Snowflake is reportedly cutting around 400 jobs. The company had been screen-recording every documentation session for eight months, building training datasets from its senior technical writers’ workflows.

• Knowledge transfer: Senior writers spent their final six weeks transferring knowledge directly to the AI system that would replace them.

• The claim: Snowflake says documentation quality has not dropped. Management is celebrating 300% efficiency gains from the new AI documentation pipeline.

• The scale: The reduction is part of a broader workforce cut targeting technical writing and documentation teams specifically.

This is one of the most concrete examples yet of a company systematically using its own employees to train their replacements. The 300% efficiency claim will get the headlines. The eight months of screen recording will stick in people’s minds longer.

Read more at GenAI.works

Abstract

Historians generate knowledge through the labour of composition – through the friction between interpretation and evidence that makes claims open to scrutiny and challenge. This essay argues that when composition is bypassed, that structure disappears. Generative AI raises this issue in urgent fashion. Current large language models produce what the essay terms ‘stochastic history’: prose that replicates the surface forms of historical explanation without enacting the disciplinary reasoning behind them. Such prose flattens temporal complexity into chronological adjacency, inherits narrative patterns without deliberating over them and reproduces hegemonic framings without the mechanisms – archival friction, peer contestation and historiographical consciousness – through which the discipline revises itself. Recent studies measuring AI’s applicability to historical work capture transmissive tasks while remaining blind to the interpretive core; approaches that identify textual markers of historical thinking detect symptoms that can be simulated, not the compositional process producing them. The consequences extend beyond the profession. When stochastic history circulates in classrooms, policy research and public media, non-specialists encounter pasts stripped of contingency and contestability – pasts that naturalize present arrangements rather than rendering them open to challenge. The defence of writing advanced here is methodological rather than nostalgic: It preserves the conditions under which historical claims can be scrutinized and revised.

https://doi.org/10.1111/1468-229x.70092

This essay explores whether generative AI would be better understood as a cognitive and technical prosthesis. Rather than rehearse familiar debates about whether AI replaces human labor, I want to examine how it redistributes agency without redistributing responsibility, how it reorganizes the temporal and technical expectations that institutions place on writing, reading, judgment, and making, and how it opens terrain that was previously fenced by disciplinary credentialing and cognitive normativity alike. The last of these — the emergence of what might be called prosthetic competence, both technical and cognitive — has received too little attention, and it is here that the politics of the metaphor becomes most concrete.

The Prosthesis Frame

The usual framings are not wrong so much as flat. “AI is just a tool” borrows from an older moral drama about machines and human exceptionalism. “AI is coming for your job” borrows from another. Both flatten what is happening at the level where historical change actually registers: in routines, infrastructures, and the quiet redefinition of what counts as adequate performance.

“AI as prosthesis” gets closer to the grain. A prosthesis is not a hammer. You do not merely wield it. You attach it. You learn its resistances and affordances. You adjust your posture around it, and in time your world reorganizes so that the device is no longer an external object but part of a coupled system: body, habit, environment, maintenance regime. The metaphor matters because it shifts attention from the question of replacement to that of reconfiguration — how the coupling alters what a person can do, what institutions expect of them, and where governance sits once the interface becomes habitual.

The philosophical architecture for this claim rests on work that Andy Clark and David Chalmers began in 1998. Their “extended mind” thesis argued that certain external supports function as integral components of cognition when they are reliable, readily available, and routinely used (Clark and Chalmers 1998). Once the notebook or laptop is always there, always consulted, and always trusted, it stops being “outside” in any analytically meaningful sense. It is part of the thinking apparatus. Clark developed the argument further in subsequent work, insisting that the human mind is fundamentally a “leaky organ” whose boundaries extend into the material world through repeated, trust-calibrated coupling with external systems (Clark 2003, 2008). The insight was never that humans become machines. It was that cognition has always been scaffolded by material systems that stabilize memory, calculation, and attention over time. That is the sober corrective we forget whenever we talk about “intelligence” as if it lived entirely inside the skull.

Yet disability studies adds something the extended mind frame leaves underspecified: normativity. Aimi Hamraie’s work on universal design is essential here. Hamraie demonstrated that access is never a purely technical question but an argument about which frictions count, which bodies are treated as default, and which forms of dependence are rendered acceptable or shameful (Hamraie 2017). Prosthetic devices are not simply enabling. They are shaped by social demands for productivity, legibility, and compliance, and they arrive tethered to institutional scripts about what a body or mind should do. “Assistance” can be emancipatory. It can also be coercive. The line between the two is historically mobile, and it moves in the direction that institutions find convenient.

David Mitchell and Sharon Snyder’s foundational work on what they called “narrative prosthesis” adds a further dimension. Mitchell and Snyder showed that disability functions in literary and cultural forms as a device that props up meaning: a scaffolding for plot and moral argument (Mitchell and Snyder 2000). The concept names how physical or cognitive difference is recruited to do narrative work for an ableist culture, and the warning transfers with only minor modification. “AI as prosthesis” can become a story we tell to naturalize dependence, to aestheticize the interface, to treat structural capture as personal convenience. It can render the new arrangement inevitable.

If we keep Hamraie, Mitchell, and Snyder in view alongside Clark and Chalmers, “AI as prosthesis” ceases to be a cute metaphor and becomes a diagnosis. It forces us to ask not only what the device enables, but what it normalizes.

The Obvious Pleasures and the Quiet Ratchet

Start with what the prosthesis does well. The system drafts. It summarizes. It translates. It converts rough notes into a coherent outline, suggests transitions, and can serve as a tireless interlocutor for brainstorming — a way to make writing less lonely, less stuck. If you do intellectual labor for a living, the appeal is not mysterious. In fact, the appeal is precisely what makes the prosthesis frame so apt: the more useful the device, the more quickly it becomes habit, and the more habit becomes expectation.

That is where politics enters, because expectations are never evenly distributed.

In workplaces, the prosthesis will not simply make workers “more productive.” It will recalibrate what counts as adequate performance. Turnaround times shorten. Output volumes rise. The baseline shifts. The new norm is not “work plus AI” but “work as if AI were already attached.” This is a familiar historical sequence: technologies that promise to reduce burden often intensify it by creating new standards of responsiveness, availability, and polish. The device becomes ordinary. Then it becomes compulsory — first informally, then formally, and finally by the quiet violence of comparison against those who have not yet coupled.

Education is an even sharper site of contradiction. Once AI becomes prosthetic for drafting and comprehension, institutions face a choice they would rather avoid: either redesign assessment around the reality of coupled cognition, or pretend the coupling is illegitimate while everyone knows it is happening. The temptation will be to police boundary violations, because boundary policing is what institutions do when they cannot or will not redesign their own metrics. Yet the prosthesis metaphor makes the policing look strange. If cognition is now, in practice, distributed across human and machine systems, why is the student treated as an isolated unit for purposes of evaluation? And if the student is still treated as an isolated unit, why is the institution surprised when the student routes their work through the most available cognitive infrastructure? This is not a moral question first. It is a design question that institutions keep trying to resolve through punishment.

Prosthetic Technical Competence

The coupling extends beyond writing and thinking into terrain that has been less examined: the capacity to build. Before generative AI, the ability to create a functioning web application, construct a relational database, generate interactive data visualizations, or write a script that processed thousands of archival records required either years of technical training or sufficient institutional resources to hire someone who had that training. The boundary between those who could make computational objects and those who could not was sharp, credentialed, and largely coterminous with disciplinary affiliation. Historians, anthropologists, literary scholars, and artists occupied one side. Computer scientists and software engineers occupied the other. Digital humanities programs tried to bridge the gap, but the bridge was narrow, and the toll was high — measured in semesters of coursework that competed with disciplinary demands rather than complementing them.

Generative AI is dissolving that boundary. Not by making everyone a programmer, but by converting programming from a skill requiring years of syntactic fluency into a collaborative practice requiring domain knowledge, clear specification, and iterative judgment. The historian who can articulate what a database should organize, what an interface should display, and what an archival query should retrieve can now produce a working prototype by describing those requirements to an AI system that translates specifications into code. The prosthesis does not replace the need to think carefully about structure, logic, and purpose. It replaces the need to memorize a programming language’s grammar in order to express that thinking as executable software.

The implications are significant. A scholar studying maritime trade routes can build a geospatial application that renders shipping data as interactive corridor maps — not by learning JavaScript and Leaflet from scratch, but by specifying what the visualization should show, testing the output, identifying errors, and iterating. A researcher working with thousands of newspaper articles can construct a script that extracts date-stamped keyword frequencies across multiple publications, producing the kind of longitudinal dataset that would otherwise require either a funded research assistant or months of manual tabulation. A teacher who wants students to interact with primary sources can build a web-based interface tailored to the specific pedagogical problem — sorting, comparing, annotating — without submitting a grant application to a digital humanities center and waiting eighteen months for a developer’s time.

What is happening here is not the democratization of coding in the triumphalist sense preferred by Silicon Valley. It is something more specific and more interesting: the redistribution of technical competence along lines that privilege domain expertise over syntactic fluency. The person who knows what the archive should organize, what the dataset should reveal, what the interface should render visible, gains the capacity to build the thing that does it. In effect, the prosthesis recodes the boundary between “having an idea for a digital project” and “being able to make one” from a credentialing barrier into an iterative conversation.

That conversion has particular force in fields where computational methods have long been recognized as valuable but remained inaccessible to most practitioners. In historical research, the gap between what scholars know they could learn from large-scale textual analysis, network mapping, or spatial modeling and what they can actually build has been a persistent structural constraint. Generative AI does not close that gap entirely — the prosthesis has its own limitations, its own patterns of error, its own tendencies to produce plausible but incorrect output. Yet it narrows the gap enough to shift who can do computationally intensive work from a credentialed minority to a much wider population of domain specialists willing to learn the collaborative discipline of specifying, testing, and repairing.

The skeptic’s objection is fair and needs to be absorbed rather than dismissed: code produced through prosthetic coupling may be functional without being robust, legible without being secure, and sufficient for a prototype without being adequate for production deployment. This is accurate as far as it goes. Yet the objection draws the wrong boundary around the problem. For the historian building an interactive archive, the literary scholar constructing a textual analysis pipeline, or the teacher designing a classroom tool, the relevant standard is not whether the code meets enterprise software engineering criteria but whether it produces a reliable, functional object that would not otherwise exist. The alternative to prosthetic code is not better code. In most cases, there is no code at all. The prosthesis creates capacity where there was none, and the question of quality, as real as it is, remains secondary to the question of possibility.

Prosthetic Access and Cognitive Normativity

The redistribution of competence extends beyond disciplinary boundaries. It reaches into the politics of cognition itself.

Institutions have always defined “competence” through specific output formats: linear prose, timed examinations, structured oral presentations, and sequential task completion. What presents as a neutral standard of intellectual performance is, in practice, a sorting mechanism calibrated to neurotypical processing. Hamraie’s analysis of how design regimes produce “the normate” — a subject position whose unmarked status renders its own contingency invisible — applies directly to cognitive assessment (Hamraie 2017, 19–22). The person with dyslexia is not less capable of historical analysis; they are penalized by an evaluation regime that treats fluent written production as a proxy for analytical thinking. The person with ADHD is not less capable of sustained intellectual work; they are penalized by institutional temporalities that treat linear task sequencing as the only legitimate workflow. The autistic researcher whose pattern recognition exceeds their neurotypical peers is nonetheless disadvantaged by assessment structures that reward a narrow repertoire of communicative conventions. In each case, the institution measures deviation from a cognitive default and records that deviation asa deficit. The deficit is institutional. The label lands on the person.

