Tag: #data

Posted on

Carbon / Renewables, and Artificial Intelligence: Infrastructure, Power, and the Geography of Computation

The most important assumption embedded in the AI boom is not that intelligence demand will grow. It is that intelligence will remain centralized. This assumption is rarely examined because it has been inherited from the economics of fossil fuels and cloud computing. Yet the rise of local models raises the possibility that intelligence may increasingly resemble renewable energy: abundant, distributed, and locally generated. If so, the consequences extend far beyond technology markets. They would reshape the geography of power itself.

The artificial intelligence boom rests on a surprisingly narrow assumption. Across financial markets, technology firms, governments, and the media, there is broad agreement that demand for computation will continue rising for years, perhaps decades. This expectation underpins hundreds of billions of dollars in investment in data centres, power generation, semiconductor fabrication, and the infrastructure required to support them. Every new hyperscale facility is celebrated as evidence of an inevitable future.

Yet the AI boom assumes more than rising demand. It assumes that intelligence itself will remain centralized.

History suggests this distinction matters. The most consequential technologies are often remembered not for what they did, but for how they were organized. Coal transformed the nineteenth century because it created new relationships between capital, labour, and the state. Oil reshaped the twentieth because its production favoured scale, concentration, and control over strategic resources. The internet altered communication because it reorganized the architecture through which information flowed. Infrastructure, not technology alone, determines who holds power and who depends on whom.

This perspective echoes a broader insight found in the work of Adam Tooze (see my review of his LRB speech from last year) and earlier scholars such as Lewis Mumford and Thomas P. Hughes. Modern power is exercised through infrastructure. Energy systems, financial networks, logistics chains, communications platforms, and technical standards shape economic and political outcomes as surely as governments or markets. The central question is not who possesses resources, but who controls the systems upon which others depend.

Viewed this way, artificial intelligence appears less as a technological breakthrough than as the latest chapter in a much older story. The critical issue is not whether AI becomes more capable. It is whether AI reinforces an existing logic of concentration or introduces a new logic of distribution.

Technological revolutions are often portrayed as disruptions that overturn existing hierarchies. More often, they reinforce them. The industrial revolution concentrated production in factories. Electrification produced centralized grids. The oil economy elevated a small number of producers, refiners, and states to strategic importance. In each case, the enduring consequences flowed from the infrastructure surrounding the technology rather than the technology itself.

Artificial intelligence may prove no different. Today’s debate is dominated by benchmarks, reasoning capabilities, and the race toward increasingly powerful models. Future historians may care less about which model first surpassed a particular threshold than about who owned the computational infrastructure, who controlled access to intelligence, and who captured the resulting economic rents.

The current trajectory clearly favours concentration. Frontier AI models require enormous amounts of capital, energy, specialized hardware, and technical expertise. These requirements create formidable barriers to entry and strengthen the position of a small number of firms whose advantages compound over time. More compute produces better models; better models attract more users and investment; investment finances still larger infrastructure. Scale begets scale.

The pattern closely resembles the political economy of fossil fuels. Oil production rewarded concentration because extraction, refining, and distribution required large-scale infrastructure. The result was not merely economic power but geopolitical influence. Control over energy became a source of leverage; access became a strategic concern. Contemporary discussions of AI increasingly adopt the same language. Semiconductors are treated as strategic assets. Data centers are becoming critical infrastructure. Electricity generation is framed as a prerequisite for national competitiveness.

Yet there is another possibility. While investment flows toward larger centralized systems, local AI models are improving at extraordinary speed. Tasks that recently required cloud-scale resources can increasingly be performed on consumer hardware. More efficient models, better chips, and advances in compression are steadily reducing computational requirements.

This raises a question largely absent from current forecasts: what if intelligence follows a trajectory closer to renewable energy than fossil fuels?

A photo of the hydro generation near IJmuiden aan Zee in the Netherlands.

The comparison is ultimately about power. Oil is concentrated; solar is distributed. Oil creates dependency because production is controlled by a relatively small number of actors. Solar reduces dependency by allowing energy to be generated where it is consumed. The transition from fossil fuels to renewables is therefore not merely an energy transition; it is a transformation in the architecture of power.

