exabytes, blockchains deal with data along a qualitative axis, working with properties such as immutability, auditability, verifiability, censorship resistance, Sibyl resistance, unbreakable (including quantum-resistant) encryption, long-term (decades to centuries) persistence, and provenance.

If the authors of the 2009 Google paper had thought to cover blockchains (an unreasonable expectation, since the Bitcoin whitepaper was published in October 2008), they might have speculated about the “unreasonable economics of validated data.”

The main problem with scaling blockchains is scaling their historical memory capacity without weighing them down with the enormous computational demands of validating it. Even with the gradual ascendancy of proof-of-stake blockchains and the increasingly favorable comparison with the energy usage of AI compute, blockchains remain energy-hungry beasts. And a big part of that hunger has to do with data handling and memory preservation rather than proof-of-work computations.

The Bitcoin world (which has remained attached to PoW) has scaled back its ambitions, contenting itself with the idea of serving as a post-nation-state payment-rail and store-of-value infrastructure. The Ethereum world will soon embark on what is perhaps the most difficult chapter of its roadmap — what Vitalik Buterin has dubbed the “Purge” phase — advances meant to lower the burden of maintaining the increasingly voluminous historical records in the immutable ledger.

And if you think blockchains and AI represent the peak of the resource demands from the emerging computational cosmopolis, think again. Robotics, sensor networks, mixed reality infrastructures, and the Internet of Things (which may have disappeared from the headlines but continues to grow robustly in the background) will likely increase the energy demands of computation by at least another factor of 2-3.

We’re just beginning to understand that these frontiers of computation too, are primarily defined and shaped by their data-management demands. A robot is primarily a locus of edge-learning with a local firehose of camera and sensor data, and a memory-point-of-view. Every IoT device pumps streams of data in and out of the world’s compute and memory infrastructure, articulating, perhaps for the first time in history, something like a proprioceptive memory in the body politic of the computational cosmopolis. Climate technologies, once we are past the current phase of head-in-sand denialism, will add their own data firehoses to the mix.

Perhaps what we are witnessing is the birth of a cosmopolis characterized by a vastly deeper and more comprehensive kind of fixity than that introduced by print technology. An articulation of civilizational memory so rich, deep, and alive, it constitutes something like a planetary awakening, not merely into a new consciousness, but a new memory of itself.

Perhaps what is happening is that the emerging articulation of civilizational memory is turning eternity — an objective temporal infinity — into immortality (a subjective temporal infinity). This is the essence of the computational cosmopolis.

In my talk on Wednesday, I hope to dive into some of the specifics of what I think is happening, or about to happen, and how best we can situate ourselves within the transformation. I will also share more about the Protocol Symposium, and why that might be a portal for entering the emerging computational cosmopolis for at least some of you. If you’re interested in participating in any component, I recommend you register sooner rather than later. The deadline, again, is August 22.

1. A decade ago, I offered up a notion of “generative pluralism” as my synthesis of technological determinism and social determinism, which constructed the future as, in some sense, a necessary function of the ascendant technologies of an era. Now, while I still think the future is a necessary function of the technological present, I don’t think technology is sufficient to define the future. This opens the door for multiple technological futures. So for instance, there is more than one world that corresponds to the forecast that “software is eating the world.” There is more than one software-eaten world out there, in the fan of futures we are navigating. The subset we end up inhabiting constitutes a set of cosmopolises.