Margaret Price’s work on “mental disability” in academic culture is clarifying here. Price argued that the norms governing academic discourse — the seminar, the conference presentation, the written examination — are not neutral containers for intellectual exchange but embodied practices that systematically privilege certain kinds of minds while rendering others incoherent or invisible (Price 2011). The classroom is not merely a place where content is assessed. It is a site where cognitive normativity is performed and enforced. What counts as “participation,” “clarity,” and “rigor” are not objective descriptors but institutional conventions that carry the full weight of ableist assumptions.

AI as a cognitive prosthesis intervenes at exactly this joint. It converts between cognitive modes: translating spatial or associative thinking into linear prose, handling surface-level mechanics so that a writer with dysgraphia can concentrate on argumentation, and providing structural scaffolding for someone whose executive function processes organization differently. The system can serve as an intermediary between how a person actually thinks and the institutional format that has historically gatekept the expression of that thinking. In effect, the prosthesis separates the idea from the narrow channel through which institutions have demanded it travel.

The implications for cognitive equality are real but require careful specification. For neurodivergent users, the prosthesis does not fix a deficit. It routes around an institutional barrier — one that was never a measure of intellectual capacity but rather a measure of conformity to a particular cognitive style. The student with dyslexia who uses AI to handle sentence-level mechanics and produce polished academic prose is not cheating. They are accessing the same expressive range that neurotypical students reach through a processing pathway the institution happened to privilege. The researcher with ADHD who uses AI to organize scattered but brilliant associative insights into a structured argument is not being “helped” in the condescending sense. They are using a prosthesis that compensates for an institutional design failure: the failure to recognize that analytical capacity and linear organizational fluency are separate competencies that happen to have been bundled by convention.

Yet the critical edge matters here, and there are at least two tensions worth holding in view. The first is the question of whether prosthetic access produces genuine cognitive equality or better masking. If the institution never revises its norms — if it continues to treat linear prose, timed output, and sequential processing as the unmarked standard — then AI does not accommodate neurodivergent cognition so much as make neurodivergent people more legible to systems that never examined their own assumptions. The person still has to pass through the institution’s preferred format. The prosthesis simply makes the translation invisible. That is access of a kind, but it leaves the normative architecture intact and places the burden of adaptation entirely on the user rather than on the institution that designed the barrier. What presents as accommodation is, in practice, assimilation by other means.

The second tension is that the dependency problem strikes harder here than anywhere else in the prosthetic landscape. If the device is what makes the playing field level, then a pricing change, a model update, or an expired institutional license is not an inconvenience. It is a revocation of access. For the neurotypical user, losing the prosthesis means slower writing, more effortful organization, reduced output — a degradation of convenience. For the neurodivergent user whose equitable participation is routed entirely through the prosthetic interface, losing the prosthesis can mean losing the capacity to participate at all. Dependence, in this case, is not a market relation. It is a civil rights question dressed as a subscription fee.

From Prosthesis to Cyborg

At some point, the prosthesis frame encounters its own limits. A prosthesis assumes separability: you attach the device, you can detach it, the boundary between user and system remains analytically legible even when practically blurred. Yet the cases accumulating in this essay press against that assumption. The historian whose research methodology now presupposes iterative human-machine conversation, the neurodivergent student whose institutional participation is routed entirely through the interface, the scholar whose workflow has reorganized so thoroughly around AI coupling that removal would not restore a prior state but produce a different and diminished subject — these are not cases of a person using a detachable tool. The coupling has become constitutive.

The concept that names this threshold predates current technology. Manfred Clynes and Nathan Kline coined “cyborg” in 1960 to describe an organism-machine integration so thoroughgoing that the boundary itself dissolves — not as metaphor but as a functional description of a self-regulating human-machine system in which the mechanical components become as integral as the organic ones (Clynes and Kline 1960). Their original context was space travel: an astronaut whose physiological regulation was partially offloaded to implanted devices would be free to concentrate on exploration rather than survival. The insight was that coupling, once sufficiently deep and habitual, ceases to be supplementary. It becomes structural.

Donna Haraway radicalized the concept by insisting that the cyborg was already the condition of late-twentieth-century subjectivity, not a future to be anticipated but a present to be reckoned with politically (Haraway 1985). Haraway’s cyborg refused the clean boundaries between organism and machine, between nature and culture, between the self-sufficient liberal subject and the technologically enmeshed one. The “Cyborg Manifesto” argued that these boundaries were never stable to begin with and that their maintenance served ideological functions: naturalizing certain forms of autonomy while rendering others deviant. For Haraway, the political task was not to resist cyborgization but to claim it — to insist on the right to narrate one’s own coupling rather than having it narrated by the institutions and markets that profit from it.

Cyborgization — treated as a process rather than a fixed state — names what happens when the prosthesis passes the habit threshold. Clark’s later work on “natural-born cyborgs” converges here from a different direction, arguing that human beings are by nature prone to deep integration with cognitive technologies and that the boundaries we draw between “self” and “tool” are matters of convention rather than ontology (Clark 2003). The question is not whether cyborgization occurs. It always has, from writing systems to eyeglasses to smartphones. The question is what changes when the coupling involves a system that generates, reasons, and adapts, and when the infrastructure that sustains the coupling is controlled by commercial platforms operating at a planetary scale.

The transition from a prosthetic to a cyborg relation changes the governance stakes. You govern a prosthesis through regulation of the device: its safety, its availability, and its terms of use. You govern a cyborg relation by regulating the person-system composite, thereby shaping the conditions under which cognition itself is constituted. No existing institutional or legal framework is designed to do that. Employment law assumes that an employee's capacities are their own. Educational assessment assumes a student whose cognition is bounded by their skull. Intellectual property law assumes an author whose creative contribution is distinguishable from the contributions of their tools. Each of these assumptions becomes unstable once the coupling is deep enough that separating human contribution from machine contribution is not merely difficult but conceptually incoherent.

For neurodivergent users, the stakes of cyborgization are sharpest. If the AI system is not an optional enhancement but the infrastructure through which equitable cognitive participation becomes possible, then the person-system composite is not a convenience but a civil rights configuration. Regulating, restricting, or withdrawing the technological component is not analogous to revoking a workplace perk. It is analogous to removing a wheelchair ramp — except that the ramp is owned by a corporation, priced as a subscription, and subject to unilateral modification. The costs of that arrangement are absorbed entirely by the user.

The Governance Problem

The deeper problem is governance, and here the prosthesis metaphor becomes uncomfortably literal. Prostheses require maintenance, calibration, and an ecosystem that decides what the device can do, how it behaves when it is uncertain, and how it fails. With AI, that ecosystem is not your body’s physiology or a local clinic. It is a platform stack: training data, compute infrastructure, moderation policy, pricing tiers, terms of service, enterprise procurement, and update schedules that can alter the device without warning. If the prosthesis becomes habitual — and certainly if the relation becomes cyborg — then governance becomes intimate. The interface sits inside your cognitive routine, but it is administered elsewhere.

This is where the political economy stops being an optional background and becomes constitutive. Shoshana Zuboff’s account of surveillance capitalism, whatever one thinks of its emphases, remains clarifying on this specific point: platforms extract value by positioning themselves as intermediaries and converting mediated activity into predictive assets (Zuboff 2019). The more the prosthesis becomes the default route for everyday thinking-work and making-work, the more that work is routed through systems whose incentives are not aligned with epistemic care. The intermediary recodes the relationship: what appears to be assistance functions as extraction.

Nick Couldry and Ulises Mejias sharpen the stakes by framing large-scale data extraction as a colonial relation — not as an analogy but as a structural claim about appropriation. Drawing on a long tradition of dependency theory and postcolonial critique, Couldry and Mejias argued that modern capitalism extends itself by annexing human experience and interaction as raw material for value (Couldry and Mejias 2019). Within that frame, “AI as prosthesis” names a double movement: the device helps you, and the device appropriates the traces of your helping yourself. When the historian builds an application through iterative conversation with an AI system, each exchange refines the platform’s training data while producing an artifact the historian needs. Dependence is not merely psychological. It is infrastructural.

Kate Crawford’s insistence that AI is not an abstraction but an extractive industry provides the material counterweight (Crawford 2021). Crawford demonstrated that the planetary costs of AI systems are distributed along familiar lines of geopolitical asymmetry: minerals are extracted from mines in the Global South, data annotation labor is performed by low-wage workers, and the environmental costs of computation are externalized onto communities with the least political leverage to resist them. The prosthesis has a supply chain. So does the cyborg.

For prosthetic technical competence, the governance problem carries an additional dimension. The non-specialist who builds through AI coupling depends not only on the platform’s continued availability and pricing but also on its continued capability. A model update that changes how the system handles code generation, a policy revision that restricts certain outputs, a corporate decision to deprecate a feature — any of these can degrade or destroy a workflow that the user has built their practice around. The prosthesis is not yours. It is leased, and the lease terms are written by the lessor.

Responsibility Without Redistribution

Prosthetic systems distribute agency. They rarely distribute liability. When AI produces an error — a fabricated citation, a flattened argument, an invented statistic, a broken function — the reputational and professional risk sticks to the human user. That is not an accident. It is a governance choice. Platforms insist the system is “assistive.” Organizations insist the human remains accountable. The gap between the two becomes the user’s burden: auditing, verifying, cross-checking, and disclaiming. Prosthetic cognition reduces some kinds of labor while creating new ones, especially supervision and repair. The work does not disappear. It is translated into a different register and shouldered by a different party.

The same asymmetry applies to prosthetic code. The scholar who builds an application through AI coupling bears full responsibility for its accuracy, its functionality, and its failures — while the platform that generated the code bears none. If the script misparses a dataset, the error is the researcher's responsibility. If the web application breaks after a library update, the repair falls to the builder. The prosthesis does not simply extend capacity. It disciplines the user into a managerial posture toward its own supports, converting the act of making into a continuous practice of inspection and maintenance. What presents as creative empowerment doubles as unpaid quality assurance.

Mitchell and Snyder’s concept of narrative prosthesis is worth returning to here. If “AI as prosthesis” becomes the dominant story — the comfortable metaphor that frames dependence as enhancement and structural capture as personal convenience — then the narrative itself does political work by foreclosing harder questions about who profits from the coupling and who bears its costs (Mitchell and Snyder 2000). Haraway would add that the political task is not to refuse the coupling but to insist on knowing its terms: to be a cyborg with a politics rather than a consumer with a subscription (Haraway 1985).

Five Tests

If the metaphor is to do analytical work rather than serve as atmosphere, it needs to be operationalized. Building on the frameworks outlined above — Clark and Chalmers on cognitive extension, Hamraie on normative access, Haraway on cyborg politics, Zuboff on platform capture, Couldry and Mejias on data colonialism, and Crawford on material extraction — I would propose five tests.

The first is habit. The decisive threshold is not whether you use AI but whether your workflow reorganizes around it — whether going without feels like losing capacity rather than choosing a different method. Clark’s “natural-born cyborg” argument suggests this threshold will arrive sooner and more quietly than most users expect (Clark 2003). Prostheses change history when they become ordinary. They become cyborg relations when their removal is no longer experienced as an inconvenience but as an amputation.

The second is governance. Who sets the parameters of the prosthesis, and how legible are those parameters to the user? What genres does the system compress into templates? What uncertainty signals does it provide or withhold? What does it refuse, and on whose authority? These are epistemic politics encoded as interface design — and opacity, as Zuboff’s framework reminds us, is not a bug but a revenue model (Zuboff 2019).

The third is norm formation. Once prosthetic cognition and prosthetic technical competence become common, which new expectations of speed, output, and capability become naturalized — and which people are punished for failing to meet them? The device that “helps” the historian build a digital archive today becomes the baseline expectation that renders the non-digital historian inadequate tomorrow. The prosthesis that gives the neurodivergent student equitable access today becomes the invisible infrastructure that institutions use to avoid redesigning their own assessment norms. Hamraie’s analysis of how “the normate” structures design regimes is directly applicable (Hamraie 2017). The risk is that prosthetic access substitutes for institutional transformation — praised as inclusion, when in practice, it is merely delegation.