A similar distinction may emerge in AI. Frontier models resemble large power stations, requiring immense capital and centralized infrastructure. Local models resemble distributed generation. They may not match frontier performance, but they offer advantages in autonomy, resilience, privacy, and cost.

The conventional response is that efficiency increases demand. Following Jevons (Jevon’s Paradox), many argue that cheaper AI will simply generate more AI usage, just as cheaper storage produced more data and cheaper computation produced more software. This is likely correct. But it misses the crucial point. Efficiency can increase demand while simultaneously changing where that demand is satisfied.

The rise of personal computers increased demand for computation without requiring all computation to remain centralized. Smartphones expanded computing while placing substantial capability directly into users’ hands. Growth and distribution are not mutually exclusive.

The key question, therefore, is not whether demand for intelligence will grow. It almost certainly will. The more important question is whether that demand requires centralized infrastructure. Most users do not need frontier performance; they need systems that are good enough. If local models satisfy most everyday tasks, intelligence may become increasingly distributed even as overall demand continues to rise.

The implications extend far beyond technology markets. Infrastructure shapes political and economic power. A world dominated by centralized AI would deepen dependence on a small number of firms and states. A world of widely distributed intelligence would produce a different balance between autonomy and dependency, concentration and diffusion.

The most important assumption embedded in the AI boom is not that demand for intelligence will continue growing. It is that intelligence will remain centralized. If that assumption proves wrong, the consequences will extend far beyond technology. They will reshape the political economy of computation and, potentially, the geography of power itself.

This argument builds most directly on the work of Lewis Mumford and Thomas P. Hughes. In Technics and Civilization (1934), Mumford argued that technologies are never merely technical; they embody social and political choices, often reinforcing either centralized or decentralized forms of power. Hughes, in Networks of Power: Electrification in Western Society, 1880–1930 (1983), shifted attention from individual inventions to the large technological systems that emerge around them, showing how infrastructure, institutions, capital, and governance become inseparable. Together, they suggest that the key question about AI is not how intelligent it becomes, but what kind of system it creates. This perspective has been extended by Alfred Chandler (The Visible Hand, 1977), Manuel Castells (The Rise of the Network Society, 1996), James C. Scott (Seeing Like a State, 1998), Timothy Mitchell (Carbon Democracy, 2011), David Edgerton (The Shock of the Old, 2006), and more recently Adam Tooze, all of whom examine how infrastructure shapes economic, political, and social power.
Posted on

Translating Administrative Time: Data as Archive, Infrastructure as History in the Formation of Canadian Immigration

This project advances contemporary historiography by treating administrative data as active agents in knowledge production, showing how classification and archival practices shape what is knowable and who is visible. By integrating data-driven methods with historical inquiry, it expands methodological and epistemological approaches while highlighting the politics and contingencies of producing historical knowledge.

From the first moments I began working with immigration records, I was drawn not simply to their volume but to their structure, their silences, and the ways in which they delineate what counts as knowable. Administrative forms, legacy systems, and coding schemes do not merely record phenomena; they enact regimes of legibility that make certain lives, movements, and decisions visible while leaving others obscure. My historical purpose is to investigate immigration data as epistemological infrastructure; to trace the historical logics embedded within the records themselves; and to interrogate how these infrastructures have shaped the knowledge, governance, and social integration of migrants over time. In Canada, where immigration is central to demographic, social, and political life, this investigation carries particular significance. The distinctions embedded in administrative systems: temporary versus permanent, refugee versus economic, authorized versus unauthorized, are not neutral descriptors. They mark differential inclusion and exclusion, structure access to rights and opportunity, and channel life trajectories in ways that unfold across decades and even generations.