The fourth is dependency and repair. What happens when access is revoked, when pricing changes, when institutional licenses expire, when models shift, when privacy regimes tighten or loosen? Dependence is not a feeling. It is a relation to an infrastructure, and it is most visible when the infrastructure withdraws. For users whose equitable cognitive participation depends on the prosthesis, withdrawal is not inconvenient. It is exclusion. Couldry and Mejias’s framework clarifies the structural dimension: what appears to be individual dependency is, at scale, an extractive relation between the platform and the population (Couldry and Mejias 2019). The subscription renews. The terms do not.

The fifth is responsibility. How is liability assigned when an agency is distributed? In practice, the default resolution is to make the human user responsible for the system’s failures — convenient for platforms, punitive for workers, scholars, and students. If AI is prosthetic, then the politics of audit, disclosure, and liability are central, not an afterthought. If the relation has become cyborg, then the question is not merely who is responsible but who gets to narrate the coupling and on whose terms, which is, as Haraway insisted, a question about power before it is a question about policy (Haraway 1985).

The Terrain

None of this requires us to treat AI as literally analogous to a prosthetic limb, nor to treat the language of cyborgization as a science-fiction prediction. Metaphors need not be anatomically faithful to be analytically sharp, and process concepts need not name a completed transformation to identify a trajectory. What “AI as prosthesis” does, at its best, is keep a contradiction in view: the system can genuinely extend capacity — including the capacity to build computational objects, and including the capacity of neurodivergent people to participate equitably in institutions that never redesigned their norms — while simultaneously tightening governance, intensifying expectations, and deepening dependence on infrastructure administered by others. What cyborgization adds is the recognition that, at some threshold of integration, the language of “using a tool” gives way to that of constitutive coupling, and the governance questions shift accordingly. That contradiction is not a temporary glitch on the road to a stable settlement. It is the terrain.

The question that matters most is not whether AI becomes more capable, but whether it becomes habitual, and for whom. The historian who now builds applications, the teacher who now constructs interactive archives, the scholar who now generates datasets from primary sources, the neurodivergent student whose analytical capacity finally finds institutional expression — each has gained something real. Each has also entered into a dependency relationship with a platform whose incentives, longevity, and governance are beyond their control. Prostheses are political when they become ordinary. They become cyborg relations when the ordinary becomes constitutive. And they are most political of all when they redistribute competence across boundaries that institutions have spent decades treating as natural.

References

Clark, Andy. 2003. Natural-Born Cyborgs: Minds, Technologies, and the Future of Human Intelligence. New York: Oxford University Press.

Clark, Andy. 2008. Supersizing the Mind: Embodiment, Action, and Cognitive Extension. New York: Oxford University Press.

Clark, Andy, and David J. Chalmers. 1998. “The Extended Mind.” Analysis 58, no. 1: 7–19.

Clynes, Manfred E., and Nathan S. Kline. 1960. “Cyborgs and Space.” Astronautics, September, 26–27, 74–76.

Couldry, Nick, and Ulises A. Mejias. 2019. The Costs of Connection: How Data Is Colonizing Human Life and Appropriating It for Capitalism. Stanford: Stanford University Press.

Crawford, Kate. 2021. Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. New Haven: Yale University Press.

Hamraie, Aimi. 2017. Building Access: Universal Design and the Politics of Disability. Minneapolis: University of Minnesota Press.

Haraway, Donna J. 1985. “A Manifesto for Cyborgs: Science, Technology, and Socialist Feminism in the 1980s.” Socialist Review, no. 80: 65–108.

Mitchell, David T., and Sharon L. Snyder. 2000. Narrative Prosthesis: Disability and the Dependencies of Discourse. Ann Arbor: University of Michigan Press.

Price, Margaret. 2011. Mad at School: Rhetorics of Mental Disability and Academic Life. Ann Arbor: University of Michigan Press.

Zuboff, Shoshana. 2019. The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power. New York: PublicAffairs.

The great computer science exodus (and where students are going instead)

Connie Loizos

12:40 AM PST · February 15, 2026

Something strange happened at University of California campuses this fall. For the first time since the dot-com crash, computer science enrollment dropped. System-wide, it fell 6% this year after declining 3% in 2024, according to reporting this past week by the San Francisco Chronicle. Even as overall college enrollment climbed 2% nationally — according to January data from the National Student Clearinghouse Research Center — students are bailing on traditional CS degrees.

The one exception is UC San Diego — the only UC campus that added a dedicated AI major this fall.

This all might look like a temporary blip tied to news about fewer CS grads finding work out of college. But it’s more likely an indicator of the future, one that China is much more enthusiastically embracing. As MIT Technology Review reported last July, Chinese universities have leaned hard into AI literacy, treating AI not as a threat but instead as essential infrastructure. Nearly 60% of Chinese students and faculty now use AI tools multiple times daily, and schools like Zhejiang University have made AI coursework mandatory, while top institutions like Tsinghua have created entirely new interdisciplinary AI colleges. In China, fluency with AI isn’t optional anymore; it’s table stakes.

U.S. universities are scrambling to catch up. Over the last two years, dozens have launched AI-specific programs. MIT’s “AI and decision-making” major is now the second-largest major on campus, says the school. As reported by the New York Times in December, the University of South Florida enrolled more than 3,000 students in a new AI and cybersecurity college during its fall semester. The University at Buffalo last summer launched a new “AI and Society” department that offers seven new, specialized undergraduate degree programs, and it received more than 200 applicants before it swung open its doors.

The transition hasn’t been smooth everywhere. When I spoke with UNC Chapel Hill Chancellor Lee Roberts in October, he described a spectrum — some faculty “leaning forward” with AI, others with “their heads in the sand.” Roberts, a former finance executive who arrived from outside academia, was pushing hard for AI integration despite faculty resistance. A week earlier, UNC had announced it would merge two schools to create an AI-focused entity — a decision that drew faculty pushback. Roberts had also appointed a vice provost specifically for AI. “No one’s going to say to students after they graduate, ‘Do the best job you can, but if you use AI, you’ll be in trouble,’” Roberts told me. “Yet we have faculty members effectively saying that right now.”

Parents are playing a role in this rocky transition, too. David Reynaldo, who runs the admissions consultancy College Zoom, told the Chronicle that parents who once pushed kids toward CS are now reflexively steering them toward other majors that seem more resistant to AI automation, including mechanical and electrical engineering.

But the enrollment numbers suggest students are voting with their feet. According to a survey in October by the nonprofit Computing Research Association — it members include computer science and computer engineering departments from a wide range of universities — 62% of respondents reported that their computing programs saw undergraduate enrollment declines this fall. But with AI programs ballooning, it’s looking less like a tech exodus and more like a migration. The University of Southern California is launching an AI degree this coming fall; so are Columbia University, Pace University, and New Mexico State University, among many others. Students aren’t abandoning tech; they’re choosing programs focused on AI instead.

It’s too soon to say whether this recalibration is permanent or a temporary panic. But it’s certainly a wake-up call for administrators who’ve spent years wrestling with how to handle AI in the classroom. The debate over whether to ban ChatGPT is ancient history at this point. The question now is whether American universities can move fast enough or whether they’ll keep arguing about what to do while students transfer to schools that already have answers.

The newsletter GenAI.works reports:

Spotify put a weirdly specific claim on the record during its Q4 2025 earnings call. Spotify executive Gustav Söderström said the company’s “best developers… have not written a single line of code since December,” because AI tools got good enough to build and ship work without traditional coding. The stake is its users’ trust and product control because a lack of oversight from human developers could build on ineffective features.

Spotify is selling a future where software becomes cheap and constant. That can feel great when it means fewer bugs and faster improvements. It can also mean more nonstop experiments, more synthetic content, and blurrier responsibility when something goes wrong. If engineers stop writing code, Spotify’s next trust test is whether users still feel like humans and who owns the outcome. The challenge lies in striking a balance between applying AI for efficiency while maintaining the human touch that users value.

© 2026 GenAI Works, Inc.

In January 2025, the Video Game History Foundation launched free public access to a digital archive of more than 1,500 fully digitised video game magazines spanning several decades of print, from long-running titles such as Electronic Gaming Monthly and Nintendo Power to short-lived publications such as Girl Gamer and Digital Diner (VGHF 2025a). All of it is full-text searchable. For historians of media, technology, and popular culture, this is not a curiosity. It is a primary source base.

The magazines were never designed to survive. Monthly publication cycles enforced disposability; each issue lost its commercial relevance the moment the next one shipped. When publishers folded or merged, back catalogues were shredded during asset transitions rather than transferred to libraries or archives. Developers and editors treated these objects as operational tools: vehicles for advertisements, previews, and reviews that served business functions rather than cultural ones. The result was the systematic destruction of a print record that documented how an entire medium was experienced, marketed, and understood in real time, its evidentiary value rendered invisible by the very disposability that defined the format.

The VGHF’s digital library encompasses more than 30,000 curated files: not only magazines but also development documents, artwork, press kits, promotional materials, and over 100 hours of production footage from the making of the Myst series by developer Cyan (VGHF 2025a). The Foundation built a custom optical character recognition toolset to handle the bold layouts, irregular typography, and dense colour schemes that characterise video game print media, a technical challenge that had previously made these sources resistant to full-text digitisation (Medley 2025). Phil Salvador, the Foundation’s Library Director, has described the platform as “a powerful tool for video game research with best-in-class discovery features,” powered by ArchivesSpace and Preservica (VGHF 2025a). The infrastructure is not makeshift. It is built to professional archival standards, and that distinction matters: what appears to be a fan project is in fact an institutional intervention, designed to convert scattered ephemera into a searchable evidentiary base.

Ephemera as Evidence

Playing a game tells you what the software does. It does not tell you how the game was promoted, who was presumed to be the player, or what cultural assumptions shaped the critical response. Magazine covers, advertisements, and editorial framing reveal the gendered and racialised marketing logic that surrounded games at the moment of their release: evidence visible only in the print record, not in the software itself. Frank Cifaldi, the Foundation’s founder and co-director, has framed the problem bluntly. The game is an artefact of design. The magazine is an artefact of reception. Both are necessary; neither alone is sufficient.

That distinction is familiar in media history. Scholars of film, radio, and television have long recognised that the institutional apparatus surrounding a cultural work: its marketing, its audience correspondence, its trade press coverage, constitutes a layer of evidence distinct from, and often more revealing than, the work itself. Yet the preservation infrastructure for video games has, until recently, treated the software as the primary object worth saving, while the print ecology that gave it meaning was left to decay. What was preserved was the product; what was lost was the regime of reception that made it legible.

Cifaldi has drawn explicit parallels with film. In a 2016 talk at the Game Developers Conference, he noted that an estimated 75 per cent of American silent films are now permanently lost, and expressed concern that the early history of video games was on a similar trajectory (Cifaldi 2016). The Film Foundation, Martin Scorsese’s preservation nonprofit, served as a direct model for the VGHF’s creation (Lowood 2022). The analogy is instructive but imperfect. Film scholars have access to studio correspondence, shooting scripts, and production records because the film industry eventually developed preservation norms. The games industry has not. As Cifaldi told Atlas Obscura in 2017, “We’re hitting the point where this industry is old enough that people are starting to pass. It’s time to take this stuff seriously” (Beck 2017). That urgency is not rhetorical. It is actuarial.

The scale of disappearance compounds the problem. A 2023 study conducted by the VGHF and the Software Preservation Network found that 87 per cent of classic video games released in the United States before 2010 are no longer commercially available: a rate of disappearance comparable to pre-World War II audio recordings and American silent films (Salvador 2023). Only 13 per cent of the games surveyed remained in print, with no five-year period exceeding 20 per cent availability. The Commodore 64 library, treated as an “abandoned ecosystem” in the study, registered at 4.5 percent. Even the PlayStation 2, a relatively recent platform with a massive installed base, managed only 12 per cent. As Salvador concluded, “The interests of the marketplace do not align with the needs of video game researchers” (Salvador 2023).

If the games themselves are vanishing at this rate, the print materials that documented their cultural life are disappearing faster still. Studios rarely preserved internal records. Game source code was routinely lost or overwritten. The developer Tim Cain has described being forced by his employer, Interplay Productions, to destroy all personal copies of the original Fallout source code upon leaving the studio, only to later learn that the company had lost the material. In this vacuum, third-party artefacts such as magazines are not supplementary to the historical record. They are the historical record, or what remains of it.

Historical Narratives in the Margins

The distinction between an object and the institutional frame that surrounded it is one that historians of nearly every field have learned to treat as analytically productive: how a source was classified, circulated, promoted, or ignored often reveals more about the governing assumptions of a period than the source itself. The pages of Edge, Mean Machines, and Next Generation preserve exactly this kind of evidence. Industry leaders guessed wrong. Fans expressed doubts about the PlayStation’s commercial prospects. Critics reversed opinions about games now treated as canonical. What survives in these magazines is not a polished narrative but the dialectic between prediction and reception: a record of how audiences engaged with a medium whose history has been heavily mythologised after the fact, with the contingency flattened out in retrospect.

John O’Shea, Creative Director of the National Videogame Museum in Sheffield and a historian who has worked extensively with game print materials, describes this as “resistance to linear ideas of history” (Stuart 2025). The characterisation is precise. Post hoc accounts of the games industry tend toward technological determinism: the assumption that consoles and genres succeeded because of intrinsic merit or technical superiority. The magazine archive disrupts that logic. It restores the mess of bad predictions, promotional spectacles, and letter-column debates about whether polygonal graphics were worth the trade-off in frame rate. Even reader correspondence: letters to the editor, fan art, cheat submissions, and documents a participatory culture that existed well outside corporate frameworks. What presents as nostalgia collecting is, read differently, the recovery of a public sphere that commercial retrospectives have largely written out.

As Laine Nooney has argued, rigorous scholarship contributing to what she calls “critical historical studies of video games” is long overdue (Nooney 2013). Nooney’s own work on Sierra On-Line drew extensively on non-software sources: corporate records, regional archives, and oral histories to reconstruct the labour conditions and economic structures behind game production rather than treating games as autonomous design objects (Nooney 2020). The VGHF archive makes a similar move possible on a much larger scale, granting researchers access to the contextual materials that have been absent from game historiography: trade press coverage, marketing language, editorial framing, and reader responses. These are the materials that convert a history of products into a history of culture.

The point extends beyond North America. In Japan, the video game magazine ecosystem was larger, more differentiated, and more deeply integrated into consumer culture than its Western counterpart. Famitsu (originally Famicom Tsūshin, ファミコン通信), first published in 1986, became the most widely read game publication in the country, moving to weekly publication in 1991 and reaching a circulation of 500,000 copies per issue. Alongside Famitsu ran dozens of platform-specific and genre-specific magazines: Dengeki PlayStation, Saturn Fan, Neo Geo Freak, The PlayStation, among others. The VGHF has begun accepting donations of Japanese-language publications, and a collection spanning 1992 to 1998 was recently accessioned (VGHF 2025b). Yet the vast majority of this material remains undigitised and inaccessible outside private collections. For historians of Japanese media industries and the transpacific circulation of cultural goods, the gap is significant. The story of how Japanese game franchises were received, adapted, and marketed in Western markets, and how Western titles were repackaged for Japanese consumers, is largely told through these magazines, and it remains only partially recoverable.

Fan Preservation and Institutional Reclamation

That any of this material survives at all is largely due to collectors and volunteer scan communities rather than to any institutional foresight. Groups such as Retromags and the Out-of-Print Archive spent years salvaging and digitising magazines without institutional backing, absorbing the costs of preservation labour that neither publishers nor libraries were willing to shoulder. The VGHF has since partnered with these groups, integrating what was essentially amateur labour into formal archival infrastructure. Hundreds of magazine scans in the digital library were contributed by community members, and the Foundation has described its platform as “a force multiplier for citizen archivists who have been preserving the history of games” (VGHF 2025a). The phrase is revealing: it converts unpaid labour into a valorised input, folding informal preservation into institutional legitimacy without fully acknowledging the asymmetry between the two.

The Foundation’s broader approach treats game history holistically, encompassing not only software and magazines but also developer notes, promotional flyers, strategy guides, VHS recordings, retailer catalogues, and press kits. One of the library’s flagship collections, the Mark Flitman papers, comprises over two decades of documentation from a retired game producer who worked at Konami, Acclaim, Midway, and Mindscape during the 1990s and 2000s. Flitman allowed the VGHF to digitise the paperwork and digital file backups he had kept in his basement, materials that Salvador has described as “an incredible record of the business of video game production and marketing” (VGHF 2025a). In 2019, the Foundation undertook a five-week archival mission at the offices of Game Informer magazine, digitising over 25 terabytes of historical material: press kits, promotional CD-ROMs, and corporate ephemera dating to the magazine’s founding in 1991 (Reiner 2019). As Cifaldi has observed, “the greatest discoveries we’re going to find are on materials that people don’t know are important” (Lowood 2022). That formulation captures both the promise and the precarity of the enterprise: the most valuable sources are precisely those most likely to have been discarded.

Curatorial practice is shifting in parallel. The National Videogame Museum in Sheffield, which holds one of the UK’s largest collections of nearly 5,000 videogame objects, now accepts magazine collections as records of individual gaming journeys, preserving both the objects and the subjective meaning donors attach to them. O’Shea has described the museum’s mission as one of collecting, preserving, exhibiting, and interpreting video games “for everyone. Forever” (NVM 2025). What was household clutter a decade ago is now treated as cultural evidence. Scholars have already begun drawing on game magazines to reconstruct the gendered and racialised assumptions embedded in early 1990s advertising, to trace how Japanese media franchises diffused across Western markets, and to document the emergence of game journalism as a critical genre in its own right, the contours of a new field rendered visible in materials that no one thought worth saving.

The legal environment, however, remains hostile to the logic of open access. In October 2024, the US Copyright Office declined to grant a new exemption under the Digital Millennium Copyright Act that would have allowed libraries and archives to share digital copies of out-of-print games remotely with researchers. The Entertainment Software Association had argued that the industry’s own commercial efforts were sufficient to preserve its history: a claim directly contradicted by the VGHF’s 2023 study showing 87 per cent commercial unavailability (VGHF 2023). As things stand, libraries can digitally preserve but not digitally share video games, and can provide on-premises access only. Books, films, and audio recordings enjoy broader exemptions. The asymmetry is structural rather than incidental: the costs of restricted access are borne by scholars and public institutions, while the rights holders who lobbied for restriction bear none.

Towards a Print-Culture History of Gaming

None of this is unprecedented. Ephemera: trade cards, pamphlets, broadsides, advertising circulars, have been reclaimed before by historians who recognised that the disposable objects of one era become the documentary evidence of the next. Video game magazines fit this pattern precisely. Their temporality, what they highlighted, what they ignored, how they speculated, is what makes them valuable now. They offer something the games alone cannot: a window into the cultural and commercial world that produced, consumed, and discarded them.

The VGHF’s library has already been used in commercial video game projects, including Mortal Kombat: Legacy Kollection, GEX Trilogy, Bubsy in: The Purrfect Collection, and Atari 50: The Namco Legendary Pack (VGHF 2025c). It has supported the digital archives of Electronic Gaming Monthly and Game Informer, and its materials have appeared in museum exhibitions, including the FMV Zone exhibit at the Portland Retro Gaming Expo. The archive is being used not only by historians but by an industry that is beginning to recognise, belatedly, that its own past requires the kind of contextual documentation that magazines provide. The irony is quiet rather than dramatic: the same industry that treated these publications as disposable now turns to them as a resource, having destroyed the internal records that might have served the same purpose.

The archive is open. The question now is what historians will build from it.

References

Beck, Julie. 2017. “The Holy Grail of Video Game History Is Probably an Office Memo.” Atlas Obscura, 30 August 2017. https://www.atlasobscura.com/articles/video-game-history-foundation-grail-preservation-ephemera.

Cifaldi, Frank. 2016. “’It’s Just Emulation!’ — The Challenge of Selling Old Games.” Talk presented at Game Developers Conference, San Francisco.

Lowood, Henry. 2022. “Frank Cifaldi: A Conversation on Game Preservation and Documentation.” ROMchip: Journal of Game Histories 4 (2). https://romchip.org/index.php/romchip-journal/article/view/171.

Medley, Sam. 2025. “Video Game History Foundation Launches Digital Library of Old Video Game Magazines, Development Assets, and More.” NotebookCheck, 31 January 2025.

Nooney, Laine. 2013. “A Pedestal, A Table, A Love Letter: Archaeologies of Gender in Videogame History.” Game Studies 13 (2). https://gamestudies.org/1302/articles/nooney.

Nooney, Laine. 2020. “The Uncredited: Work, Women, and the Making of the U.S. Computer Game Industry.” Feminist Media Histories 6 (1): 119–146. https://doi.org/10.1525/fmh.2020.6.1.119.

NVM (National Videogame Museum). 2025. National Videogame Museum, Sheffield. https://thenvm.org/.

Reiner, Andrew. 2019. “Keeping History Alive.” Game Informer, 5 December 2019. https://gameinformer.com/2019/12/05/video-game-history-foundation-keeping-history-alive.

Salvador, Phil. 2023. Survey of the Video Game Reissue Market in the United States. Video Game History Foundation and Software Preservation Network. Zenodo. https://doi.org/10.5281/zenodo.7996492.

Stuart, Keith. 2025. “Video Game Magazines Were Disposable—Now They’re a Vital Archive of Gaming Culture.” The Guardian, 28 January 2025.

Video Game History Foundation (VGHF). 2023. “87% Missing: The Disappearance of Classic Video Games.” https://gamehistory.org/87percent/.

Video Game History Foundation (VGHF). 2025a. “The VGHF Library Opens in Early Access.” 30 January 2025. https://gamehistory.org/vghf-library-launch/.

Video Game History Foundation (VGHF). 2025b. “Japanese Game Magazine Donation.” VGHF Library Catalogue. https://library.gamehistory.org/repositories/2/accessions/50.

Video Game History Foundation (VGHF). 2025c. “Our Impact in 2025.” 2 December 2025. https://gamehistory.org/our-impact-in-2025/.

Christopher Harding

The nature of the Japanese language presents fascinating challenges to the development of artificial intelligence.

A favourite tourist board cliché about Japan has long been that it’s the place where tradition and refinement meets high technology. Visitors taking the so-called Golden Route in Japan fly into Tokyo and are transported from a hyper-modern megalopolis via futuristic-feeling shinkansen trains to the ancient shrines and temples of Kyoto.

But the shinkansen started out as 1960s technology, as for the most part did Japan’s ubiquitous vending machines. Yes, there is something about their design, efficiency and place within a larger omotenashi culture – a premium placed on putting customers at ease – that contributes to an enduring sense of Japan being ahead of most of the rest of the world. But in terms of cutting-edge technology, Japan isn’t quite the all-conquering force that it once was. Might that situation be at last about to change?

Much of the technology for which Japan became well known in the second half of the 20th century was western in its origin but Japanese in its perfection. That included automobile manufacture and electronics, thanks in no small part to licensing agreements that allowed Japanese firms to get hold of and adapt inventions like the transistor. They refined the concepts and revolutionised production methods, churning out precision hardware on an extraordinary scale and with previously unheard-of levels of quality control.