The conceptual lens I adopt situates this work within the contemporary data turn. Just as the linguistic turn revealed that language constitutes reality as much as it describes it, the data turn compels us to recognize that administrative records do not passively capture migration. They produce particular ways of seeing, categorizing, and governing mobility. The epistemological stakes of this shift are profound; knowledge is neither transparent nor self-evident. Databases, coding conventions, and legacy infrastructures act as mediators of understanding; they render some patterns readable, some phenomena legible, and others invisible. The work of a historian in this context is to unpack the structures, logics, and assumptions embedded in these systems; to interrogate how these data infrastructures themselves constitute knowledge; and to render visible the historical processes through which knowledge has been produced.

In examining Canadian immigration records, I am attentive to the long-term genealogies of classification, policy, and bureaucratic logic. Categories that distinguish temporary from permanent status, refugees from economic immigrants, or authorized from unauthorized presence are not merely operational tools. They are historically contingent constructs that reflect policy priorities, social anxieties, administrative conventions, and technical constraints. Each field, code, or administrative note carries traces of decisions made by analysts, clerks, and policymakers, whose choices shape both the legibility of migrants and the possibilities for historical reconstruction. By tracing the evolution of these categories, my research illuminates how the state has historically imagined migrants, structured opportunity, and mediated social belonging. In so doing, it foregrounds the interplay between administrative infrastructure, knowledge production, and the social experience of migration.

This project is informed by a dual sensibility that bridges analytic rigour and historical imagination. Administrative records are simultaneously precise and incomplete; they encode patterns yet leave gaps, silences, and ambiguities that demand interpretive work. The historian’s task is therefore translational: to render administrative time legible to analytical and historical time, to preserve provenance and integrity, and to enable longitudinal reconstruction while remaining attuned to the contingencies and biases embedded in the source material. In practical terms, this involves the harmonization of legacy systems such as FOSS, CAIPS, LIDS, and VIDS into contemporary platforms such as GCMS and, in the future, DPM3, while maintaining awareness of the temporal, technical, and policy contexts that shaped their design and evolution. It also entails linking these administrative records to longitudinal datasets such as the IMDB, provincial vital statistics, and Statistics Canada holdings such as the Census, thereby enabling a historically grounded understanding of migration trajectories and outcomes.

A defining dimension of this work is its methodological reflexivity. Immigration data is produced for operational purposes; it emerges from rhythms, constraints, and logics designed to facilitate case management rather than historical reconstruction. As such, the historian must engage in a form of translation that renders these operational temporities and structures legible to long-term analysis. This involves attending to provenance, documenting the evolution of codes, and creating linkages across disparate systems and historical periods. Such work is not merely technical; it is interpretive, epistemological, and historical. Every decision about how to harmonize, integrate, or interpret records is informed by an awareness that data is never neutral.

For instance, consider the historical distinction between temporary and permanent status in Canadian immigration records. These categories are operational; they guide processing, eligibility, and access. Yet they are also epistemic; they shape how analysts, researchers, and policymakers interpret migration flows, integrate newcomers, and assess policy outcomes. The thresholds, definitions, and coding conventions associated with these categories have shifted over time, reflecting evolving policy priorities, social pressures, and technical constraints. Reconstructing these categories longitudinally requires attention to their historical contingency and interpretive framing. It requires tracing not only what was recorded, but how it was recorded, and why it was recorded in particular ways. The historian must interrogate the temporal, institutional, and social processes that produced the data itself, and the consequences of those processes for what can be known and who can be represented.

This methodological reflexivity extends to the integration of legacy systems into contemporary analytical environments. FOSS, CAIPS, LIDS, and VIDS were designed to address discrete operational challenges; they did not anticipate integration into longitudinal analysis spanning decades. Harmonizing these records with GCMS, linking them to the IMDB and provincial datasets, and maintaining categorical integrity are acts of translation, mediation, and interpretation. Each harmonization decision carries epistemic consequences; categories may be redefined, temporal boundaries aligned, and linkages established in ways that preserve analytical fidelity while revealing the historical logic embedded in each system. The historian’s role is to make these processes legible, to document the choices and contingencies involved, and to reflect on how the resulting data architecture shapes both historical interpretation and contemporary knowledge production.