By the late 1980s, Japanese companies controlled over half of the global chip market. Electronics giants such as Sony and Panasonic, alongside automotive firms like Toyota, were household names. The world was not unvaryingly happy about these developments. Japan, like Germany, had to put up with ‘Who won the war anyway?’ resentment in some quarters at their return to prosperity and competitiveness. The damage done to American car manufacturing yielded such anger that there was a time when in parts of Detroit you could pay a dollar to pick up a baseball bat and help smash up a Japanese car.

Around the time that Japan’s economic bubble burst at the turn of the 1990s, its fortunes as a global technology leader went into decline. There were all sorts of reasons for this. Japan gave in to heavy political pressure from the United States to go easy on American firms, in various areas of trade including semiconductors – whose production in any case began to shift from the same companies designing and manufacturing to the outsourcing of the latter, often to emerging Asian economics. Taiwan began its rise to semiconductor supremacy around this time.

More broadly, technological innovation began to be less about the precision manufacturing in which Japan excelled and more about software. It was soon the turn of American companies, like Microsoft, Apple, Google and Paypal, to become household names. While firms like these were revolutionising the how-to of innovation in the 1990s and 2000s – embracing venture capital, platforms over products and the use of ‘fail fast’ startups as a form of R&D – Japan became notable for its continued reliance on fax machines and fistfuls of cash.

Tempting though it is to understand Japan’s relative stagnation in the world of tech purely in terms of a conservative corporate culture – consensus-driven and perfectionist – bumping up against ‘move fast and break things’ disruptors in the US, the situation is at once more complex and more interesting. This is especially true of the technology of the moment, artificial intelligence, and the slightly older technologies on which it rests.

Thirty years ago, internet uptake in Japan was frustrated by bureaucratic conservatism and government inertia, alongside the dominance of telecommunications by Nippon Telephone and Telegraph. NTT made accessing the early internet expensive by charging users for each local call rather than offering a flat monthly fee as in the US. But the Kobe earthquake of 1995 helped clarify the case for the internet, showcasing its potential in sharing time-critical information. Within a few years, Japan led the world in mobile internet technology.

You really felt this, visiting or living in Japan during the mid-2000s. I vividly recall finding the idea that you could switch between sending a text message and sending an email – from your phone! – nothing short of revolutionary. You could also take grainy photographs – from your phone! From micro-payments to micro-blogging, a population always looking for things to fill its famously long work commutes – up to two hours one way – was again wowing the world with its innovation and embrace of high technology.

Even then, there were signs of the difficulty that Japan would face, the more that technology and trade became truly global. Some of the earliest desktop web browsers struggled to handle Japanese characters, while English quickly became the lingua franca of the worldwide web. The knock-on effect, 30 years later, has been that when it comes to training large-language models (LLMs) such as ChatGPT, companies based in the English-speaking world have enjoyed an enormous advantage. The sheer volume and range of training data available, from science to the arts, is such that Japan simply cannot compete. Around 55 per cent of web content is in English. Japanese content accounts for between two and three per cent.

The nature of the Japanese language itself presents fascinating challenges to the development of artificial intelligence. Japanese uses three scripts: kanji (based on Chinese ideographs), a simplified syllabary called hiragana and a second syllabary called katakana, often used for foreign loanwords. Many words in Japan can be written in multiple combinations of these scripts, and they usually appear in streams of text where words are not separated – it’s up to the reader to disaggregate them, causing yet more problems for machine learning. Add to this the importance of honorifics and varied levels of politeness in everyday speech, and you have quite the challenge for training LLMs.

On top of all this, one of the beauties – and frustrations – of the Japanese language as used in real life is that it’s highly dependent on context. People rarely speak in what users of English might consider full sentences. Instead, a few well-chosen words deployed at the right moment do a great deal of heavy lifting. The success of this depends on people being able to intuit the context – something with which AI, alongside non-native speakers, often struggles.

LLMs are mathematical at heart. They transform words, parts of speech and punctuation marks into ‘tokens’ and then track both logical references within sentences and repeated combinations found in the training data with which AI engineers supply them. Feed a machine ‘The cat sat on the mat’ enough times and pretty soon it can complete the sentence if given just the first five words, based on probability theory among other mathematical innovations.

Japanese is structurally more regular than English, as many a frustrated Japanese learner of English will be happy to tell you. But compared with English overall, there is a smaller role for logical reference and a larger one for situational inference.

Take the sentence: Mō owatta kara, hon o ageta. It literally means: ‘Because already finished, gave book.’ What’s finished? Who gave who a book? If you’re in the room, whether physically or virtually as part of a story, you likely won’t struggle with these questions. But pity the machine that has to make the connections and work them into the picture that it’s forming of how the Japanese language works.

Social and cultural contexts add challenges of their own. To non-native speakers of Japanese, using the language in real life is a little like driving through a busy city having only studied the theory behind how a car works alongside a smattering of the highway code. You must very quickly try to understand why a range of different drivers do what they do in different situations and what is expected of you in response. It all takes place at high speed and the stakes can often be high.

It’s not that you can’t train an LLM to learn these contexts and make these inferences, but it’s more difficult and requires more data, which takes you back to the problem of the comparative lack of Japanese online.

Chinese AI companies have enjoyed advantages here that Japan doesn’t have – and some of which most in Japan would not want. They include lavish state support, with political strings attached, for both AI and the physical infrastructure (such as data centres) on which it relies. China’s highly-controlled internet environment meanwhile reduces the chances of AI models producing unwanted output. Still, both Chinese and Japanese AI engineers have to grapple with the difficulty of working with ideographs and context.

Some of Japan’s LLMs begin with existing models that have been pre-trained on English-language data, since much of this doesn’t change around the world: basic facts, information about global organisations, patterns of general reasoning and even safety filters. They then continue the pre-training process using Japanese data. This helps to fix problems with context and idioms. Finally, these LLMs can be fed more specialised Japanese data relevant to their intended applications in finance, medicine and elsewhere.

Even so, no Japanese LLM yet comes close to challenging the dominance in Japan of ChatGPT, which since GPT-3 has been trained multilingually and has become better at parsing non-English languages with every iteration. Ongoing feedback from Japanese users together with various technical innovations help it to adjust to Japanese conversational norms: avoiding over-directness, culturally unintelligible or inappropriate humour, and mistakes with ever-changing slang.

Not only do we not yet have a breakthrough Japanese LLM. The uptake of AI generally in Japan remains comparatively slow. Fewer than 50 per cent of Japanese companies had plans to incorporate generative AI in their workflows in 2024, compared with more than 80 per cent of companies in China and the US. And fewer than 50 per cent of Japanese university students used generative AI that year, compared with 66 per cent of British undergraduates (the number in Britain has subsequently risen to more than 90 per cent).

Japan meanwhile ranks close to the bottom of surveys tracking consumer knowledge about and trust in artificial intelligence – a striking finding given the old saw that Japanese people are more likely than their western counterparts to trust other high-tech innovations, including humanoid robots. This survey evidence is all the more remarkable because qualitative research comparing Japanese and American participants found that in a range of scenarios Japanese were actually more willing to work with AI agents than their US counterparts.

What’s going on here?

A claim often made is that Japanese culture, shaped by Shinto and Buddhism, is more at home than cultures underpinned by Abrahamic religions with the idea that non-human entities in the world possess spirit and even agency. This means that in Japan the so-called ‘uncanny valley problem’ – where trust in human-like robots declines the more human they seem – doesn’t pertain to the extent that it does in countries like the US. It also means, some claim, that people in Japan are more likely to attribute a kind of consciousness or social potential to AI that means they are willing to cooperate with rather than exploit or fear it.

While this may be true – although when it comes to Japan there’s a tendency to exaggerate cultural arguments – the real problem seems to be more prosaic: people in Japan don’t yet trust institutions to use artificial intelligence responsibly or usefully. Private companies that have spent decades building trust with consumers have a slight advantage over government here. Where the rollout of digital ID in Japan has been frustrated by instances of fraud and general privacy concerns, established brands such as Rakuten have found consumers willing to take a bet on some of their own innovations. These include autonomous mobile robots (AMRs), trialled in Tokyo earlier this year for delivering goods.

Attitudes to AI in Japan may well change the more that it becomes a component in obviously benign innovations like ‘co-bots’: robots designed to assist human beings in the pursuit of clear and positive aims, from productivity in industry to caring for the elderly and the infirm. There are also a range of things that a forward-looking Japanese government might do to deliver on long-standing promises of turning Japan into one of the leading digital societies in the world.

First of all: money. Commentators are calling for a decisive shift from ploughing government funds mainly into universities, established research institutions and large corporations towards favouring small startups. Successful AI startups already exist, including Sakana.ai and Preferred Networks Inc. The latter has taken off by combining deep learning with the ‘internet of things’ and by collaborating with major firms including Toyota. Still, as of now, aspiring AI startups in Japan are being advised to seek capital internationally rather than primarily at home.

Second, Japanese companies could be supported in taking part in a trend towards Asia-Pacific AI startups developing products with cross-border implementation in mind right from the beginning – encompassing languages like Japanese, Chinese, Korean and Thai.

Third, some business investors are calling for the Japanese government to offer more clarity about the future direction of regulation in this sector. That will make it easier for significant investment decisions to be made. Much depends, as it does elsewhere in the world, on how public sentiment about AI and privacy shifts over the next few years.

Finally, AI could be integrated more deeply into broader Japanese plans for a highly automated society. Public and political pressure to achieve this is growing by the year. Two of Japan’s most pressing problems are a rapidly declining population – shrinking by some 2,000 people every day – and persistently low rates of productivity.

At the same time, the prospect of large-scale migration into Japan remains unpopular and is a solution that no Japanese government has been willing to champion to any great extent. Politicians have preferred to create opportunities for foreign workers in targeted and largely unadvertised ways, to avoid the likely backlash. The situation doesn’t look like changing anytime soon, given the rise of populist parties on the right of Japanese politics in recent years. These are more openly critical than their predecessors of the level of migration in Japan and the impact that they claim it is having.

Against this backdrop, various innovations are already being trialled to prepare for a day when Japan is a smaller but smarter society. Some are relatively prosaic. Increasing numbers of vending machines are equipped with enough deep learning to suggest products based on previous purchases (accessed by scanning a customer’s face) or on the weather that day. A small number even track footfall around the machine and adjust product prices accordingly – particularly with perishable items like salads. Other innovations are more striking, such as the trialling of unmanned convenience stores. These use ceiling-mounted cameras to track customers and products, requiring people only to tap a payment card as they leave the shop.

One experiment to watch has just opened its gates this autumn. Toyota’s ‘Woven City,’ located at the foot of Mount Fuji, is a small community where a few hundred people (rising to around 2,000) will live alongside the latest tech innovations and the entrepreneurs and startups who create them. The point is to allow innovations to be tested in real time, from new kinds of robotics to autonomous vehicles.

Politicians in Japan have a track record of promising a great deal on digital innovation while being slow to deliver. But as the potential benefits to Japanese society of AI and associated technologies become clearer, there is every chance that this once-great tech giant – which still punches above its weight in many areas, including engineering and the natural sciences – may make a long hoped-for return to form.

Author

Christopher Harding

More about Christopher Harding

This reaction essay scrutinises the spatial and infrastructural topographies of East Asia’s digital‑game industries by way of a comparative investigation of Japan and South Korea. It explores Martin Roth’s notion of “multilayered spatialization” and places it in juxtaposition with the divergent industrial trajectories of Japan and South Korea. Methodologically the study is comparative and infrastructural, blending metadata‑informed cultural analysis with the regional policy frameworks to evaluate how platform constraints, localisation regimes and state intervention configure game production and circulation. The comparison reveals distinct strategies of global engagement and regulatory design, thereby problematizing simplistic binaries of global versus local in East Asian cultural industries.