The translational work I undertake is also inherently historical. Data does not exist in a vacuum; it is embedded in social, political, and institutional contexts. Categories, codes, and records encode assumptions about identity, status, and belonging. By tracing these assumptions, we can reconstruct not only patterns of migration, but the epistemic and moral frameworks that underlie them. Administrative distinctions such as refugee versus economic migrant, temporary versus permanent, for example, carry enduring effects on social integration, access to rights, and the life courses of migrants. Longitudinal reconstruction allows us to see these effects across decades and generations, revealing how knowledge infrastructures mediate both historical outcomes and contemporary understanding.

Knowledge production is inseparable from the infrastructures that enable it. In the case of immigration, the categories, fields, and codes embedded in administrative systems are themselves agents of historical formation; they shape what is recorded, what is legible, and what can be interrogated. They establish epistemic boundaries around human movement, differentiating between those whose lives are visible to the state and those who remain peripheral, undocumented, or hidden. To study these infrastructures historically is to recognize that knowledge is not merely extracted from reality; it is enacted, performed, and maintained through bureaucratic, technical, and policy frameworks. This insight compels a dual orientation: we must attend both to the lives documented within the records and to the processes, logics, and assumptions that produced those records in the first place. The two are inseparable; neither the data nor the lived experience can be understood in isolation from the historical infrastructures that mediate them.

Administrative records are themselves temporal objects; they emerge from operational time, which often diverges sharply from the temporalities required for historical analysis. Case processing, workflow cycles, and program deadlines produce rhythms that are not aligned with longitudinal reconstruction or historical comparison. My work seeks to bridge these temporalities by developing methods that translate operational time into analytical and historical time while preserving the provenance, logic, and integrity of the original records. This involves detailed documentation of how systems were designed, how codes were defined, and how processes evolved over time. It also entails creating linkages across disparate datasets, jurisdictions, and decades, enabling historians and analysts to trace trajectories, reconstruct selection logics, and examine long-term outcomes. By treating administrative infrastructures as historical sources in their own right, I aim to render visible the processes through which knowledge is produced, structured, and constrained.

The historical significance of this work becomes clear when one considers the ways in which classification shapes social and political life. Categories such as temporary worker, refugee, or economic migrant do not merely reflect administrative convenience; they constitute frameworks for understanding social worth, civic belonging, and eligibility for rights. These distinctions operate over time, producing effects that extend far beyond the moment of record creation. A person classified as a member of a Designated Class in the 1980s experiences integration differently than an economic migrant in the same decade; their opportunities for settlement, access to services, and pathways to citizenship are shaped by policy, social perception, and the interpretive logic embedded in administrative systems. By reconstructing these categories longitudinally, historians can trace not only outcomes but the epistemic and moral frameworks that produced them. In this sense, administrative data is both archive and instrument: it preserves the historical record and simultaneously shapes the production of knowledge about social reality.

The Canadian context offers a particularly rich site for this inquiry. Immigration has been central to national identity and demographic transformation, and the Canadian state has maintained extensive administrative infrastructures for documenting and managing mobility. Legacy systems such as FOSS, CAIPS, LIDS, and VIDS reveal the historical layering of policy, technology, and bureaucratic practice; their integration into contemporary platforms such as GCMS illustrates the persistence and adaptation of epistemic structures over time. Linking these records to the IMDB, provincial vital statistics, and Statistics Canada holdings allows for the reconstruction of trajectories over decades, enabling scholars to examine long-term outcomes in settlement, health, education, and civic participation. It also allows us to interrogate the evolution of classificatory regimes, showing how policies, categories, and operational logics have shifted in response to political priorities, social anxieties, and technical constraints.

This approach is not merely technical; it is profoundly interpretive. Every choice in data harmonization, categorization, or linkage carries epistemic weight. To collapse temporal variation, reconcile divergent codes, or align fields across systems is to make an interpretive claim about continuity, equivalence, and historical meaning. The historian must therefore be reflexive about the assumptions and consequences embedded in these decisions. Translation is never neutral; it mediates between operational intent and analytical possibility, between past practices and present understanding. By foregrounding these processes, this work makes explicit the epistemic and moral stakes of historical reconstruction and demonstrates that data infrastructures are themselves sites of historical knowledge production.