Japan: Metadata and the Persistence of Regional Logics

Roth’s central argument dismantles the assumption that Japan’s console‑game industry is either a wholly self‑contained national domain or seamlessly subsumed within global entertainment markets. Instead, he shows that Japanese games circulate within overlapping circuits of domestic production and transnational distribution: this dynamic—termed by Roth “multilayered spatialization,” which involves the convergence of infrastructural constraints (such as platform region‑locks), economic strategies (publisher partnerships) and linguistic localisation regimes. (Roth 2025)

Analysing metadata drawn from repositories such as MobyGames and MADB, Roth identifies a persistent regional segmentation, particularly in the mid‑tier console and handheld markets. Even as certain flagship titles (for instance, Dark Souls and Elden Ring) adopt global release strategies, much of Japan’s output remains anchored domestically, reinforcing what Roth describes as spatial instability in the uneven, fragmented and contingent geographies of cultural distribution (Gerteis 2025).

A recurring case study is FromSoftware, whose evolution from Japan‑first releases (e.g., King’s Field) to simultaneous global launches (such as Elden Ring) epitomises infrastructural transition. Yet, even here, Roth cautions against assuming global uniformity: FromSoftware retains core production in Japan and continues to craft region‑specific content, reflecting a dual orientation that blurs the conventional national–global dichotomy (Roth 2025).

Roth observes how Japan’s industry exhibits a tension between outward‑looking global release strategies and inward‑oriented regional logics that persist via infrastructural, economic and linguistic mediations.

South Korea: Platform Governance and State‑Led Acceleration

By contrast, South Korea’s gaming sector has emerged via a distinct constellation of infrastructural and policy mechanisms. From the late 1990s the Korean government invested proactively in building a digital‑content ecosystem: supporting broadband expansion, establishing rating systems and funding export‑oriented content development (Chung 2015). These interventions formed a critical backdrop for the rise of online gaming, e‑sports and mobile‑first development patterns.

Unlike Japan’s console‑centred industry, Korea’s domestic market is dominated by PC and mobile games, facilitated by ubiquitous access to PC bangs (gaming cafés) and state‑backed initiatives promoting content export. Major firms such as NCSoft and Nexon have thrived in this environment, embracing live‑service models, international franchising and platform analytics as core design logics. Here, spatialisation manifests less through region‑locks and delayed localisations and more through platform governance and accelerated global release cycles (Na, Kim & Kim 2022).

In this schema, Korea’s industry exhibits what might be termed “infrastructural convergence”: production is tightly coupled with export‑readiness, supported by robust localisation pipelines and monetisation strategies tailored for global audiences. Yet this convergence is not without its frictions: as Na, Kim & Kim (2022) demonstrate in their COVID‑era analysis, the industry remains vulnerable to systemic shocks, thereby highlighting structural dependencies on platform infrastructures, labour models and global distribution channels.

Thus, Korea’s model emphasises a different modality of spatial logic that is less about fragmentary regional segmentation and more about integrated global‑facing production, albeit one that remains shaped by infrastructural and policy contingencies.

Shared Regional Challenges: Localisation and Cultural Mediation

Although the national infrastructures of Japan and South Korea diverge, both industries grapple with the centrality of localisation as a spatial and functional process. Roth treats localisation not merely as translation but as a mechanism of cultural and infrastructural adaptation whereby access to global markets is mediated.

Decisions about which titles are localised, when and how. They are inseparable from platform affordances, legal regulation and publisher strategic priorities (Roth 2025).

In the South Korean case, localisation likewise functions as a strategic tool: the lifting of Japan’s game‑import ban in 2004 compelled Korean firms to recalibrate competitive strategies and simultaneously invest in global‑ready IPs capable of bypassing domestic saturation (Chung 2015). Thus, localisation operates as a threshold mechanism structuring not only access but the assignment of value in transnational circuits.

In both contexts we observe that global circulation is not automatic but is mediated by infrastructures that encode spatial hierarchies: games do not simply “go global”; rather, they are routed, delayed, adapted or foreclosed through technical, legal and cultural filters. These constraints become especially visible in mid‑tier or niche releases, which often remain region‑locked or poorly localised, thereby reinforcing the uneven geography of global gaming.

Toward a Comparative Infrastructural Method

What emerges from this comparative vantage is the value of infrastructural and metadata‑driven methods for understanding East Asian media industries. Roth’s application of metadata analytics to trace patterns of production, platform and localisation offers a model for treating games as both cultural texts and industrial artefacts. In contrast, South Korean scholarship and industry analysis foreground policy instruments, broadband infrastructure and platform design as the key vectors of transformation (Chung 2015; Na, Kim & Kim 2022).

These perspectives are not mutually exclusive. Their convergence signals a methodological imperative to treat media industries as assemblages of code, labour, policy and infrastructure. Whether via the absence of platform compatibility in Japan or the ubiquity of global‑facing design in Korea, both systems reveal the contingencies of cultural production in the digital era.

Thus, a comparative infrastructural method can illuminate how differing national legacies, technological architectures and policy regimes shape cultural circuits—without recourse to teleological narratives of globalisation.

Concluding Thoughts

Japan and South Korea present two structurally distinct yet interlinked models of videogame production and circulation. Roth’s spatialisation framework enables us to see Japan’s industry as defined by infrastructural frictions and selective globalisation, while Korea’s model emphasises state‑facilitated convergence and export logic. Both cases challenge teleological schemas of globalisation and underscore the necessity of comparative, infrastructural methodologies in media studies. As digital media continue to traverse borders, the circuits through which they move, and stall, remain central to understanding East Asia’s position in the global cultural economy.

References

Chung, Pei-chi. 2015. “South Korea.” In Video Games Around the World, edited by Mark Wolf, 333–351. Cambridge, MA: MIT Press. https://doi.org/10.7551/mitpress/9658.003.0034

Gerteis, Christopher. 2025. Unboxing Japanese Videogames: A Metadata-Based Approach to the Production and Distribution of Spatial Instability By Martin Roth. Cambridge, MA: MIT Press, 2025. 192 pages. Ebook, open access, ISBN: 9780262382892. Paperback, ISBN: 978-0262552226. International Journal of Asian Studies. Published online 2025:1-3. https://doi.org/10.1017/S1479591425000026

Na, Jae-Hyun, Kim, Eun-Jung, and Kim, Joon. 2022. “Data Analysis of the Impact of COVID-19 on Digital Game Industrial Sustainability in South Korea.” PLOS ONE 17 (12): e0278467. https://doi.org/10.1371/journal.pone.0278467.

Roth, Martin. 2025. Unboxing Japanese Videogames: A Metadata-Based Approach to the Production and Distribution of Spatial Instability. Cambridge, MA: MIT Press. https://doi.org/10.7551/mitpress/15321.001.0001

Anime is booming and generating record revenues year on year — but it's short on skilled animators. In response, a textbook that teaches the fundamentals of Japanese animation was released in English last month. Aimed at aspiring animators or those curious about how anime is made, the text was produced by the Nippon Anime & Film Culture Association (NAFCA), an organization founded in 2023 to improve working conditions in the Japanese animation industry, which faces significant labor shortages.

As the title implies, the "Animator Skill Test Textbook” was written in concert with a test created and administered by NAFCA since 2024. The test, which currently has two levels of difficulty, measures aptitude in the fundamental techniques for drawing dōga, the individual frames combined to create a moving image.

Continue reading HERE.

Translating video games can lead to surprising linguistic innovation. When Jeremy Blaustein had to translate “現地調達” (genchi chōtatsu, local acquisitions) for Metal Gear Solid, he coined the term OSP or “on-site procurement,” a term that went on to be used by other games and continues to be used today.

Video game translation can also be politicaland controversial. In Final Fantasy X, for example, ありがとう (arigatō, thank you) was rendered in English as “I love you,” due to the special weight of the common phrase in context. While some fans hailed it as a brilliant translation, others felt the localization process had gone too far.

Other game translations have had cultural impacts for very different reasons: The now legendary meme “All your base are belong to us” stems from a simple mistranslation of a grammatically ordinary sentence: 君たちの基地は、全てCATSがいただいた (kimitachi no kichi wa, subete CATS ga itadaita, CATS [the alien cyborg villain] has obtained all of your bases).

ローカライゼーション (Rōkaraizēshon, Localization), which should be differentiated from more general 翻訳 (hon’yaku, translation), is a fraught and complex process, though its messiness is a learning opportunity for gamers and Japanese learners alike.

To understand the language of localization, you’ll need to be able to read 片仮名 (katakana) — or should we say, カタカナ. This more angular Japanese syllabic script is used for 外来語 (gairaigo), foreign loanwords imported from other languages, and is the vocabulary that largely makes up the world of games. This starts with the word for video games ビデオゲーム (bideo gēmu), but applies to most gaming terminology: ダイアログ (daiarogu, dialogue), ステータス (sutētasu, stats/status), スキル (sukiru, skills), レベル (reberu, level) and even ノンプレイアブル・キャラクター (nonpureiaburu kyarakutā, non-playable character/NPC). Since katakana is less common, it can sometimes trip up and slow down beginner Japanese readers. It takes time and patience to be able to smoothly process these pointy imps, but be well aware that they’ll likely take up most of your gaming discourse.

Of course, gaming vocabulary still has its fair share of 漢字 (kanji, Chinese characters), which can be seen in such common terms as 経験値 (keikenchi, experience points), 必殺技 (hissatsuwaza, killer/lethal move) and 状態異常 (jyōtai ijyō, status condition/abnormal status).

The process of game localization is a long one. Once a developer knows what languages they want a game translated into, they will often begin with a ローカライゼーションキット (rōkaraizēshon kitto, localization kit), which provides a guide to the video game including the content, brand voice and a 用語解説 (yōgokaisetsu, glossary of terms). ゲームをローカライズするために、翻訳家には背景が必要です (Gēmu o rōkaraizu suru tame ni, hon’yakuka niwa haikei ga hitsuyō desu, In order to localize a game, translators will require background).

Translation involves simply changing the 起点言語 (kiten gengo, source language) into the 目標言語 (mokuhyō gengo, target language). Localization, on the other hand, means adapting the product to the local culture and market. One common approach to localization is known as トランスクリエーション (toransukuriēshon, transcreation), which aims to adapt content from one language to another while maintaining tone, intent and style. 一貫性 (Ikkansei, Consistency), 口調 (kuchō, tone) and 文化感受性 (bunka kanjusei, cultural sensitivity) are all important.

一貫性のために、ローカライゼーションには用語解説が必要です (Ikkansei no tame ni, rōkaraizēshon yōgokaisetsu ga hitsuyō desu, For the sake of consistency, the localization will need a glossary of terms).

原作の口調と会うために、起点言語の理解と目標言語の雄弁が必要です (Gensaku no kuchō to au tame ni, kiten gengo no rikai to mokuhyō gengo no yuben ga hitsuyō desu, For the sake of matching the tone of the original work, [the localization will] need an understanding of the source language and fluency in the target language).

それから、文化感受性のために、ゲームのコンテンツも調整する必要があるかもしれません (Sorekara, bunka kanjusei no tame ni, gēmu no kontentsu mo chōsei suru hitsuyō ga aru kamo shiremasen, After that, for the sake of cultural sensitivity, [the localization] may need to adjust the contents of the game as well).

The world of game localization from Japanese to English hasn’t remained static. There is more translation talent and market interest in the exchange of games in both directions. Thanks to greater familiarity with Japanese culture in the West, Japanese words like ラーメン (rāmen), 居酒屋 (izakaya) and オタク (otaku), which used to require more loose translations like “noodles,” “pub” and “geek,” can now be left as they are. On the flip side, Japanese video game companies have also become more open to adapting to cultural sensitivities abroad, for example, by changing characters’ outfits.