At a conceptual level, this project challenges conventional understandings of knowledge and classification. The epistemology of state records is neither transparent nor self-evident; it is mediated, structured, and historically contingent. Administrative categories do not simply describe phenomena; they constitute them. To understand human mobility historically, we must therefore examine the processes through which it has been rendered knowable, the instruments through which it has been documented, and the assumptions through which it has been interpreted. This perspective situates my work within broader debates in the history of knowledge, the history of governance, and the emerging field of data studies, contributing to conversations about how epistemic infrastructures shape what can be known, acted upon, and remembered.

The intellectual trajectory that informs this research is itself interdisciplinary, bridging historical inquiry, archival practice, and the analytical rigour of data science. My engagement with legacy systems and contemporary databases has cultivated an understanding of both the technical and interpretive dimensions of knowledge production. It has taught me that precision in coding, integration, and harmonization must be paired with sensitivity to historical contingency, social meaning, and the ethical implications of classification. This dual perspective enables a historically grounded approach to longitudinal research, in which empirical analysis and conceptual reflection are inseparable. By combining these sensibilities, my work seeks to expand the methodological possibilities of immigration history and data-driven social research alike.

Historical examples illustrate the stakes of this approach. Consider the treatment of refugees in Canada during the late twentieth century: administrative categories codified notions of vulnerability, eligibility, and deservingness; they also reflected broader social and political anxieties, such as attitudes toward asylum seekers or debates over labour market needs. By tracing how these categories evolved across decades, one can reconstruct not only the patterns of settlement and integration but also the underlying epistemic frameworks that shaped public perception, policy design, and bureaucratic practice. Similarly, distinctions between temporary foreign workers and permanent residents reveal how labour needs, migration policy, and social hierarchies were encoded within administrative systems. These cases demonstrate that administrative infrastructures are not neutral repositories; they are active participants in the historical processes that structure human life, belonging, and opportunity.

The broader significance of this research extends beyond historical reconstruction. In an era dominated by the data turn, understanding the historical formation of epistemic infrastructures is essential for evaluating contemporary policy, governance, and social practice. By revealing how knowledge has been produced, mediated, and constrained, this work offers insight into the ethical and analytical responsibilities of researchers, policymakers, and institutions. It highlights the ways in which administrative categories can reproduce inequality, shape opportunity, and influence social perception. At the same time, it provides tools for rigorous longitudinal analysis, allowing scholars to reconstruct trajectories, interrogate selection logics, and examine long-term outcomes in ways that are both historically grounded and analytically robust.

Ultimately, my historical purpose is to make visible the infrastructures through which migration has been rendered knowable, to interrogate the epistemic and moral assumptions embedded within administrative systems, and to explore the consequences of these structures for both scholarship and social life. This work bridges empirical analysis, historical reflection, and methodological innovation, demonstrating that administrative data is not merely a technical tool but a site of historical knowledge production. By tracing the evolution of categories, codes, and systems, I aim to illuminate the interplay between policy, bureaucracy, and human experience; to reveal how knowledge infrastructures structure both possibility and constraint; and to contribute to a more nuanced, reflexive, and ethically aware understanding of migration in Canada and beyond.

Through this research, I seek to advance historical methodology, deepen understanding of Canadian immigration, and expand the conceptual frameworks through which data and history intersect. It is a project that integrates technical expertise with historical imagination, methodological rigour with interpretive sensitivity, and archival practice with theoretical reflection. By engaging with the infrastructures of knowledge themselves, I aim to demonstrate that history is not only about events, people, and policies; it is also about the instruments, categories, and processes through which the past becomes knowable, legible, and meaningful. In pursuing this purpose, I hope to contribute to a scholarly tradition that is attentive to the ethical, epistemological, and social dimensions of research, while offering new tools for understanding the complex interplay between data, governance, and human experience.

Relevant published works:

The Order of Things: An Archaeology of the Human Sciences by Michel Foucault
Foucault’s work examines the historical formation of epistemes, the underlying structures that make knowledge possible within a given era. For this project, it provides a conceptual foundation for understanding immigration data as historically contingent knowledge; administrative categories, coding schemes, and legacy systems are not neutral reflections of reality, but products of specific epistemic frameworks. Foucault’s analysis supports my argument that data infrastructures themselves enact knowledge, determining who and what is legible within the bureaucratic archive.

How We Think: Digital Media and the Future of the Humanities by N. Katherine Hayles
Hayles foregrounds the materiality and mediation of knowledge in digital and computational contexts, emphasizing how coding, databases, and technical infrastructures shape human understanding. This perspective is directly relevant to the translational and harmonization work in my project: legacy immigration records do not naturally yield historical insight. They must be interpreted, linked, and rendered legible across temporal and technical boundaries. Hayles’ emphasis on the interaction between human interpretive work and infrastructural mediation informs the project’s methodological approach and justifies a reflexive stance toward data as both archive and instrument of knowledge.

The Data Revolution: Big Data, Open Data, Data Infrastructures and Their Consequences edited by Rob Kitchin – this work situates data infrastructures within social, technical, and institutional contexts, highlighting that design choices, governance structures, and classification systems actively shape what can be known and what remains invisible. This aligns with my project’s focus on immigration records as epistemic infrastructure: coding schemes, legacy systems, and administrative categories not only organize information but constitute the very possibilities of knowledge about migration. Kitchin’s work provides conceptual tools for thinking about longitudinal linkages, interoperability, and the politics of classification, directly supporting my methodological and epistemological aims.

Posted on

Assembling Knowledge: Layered Pivots, Constellations, and the Limits of Statistical Aggregation

Efficiency in data work is not only a technical concern; it is a question of knowledge and power. True understanding arises when we attend to relationships and structure rather than simply accumulating more data, revealing the limits of sheer aggregation. The frameworks and lenses we use to interpret information determine which patterns are visible and which remain hidden, showing that context, relational positioning, and choices about focus are as decisive as the data itself. Nerd alert.

The Single Source Shortest Paths problem is often introduced as a simple question about travel; if you begin in one town, how can you find the shortest routes to every other town in the region. Dijkstra first presented his solution in 1956 as a story rather than a technical proof; he wanted readers to grasp what a computer could do by imagining a traveller who keeps careful track of every possible road. His method guarantees perfect accuracy, although it requires the algorithm to examine each location with the same level of attention.

More recent work shows that full precision does not depend on such rigid order. Researchers discovered that some locations matter more than others; these high value points can serve as pivots that reveal much of the network once they are checked. Instead of marching through every town one by one, the newer approach looks outward in wider bands that align with the structure of the landscape. This creates a picture of how information flows through a system when you concentrate on the most catalytic parts. The network becomes something you assemble through patterns of relation rather than simply something you trace step by step.

This distinction parallels a critique raised by Song-Chun Zhu: large-scale statistical aggregation alone, even with immense data, does not capture the structural elegance necessary for knowledge or artificial general intelligence. Aggregating observations is akin to noting all the stars in the night sky individually, without recognizing their patterns or relational significance. Zhu emphasizes that knowledge emerges not from sheer volume, but from the principled organization of relational structure. Pivot-based SSSP embodies this idea: nodes are selected and processed for structural leverage, generating insight through relational activation rather than exhaustive comparison.

To conceptualize this, imagine constructing a constellation in the night sky, such as Orion. Each node in the graph is a star, but stars are not merely observed—they are activated relationally, connected through pivotal anchor points. The three stars of Orion’s Belt serve as deep pivots, defining central structure and guiding the integration of surrounding stars such as Betelgeuse and Rigel. Other stars are integrated relative to these anchors, their positions determined by proximity and influence. Importantly, how humans perceive these stars varies: in Western tradition, we see Orion; in ancient Egypt, these stars are associated with Osiris; in Chinese astronomy, they are part of the White Tiger of the West; and in Lakota tradition, they form part of a bison. Each culture identifies different anchors and connections, reflecting relational interpretation rather than fixed identity. Graph nodes behave in the same way. Their significance emerges from how pivots organise the neighbourhood and guide the spread of influence. Similarly, in SSSP, pivots are chosen for structural leverage, not intrinsic identity. By prioritizing nodes closest to these anchors, large portions of the graph can be certified early, minimizing redundant exploration while dynamically shaping emergent structure.