Localizers have to grapple with challenging issues of tone and nuance. When a game character says, 私は元気！ (Watashi wa genki!, I am lively/well!), they have to look well beyond the literal words to the effect of the line: how it develops the personality of the speaker, and how it contributes to the story and a player’s enjoyment of the game. This more holistic approach could result in localizations ranging from “I’m on top of the world!” to “At least I’m alive,” factoring in 口調, 文脈 (bunmyaku, context) and 皮肉 (hiniku, irony).

Many localizers see this creativity as part of the joy of the job. In an interview for the J-EN Translations blog, game localizer Andre Echeverria recalls discovering that a character described as 陰キャ (inkya, a gloomy or loner character) perfectly fit the English world’s definition of “emo.” On a one-to-one basis, you could never translate 陰キャ as emo, but with the broader context in mind, the phrase worked perfectly. This allowed Echeverria to insert references to emo bands as fun イースター・エッグ (īsutā eggu, Easter eggs) that didn’t result in changes to the meaning, intent or atmosphere of the Japanese original.

Game localization works on the opposite of a simple one-to-one basis. 元気 doesn’t always mean “lively,” and ありがとう can very well mean “I love you.” To make matters even more complicated, what consumer audiences want isn’t necessarily a 雄弁な (yūbenna, eloquent) translation from Japanese to English. In a world where Japanese pop culture has captivated hearts and minds worldwide, some fans prefer a translation that seems 日本っぽい (Nihon-ppoi, Japan-ish). 突然、一貫性と口調と文化感受性は完全に違うものになってしまいます (Totsuzen, ikkansei to kuchō to bunkakanjyusei wa kanzen ni chigau mono ni natte shimaimasu, All of a sudden, consistency, tone and cultural sensitivity become totally different things).

Only the future can answer the question of where game localization goes. 雄弁な翻訳がいいか、日本っぽい翻訳がいいか、次のローカライゼーション世代が決めることだ (Yūbenna honyaku ga ii ka, Nihon-ppoi honyaku ga ii ka, tsugi no rōkaraizēshon sedai ga kimeru koto da, Whether fluent translations are better or Japan-esque translations are better will be something that the next generation of localization decides).

KEYWORDS

NIHONGO, VOCABULARY, VIDEO GAMES, TRANSLATION

Screenshot 1: The Nobunaga’s Ambition franchise was launched in the early 1980s and has since seen dozens of installments across successive generations of consoles and personal computers, evolving from rudimentary strategy grids to complex simulations that mirrored both advances in hardware and shifting popular imaginations of Japan’s Warring States past.

Nobunaga’s Ambition: Awakening welcomes players with blaring war drums, trumpets, and violins, an audio fanfare that immediately declares its dramatic historical setting. The interface is richly detailed yet accessible: menus and text can be set to English, while spoken dialogue retains original Japanese voicing for authenticity. A repository-like officer gallery allows the player to browse hundreds of Sengoku-era figures, each with their birth and death dates, statistics, and even a radial chart of their talents.

Screenshot 2: The Officer Gallery in Nobunaga’s Ambition: Awakening allows players to scroll through hundreds of historical figures. Each entry includes birth/death dates, stats, traits, and portraits, making it feel like an interactive Sengoku encyclopedia.

Gameplay feels like interacting with a living database of Japan’s Warring States; for a history student, this is both exciting and overwhelming. Names of clans, castles, and samurai scroll by quickly, making it a challenge to track who’s who at first. Yet this flood of information has a purpose; it pushes the player to learn by doing, picking up historical facts organically through play. For example, you might notice a retainer named Yoshioka Myōrin-ni labelled as an “onna-musha” (female warrior) and be inspired to look her up (Hernon 2025).

Screenshot 3: Myōrin Yoshioka’s in-game profile highlights her role as an onna-musha who defended Tsurusaki Castle. The game includes her real-life achievements and unique traits, encouraging players to research her story further.

Indeed, Myōrin turns out to be a real figure renowned for defending Tsurusaki Castle in 1586 with clever tactics and ferocity. She donned armour and led a ragtag garrison that repelled 16 assaults by a much larger Shimazu force (Turnbull 2012, pp. 47–48). Discoveries like these make the game more than just entertainment; it becomes an interactive textbook of the Sengoku period.

Historical Events and Life Lessons

Screenshot 4: Strategic map view from Nobunaga’s Ambition: Awakening, centered on Kyūshū. Players control territory like the Ōtomo domain (blue), while distant scripted events from across Japan unfold, emphasizing the broader scope of Sengoku-era conflict beyond the player’s immediate region.

Unlike many strategy games, Nobunaga’s Ambition doesn’t rely solely on freeform sandbox gameplay; it interweaves real historical events as optional cutscenes or “hearsay” vignettes that trigger as the years pass. You might be managing your fief when an event suddenly fires from distant lands, dramatized scenes of clashing clans. Choose to view it, and you witness a formative power struggle that subtly introduces the real history behind the game’s world. These scripted events function like brief history lessons amid the action, offering context and consequences beyond the player’s immediate territory (Heath 2023).

The Economy of Conquest: Rice, Koku, and Power

Beyond scripted events, Awakening offers deep strategic gameplay that mirrors the real logistical burdens of Sengoku-era warlords. Players must conquer provinces, manage resources, appease retainers, and balance economic development with military ambition. Armies aren’t paid in gold; they’re sustained with rice, tying military expansion directly to agricultural productivity. Historically, one koku (石) equalled about 150 kilograms of rice, the amount considered sufficient to feed one person for a year (Totman 2005, pp. 206–207). In-game, this economic structure is reproduced, as expanding your domain means expanding rice fields, improving ports, and developing infrastructure. Conquest is literally fuelled by grain.

In my playthrough, for example, investing in the port of Kunisaki in northeastern Kyūshū yielded considerable returns, leveraging coastal trade to generate revenue from foreign goods. Such mechanics subtly reinforce historical truths, a domain that needed not just manpower, but ample rice to feed it. The adage “an army marches on its stomach” becomes vividly real: overextension without logistical foresight results in famine and military collapse. At one point, I swept aggressively across Kyūshū, only for my forces to falter not in combat but on the march. Soldiers suffered from atrophy, a mechanical simulation of dwindling health due to poor provisioning. Entire garrisons withered before reaching the battlefield, undone not by swords but by empty stores.

The lesson recalls later examples of overreach in Japanese history, such as Japan’s fuel shortages during World War II, which undermined its military campaigns (Goralski and Freeburg 1987, pp. 316–319). Awakening quietly teaches that conquest demands patience, secure supply lines, and economic foresight. Through these systems, the game delivers a quietly radical message: in feudal Japan, true power didn’t lie in castles or swords alone; it began in the rice fields.

Islands, Castles, and Forgotten Histories: The Sō Clan, Moji, and the Ghosts of Fortresses

Nobunaga’s Ambition: Awakening occasionally gestures beyond the main island of Honshū, drawing attention to the peripheries of Sengoku Japan, places like Tsushima, where the Sō clan maintained an uneasy but strategic presence between Japan and Korea. Although only a minor faction in the game, the Sō were historically significant, as they repelled Mongol invasions in the 13th century and later monopolized trade with the Joseon dynasty in Korea through treaties, essentially acting as intermediaries between the two kingdoms. Their stronghold, Kaneishi Castle (in Izuhara, Tsushima), was designated a National Historic Site in 1995, a poignant reminder that Japan’s unification story has always existed in dialogue with worlds beyond its shores (Agency for Cultural Affairs 2017, p. 23).

One surprising feature in Awakening is the ability to adopt Christian Proselytization as a policy. In-game, it’s a minor mechanic, boosting certain officers’ loyalty or aiding diplomacy, but it gestures toward a major historical undercurrent. During the late Sengoku period, several daimyō, especially in Kyūshū, welcomed Jesuit missionaries for strategic reasons, including access to foreign firearms, trade opportunities, or even spiritual legitimacy. The Ōtomo clan, for instance, became one of Japan’s earliest and most vocal Christian patrons; its lord Ōtomo Sōrin was among the few warlords to convert to Catholicism (taking the baptismal name Francisco in 1578), and contemporaries noted that some Kyūshū lords embraced the new faith largely to attract Portuguese guns and commerce (Turnbull 2003, p. 103). Yet by 1614, the Tokugawa shogunate had reversed course. Christianity was banned outright, and its followers were persecuted (Elisonas 1991, p. 368), a stark historical reality dramatized in Shūsaku Endō’s novel Chinmoku (Silence) and brought to global audiences by Martin Scorsese’s 2016 film. (The movie, set in the 1630s, portrays the brutal consequences of that shift: underground worship, torture, and systematic violence used to root out the foreign religion.) Awakening doesn’t narrate this tragic trajectory directly, but it’s embedded in its systems and notably, in its map. If you pursue the path of Christian conversion in the game, you’re treading on precarious ground that history would later erase.

A similar sense of loss haunts the ruins of Moji Castle, once located in northern Kyūshū. In Awakening, Moji appears in certain scenarios (reflecting its real-life importance in the 1560s), but by the 1600 scenario, it vanishes entirely, mirroring historical reality. For example, in the 1560 start date, Moji is under the Ōtomo clan; by 1567, it passes to the Mōri clan, and by 1591 to the Kuroda, yet after 1600, both Moji Castle and the Ōtomo themselves are gone, erased from the map. Watching Moji Castle fall after a fierce siege, only to vanish from the campaign map, was unexpectedly moving. The game reflects reality: most Sengoku castles were ultimately lost to time, their fates sealed not by enemy armies but by political consolidation, neglect, or later redevelopment.

Historically, Moji Castle was not destroyed in a single cataclysmic event. (The 1561 Siege of Moji, which saw the Ōtomo enlist three Portuguese ships to bombard the castle, one of the first uses of Western cannon in Japan, is a famous episode, yet the fortress survived that onslaught.) Instead, Moji’s demise came from structural changes in governance. After reunifying the realm, the Tokugawa regime enacted the “One Castle per Province” law (一国一城令) in 1615, which forced every domain to demolish extra fortresses and keep only a single stronghold. Moji, being a secondary castle in the Mōri-held Chōshū domain, was thus decommissioned. A few years later (by 1617), it was abandoned entirely. During the Meiji era, the site gained new life as Moji Port became a hub of industrial modernization. The Imperial Japanese Navy built a fortress on the same hill in 1892, demolishing most of the old castle’s remnants in the process. Today, only fragments of Moji’s stone walls, Ishigaki, remain, hidden in the wooded hills above the harbour (Jcastle.info 2022).

This act of historical attrition, where castles fade from existence as authority shifts, is poignantly mirrored in the gameplay. As I played, I found myself longing for modern Japan to somehow reconstruct or memorialize these vanished sites, perhaps through XR (extended reality) technology, allowing future generations to walk their grounds once more (much as we can now virtually visit Angkor Wat). One final curiosity that emerged from this exploration is that some daimyo clans survive to this day. The Mōri clan, for example, still maintains its lineage and estate in Yamaguchi Prefecture (Mōri Museum n.d.), and descendants of once-powerful families, such as the Shimazu, continue to engage in ceremonial and cultural roles (Sengan-en 2023). In parallel with Ghost of Tsushima’s developers, who consulted with the Tsushima mayor and descendants of historical figures (Scullion 2021), one wonders what conversations might arise if game studios spoke directly with these families, the heirs of history, and advised the tellers of digital myth.

Immersion vs. Information Overload – The Push and Pull

Screenshot 5: Tutorial overlay explaining how daimyō govern from a main base and manage counties. Screens like this highlight the game’s layered strategy systems, informative, but dense enough to halt immersion until learned.