These computational strategies have clear philosophical parallels. Husserl uses the act of bracketing to set aside what is not essential; in a similar way the algorithm sets aside low impact nodes for a moment while it concentrates on high leverage pivots. Walter Benjamin describes constellations in which distinct historical fragments are placed together so that new patterns appear; pivot bands work in a comparable spirit since they draw clusters into relation based on significance rather than strict sequence. Latour’s actor network theory also resonates since each node becomes an active participant whose influence depends on its position within a web of relations; the algorithm brings certain nodes into play at key moments so that they can shape the unfolding structure.

The larger lesson concerns how we come to understand complex systems. Efficient computation mirrors efficient knowledge building; attention is directed toward high leverage elements, layered activation prevents redundant effort, and careful checks preserve coherence without requiring examination of every part. Zhu’s critique reminds us that collecting all stars or mapping every edge does not automatically produce understanding. True insight emerges through structured and relational activation; this is precisely the mechanism that gives layered pivots their power in the Single Source Shortest Paths problem.

In this light, layered pivots that prioritize proximity function much like a constellation interpreted through multiple cultural lenses; they reveal that structure, relationality, and principled activation are universal principles for both computation and knowledge. The correct sequence of activations, the appropriate prioritization of influence, and a well-chosen stopping point produce coherent insight efficiently; networks and ideas arise dynamically from the interplay of structure and relational activity rather than from raw observation alone.

Posted on

The Algorithmic Turn: Emergent Processes and the Reformation of Knowledge

This is a meditation on the shifting agency of algorithms—once confined to calculation, they have emerged as active forces in the generation of knowledge. It reflects on how this transformation unsettles conventional ideas of authorship, intention, and understanding, inviting us to reconsider the delicate interplay between human thought and machine influence in shaping our reality. A continuation of my earlier post Abstracted Intelligence: AI, Intellectual Labour, and Berkeley’s Legacy in Public Policy. A reading list is below.

The algorithm has quietly evolved from a tool of calculation into a generative force shaping the very terrain of knowledge. No longer confined to precise computation alone, it now participates actively in structuring how we understand, interpret, and create. As Wendy Chun demonstrates, these systems do more than process inputs—they habituate us, embedding themselves deeply into our cognitive and social rhythms. This evolution signals a fundamental reconfiguration of knowledge itself: no longer solely the product of human cognition or systematic observation, knowledge emerges through recursive, machine-driven processes that entwine human and computational agency.

At the heart of the algorithm lies a set of rules designed to produce outcomes, but its function has expanded far beyond problem-solving. Luciana Parisi’s insight into algorithmic speculation captures how these processes generate novelty and reshape aesthetic and epistemic landscapes rather than simply calculate or represent. Algorithms now inhabit artistic, cultural, and social realms where they do not merely answer questions but frame the very logic through which questions arise. As Alexander Galloway emphasizes, the algorithm operates at the level of interface—a mediator where legibility is constructed and constrained, and where meaning becomes both possible and limited. This shift subtly relocates authority: from human hands to encoded processes, from fixed categories to contingent and often opaque patterns.

The consequences of this shift are profound. Tarleton Gillespie’s work reveals the infrastructural labour behind these systems, which govern visibility and legitimacy in ways frequently invisible to those governed by them. Algorithms do not simply replace human decisions; they reconfigure the conditions of decision-making itself, often beneath the surface. Their generative capacity introduces complexity and opacity, producing outcomes that exceed the understanding of their creators. These recursive patterns complicate verification and accountability, exposing a form of epistemic vulnerability that challenges traditional frameworks for knowledge and governance.

Expanding this perspective, Benjamin Bratton situates algorithms within a planetary computational architecture that transcends local or institutional boundaries, reconfiguring sovereignty, cognition, and identity at a global scale. This shift implicates knowledge production in a vast technical stack that governs infrastructures of power and information flow across geographies and societies. Kate Crawford grounds these theoretical insights in material realities, illustrating how AI and algorithmic systems are embedded in extractive economies, labor conditions, and environmental costs. What may appear as immaterial knowledge production is inseparable from physical and political infrastructures that shape and constrain the possibilities of computation.

Viewed through this lens, algorithmic processes resemble dynamic narratives unfolding through layers of input, context, and recombination. Like storytellers without fixed authorship, these systems orchestrate data flows and conditional operations to produce forms that exceed their components. The outputs are not passive reflections but active interventions that reorient our relationship with knowledge—from stable transmission toward real-time interpretation and negotiation. This dynamism signals both power and precariousness, demanding ongoing reassessment of assumptions and a willingness to confront the shifting locus of interpretive authority.

The visual arts offer a vivid example of this transformation. Generative algorithms produce imagery that moves beyond imitation to invention, collaborating with human creators while introducing unpredictability and chance. This interplay opens new aesthetic spaces but carries risks: the flattening of complexity, amplification of bias, and erosion of clear boundaries between authorship, intention, and effect. The algorithm becomes a co-creator and gatekeeper, shaping the field of possibility even as it expands it.

This transformation reflects a deeper epistemological turn. Knowledge no longer appears as fixed or discrete but emerges within dynamic, recursive systems that resist containment or full comprehension. Algorithms function as agents in the production of meaning, their agency demanding reflection on not only what they enable but also what they obscure or distort. In both artistic and intellectual practice, the tension between human intention and algorithmic variation generates new possibilities while compelling vigilance. When opacity deepens and systemic influences become normalized, the risks extend beyond creativity into the realm of knowledge itself.

This challenge recalls earlier philosophical critiques of abstraction and the limits of knowledge that I have talked about before. The eighteenth-century philosopher George Berkeley, for instance, challenged the legitimacy of abstract mathematical entities—infinitesimals—that lacked direct empirical manifestation. Such critiques resonate today as we grapple with algorithmic processes that often operate as “ghostly inferences,” producing outcomes whose internal workings and assumptions remain intangible or obscured. Like Berkeley’s warning against unmoored abstractions, this calls us to critically examine the epistemic foundations and consequences of the algorithmic turn. See my post on Berkeley for more here.

Emerging from this shift is a new epistemic condition: knowledge as emergent, relational, and mediated through evolving systems. In this environment, we become not only interpreters but stewards—charged with critical engagement and ethical responsibility for the infrastructures of meaning that shape our world. This requires embracing process over product, contingency over fixity, and acknowledging the redistribution of agency from cognition to computation, from conscious intent to iterative dynamics. The challenge moving forward is to interrogate not only what these systems make possible but to ask persistently under what assumptions, for whose benefit, and at what cost.

Machine art from an exhibition a few years back.

A short reading list from sources that I have read over the last few years on this topic.

Taken together, these six works form a conceptual constellation that reframes the algorithm not as a neutral instrument, but as an active participant in the production of knowledge, culture, and power. Wendy Chun foregrounds how algorithms habituate us, not just through interface but through repetition and memory, revealing the affective and social dimensions of computation. Luciana Parisi pushes further, showing that algorithms speculate—they generate rather than merely calculate—thus altering aesthetic and epistemic landscapes. Galloway’s analysis of the interface illuminates the algorithm as a mediator of meaning, a site where legibility is constructed and constrained. Tarleton Gillespie turns to the infrastructural labour behind algorithmic systems, exposing how platforms subtly police visibility and legitimacy under the guise of neutrality. Benjamin Bratton scales this transformation globally, mapping a planetary computational architecture that reconfigures sovereignty and cognition alike. And Kate Crawford grounds these abstractions in the material and political, revealing how AI and algorithmic systems are inseparable from extractive practices, labour exploitation, and environmental cost. As a group, these texts chart a shift in thought: from seeing algorithms as tools of control to understanding them as environments—generative, recursive, and contested—within which control, creativity, and understanding are continuously renegotiated.