Nobunaga’s Ambition: Awakening is vast, intricate, and at times overwhelming. I found myself juggling multiple clan names, officer stats, alliances, and resource bars. Despite having a background in Japanese history, I often felt swamped by unfamiliar figures and mechanics. To manage the overload, I toggled the game’s tutorial and alerts settings as needed. Early on, I relied on the guidance of tooltips, which explained mechanics like diplomacy and farming. After gaining a deeper understanding, I turned off tooltips for a more immersive feel. However, I found myself reactivating it as a means to navigate in-game complexities. Realm management became a rhythm, immersion, interruption, and return. Perfect for history students willing to pause the action to study the game’s “historical textbook”. In the end, I came to embrace feeling overwhelmed, as every time I paused to learn something, a historical nuance, a forgotten castle, an economic concept, I returned with sharpened understanding.

Cultural Details: The Samurai Topknot

Screenshot 6: Samurai retainers with traditional shaved foreheads and tied topknots (chonmage) in Nobunaga’s Ambition: Awakening. This hairstyle, historically accurate to the Sengoku period, ensured helmets stayed secure and symbolized warrior status.

That curiosity extends beyond strategy. The game quietly educates players on cultural details through its character art, terminology, and systems. One example: the shaved heads and tied topknots (chonmage) worn by older samurai officers. Far from an artistic exaggeration, this hairstyle reflects authentic Sengoku-period norms. The front-half shave (sakayaki) served both practical and symbolic roles, ensuring helmets (kabuto) stayed secure during battle, keeping warriors cooler under heavy headgear, and signaling samurai status (NHK World-Japan 2024). Over time, it became a formal requirement among the warrior class. For newcomers, these half-bald generals might seem jarring, but they’re a reminder that even hairstyles carry historical significance.

Siege Warfare and Human Drama

Screenshot 7: A cutscene from Nobunaga’s Ambition: Awakening shows Toishi Castle engulfed in flames. These dramatized siege events highlight the brutality and high stakes of Sengoku warfare.

On the tactical level, the game offers a wealth of insights into Sengoku warfare. Castles are central to military campaigns, and sieges carry both strategic weight and narrative tension. In my campaign, I often took a more aggressive route, launching direct assaults and raids on fortresses rather than whittling them down over time. While costly, this brute-force approach could succeed with enough momentum. The game also allows players to soften targets through subterfuge, including blockades, sabotage, and psychological tactics such as spreading rumors, inciting revolts, or disrupting supply lines.

Once a stronghold falls, the player faces key decisions, whether to execute captured enemy officers, release them, or recruit them into their ranks. Each choice has consequences, reflecting the delicate balance between fear and loyalty that warlords like Nobunaga managed in real life.

Screenshot 8: Treatment options for captured officers subtly foreground the political calculations behind Sengoku warfare, rather than reducing conflict to mere conquest.

Historical records show both pragmatism and brutality: Nobunaga famously absorbed skilled enemies when useful, but showed zero tolerance for ideological opposition, most infamously in 1571, when he ordered the destruction of the Enryaku-ji temple complex on Mt. Hiei, slaughtering thousands of warrior monks to secure his rear flank and eliminate a powerful religious threat (Lamers 2000, pp. 75–76).

Visual Authenticity and Artistic License

Screenshot 9: Oda Nobunaga as he appears in Nobunaga’s Ambition: Awakening — polished, youthful, and adorned in ornate Western-style armour. The game's design gives him an air of theatrical charisma and stylized grandeur.

One delightful aspect of Nobunaga’s Ambition: Awakening is its commitment to historical authenticity in visuals and terminology. The portraits of characters generally reflect Japanese artistic conventions or known depictions. However, Oda Nobunaga himself is portrayed with striking features and a commanding presence, almost supernaturally composed, and far removed from how contemporary accounts described him.

In a famous 1583 portrait attributed to Kanō Sōshū, the historical Nobunaga is depicted with a modest, narrow face, a shaved pate, and plain robes reflecting his later-life Buddhist tonsure (Kanō Sōshū, 2019). This image contrasts sharply with his polished, dramatic appearance in many modern games and anime. Awakening doesn’t lean fully into fantasy, but its version of Nobunaga still exudes a leading-man aura that the historical warlord’s receding hairline likely didn’t.

Before I even pressed “New Game,” I knew this would be different. Trumpets flared, drums rolled, and Oda Nobunaga, impossibly handsome, stared out from the screen like a digital deity. As I played, I couldn’t help but reflect: “Did he really look like this?” The game makes him particularly cool, almost godly. It’s like they deified him. While not unique to Nobunaga, this “prettying up” stands out, especially given the modest and austere nature of his historical portraits.

Unifying Kyūshū: A Campaign Chronicle

Screenshot 10: The Regional Unity campaign completion sequence for the Ōtomo clan. The player is congratulated for bringing peace to Kyūshū, offering a moment of closure and reflection on the historical stakes of conquest.

The campaign to unify Kyūshū in Nobunaga’s Ambition: Awakening illustrates the game’s capacity to simulate long-term strategic struggle. Playing as the Ōtomo clan, the goal of subduing the island became a complex undertaking, marked by fierce resistance from the Shimazu, who dominated southern Kyūshū and commanded formidable defenses. Initial offensives often failed due to overextension, forcing retreats and a need to defend northern territories. Over time, aggressive advances gave way to more measured tactics, involving the capture of peripheral castles and the slow building of momentum for decisive strikes.

Eventually, Satsuma, the Shimazu’s final stronghold, fell. The game presented a more reflective moment: a cutscene in which Yoshishige Ōtomo declares that peace has been brought to the lands of Chinzei, the classical name for northern Kyūshū. This “Regional Unity” ending marked more than just military success; it evoked the layered historical weight of the island and the clans that vied for its control.

Although brute force ultimately prevailed, resource management, diplomacy, and long-term planning played supporting roles in the outcome. Policies, trade development, and officer loyalty all helped shape the pace of conquest, even on the game’s easiest difficulty setting. Historically, the Shimazu came close to unifying Kyūshū by 1586, before Toyotomi Hideyoshi’s massive Kyūshū Campaign in 1587 forced their surrender (Totman, 2005, pp. 211–212). Finishing the campaign by 1570 offered not only a sense of alternate history but also a deeper appreciation for how hard-fought real unification must have been. In that moment, I wasn’t just playing a game. I was participating in a centuries-long story, a digital echo of power, memory, and resistance. Nobunaga’s Ambition: Awakening doesn’t just simulate the past. It invites you to inhabit it, struggle with it, and, in some small way, reshape it.

Screenshot 11: Strategic map of Kyūshū under the Ōtomo clan’s control. The game’s visualisation of geography and stronghold networks reinforces the spatial logic of Sengoku unification campaigns.

Bridging Worlds Through Play

As a reviewer and M.A. student of Japanese studies, Nobunaga’s Ambition: Awakening became more than just entertainment. It was both a test of historical knowledge and a spark for new questions to arise. Games like this allow players to live through history, fight through legend, and wrestle with interpretation in ways just as engaging as any seminar. The experience rekindled my passion for the period, prompting me to research figures like Myōrin and explore pivotal moments surrounding Oda Nobunaga, even those that never unfolded in my 1570-ending campaign. It also led me to revisit the history of places I once lived in Kyūshū, where cities and castle sites I had previously passed by became newly significant. I may yet reopen the game to witness events like the Honnōji Incident firsthand, not just to see how they’re represented, but to experience how history bends when placed in the player’s hands.

As a grand strategy game, Awakening is deep and occasionally daunting. Newcomers might feel lost amid the sprawling menus and the slow, deliberate pace. While the tutorials are thorough, many (myself included) will likely skip ahead, only to suffer the consequences. However, for those who truly appreciate its depth, it becomes addictive. Each campaign is a slow-burning process of political scheming, resource balancing, and military maneuvering, and its historically grounded systems teach by doing. Whether you leave historical cutscenes on for mini-lessons or opt for a sandbox “what-if” experience, the game balances authenticity with replayability. Its art and music create a fittingly solemn atmosphere; even if its visuals aren’t flashy, they serve the tone well. And thanks to the game’s attention to historical accuracy, from daimyo alliances to rice yields, players often come away with far more knowledge than they expect. I certainly did.

Having wrapped up my Kyūshū campaign, I’ve now turned to two very different portrayals of Nobunaga and the Sengoku era: Nioh and Nioh 2. These myth-infused action-RPGs depict Nobunaga not just as a cunning warlord, but as a near-demonic legend, “the Demon King of the Sixth Heaven.” This next chapter will explore how games mythologize Nobunaga, from yokai aesthetics and temple guardians to the fictional Yasuke and beyond. The war for Japan’s land may be over in Awakening, but in Nioh, the war for its soul is just beginning.

Screenshot 12: Oda Nobunaga as “Demon King of the Sixth Heaven” in Nioh, flanked by his spectral guardian spirit. This mythologized image sets the stage for Part II of this study.

References

Agency for Cultural Affairs (2017) Japan Heritage: The Frontier Islands of Iki and Tsushima, Goto – The Ancient Bridge to the Continent. Tokyo: Government of Japan. Available at: https://www.bunka.go.jp/seisaku/bunkazai/nihon_isan/pdf/nihon_isan_pamphlet_english.pdf

Elisonas, J.S.A. (1991) ‘Christianity and the Daimyo’, in Hall, J.W. (ed.) The Cambridge History of Japan, Volume 4: Early Modern Japan. Cambridge: Cambridge University Press, pp. 301–372.

Endō, S. (1966) Silence (Chinmoku). Tokyo: Shinchosha.

Goralski, R. and Freeburg, R. (1987) Oil and War: How the Deadly Struggle for Fuel in WWII Meant Victory or Defeat. New York: William Morrow.

Heath, M. (2023) ‘Nobunaga’s Ambition: Awakening Review’, GameGrin, 7 August. Available at: https://www.gamegrin.com/reviews/nobunagas-ambition-awakening-review/

Hernon, M. (2025) ‘Legendary Japanese Onna-Musha (Female Warriors) | List of 7’, Tokyo Weekender, 31 July. Available at: https://www.tokyoweekender.com/art_and_culture/japanese-culture/legendary-japanese-onna-musha-female-warriors/

Jcastle.info (2022) Moji Castle. Available at: https://jcastle.info/view/Moji_Castle

Kanō Sōshū (2019) Oda Nobunaga Portrait. World History Encyclopedia, 5 June. Available at: https://www.worldhistory.org/image/10886/oda-nobunaga-portrait/

Lamers, J. (2000) Japonius Tyrannus: The Japanese Warlord Oda Nobunaga Reconsidered. Leiden: Hotei Publishing.

Mōri Museum (n.d.) Mōri Museum. Available at: https://mohri-museum.com/museum/

NHK World-Japan (2024) Samurai Chonmage Truth – Time and Tide. Broadcast 9 March. Available at: https://www3.nhk.or.jp/nhkworld/en/shows/3022046/

Scorsese, M. (2016) Silence. [Film]. Paramount Pictures.

Scullion, C. (2021) ‘Ghost of Tsushima devs to be made permanent ambassadors of the real island’, VGC, 5 March. Available at: https://www.videogameschronicle.com/news/ghost-of-tsushima-devs-to-be-made-permanent-ambassadors-of-the-real-island/

Sengan-en (2018) Sengan-en – Stately home and gardens of the Shimadzu family. YouTube video, 8 October. Available at:

Totman, C.D. (2005) A History of Japan. 2nd edn. Oxford: Blackwell.

Turnbull, S. (2003) Samurai: The World of the Warrior. London: Osprey Publishing.

Turnbull, S. (2012) Samurai Women 1184–1877. Oxford: Osprey Publishing.

Earlier this week, Harvard Business Review ran a piece by Mathis Bitton and Elizabeth Haas on Hamburg’s unlikely turn from zoning deadlock to large-scale refugee integration. The pivot point was not a political overhaul or a sudden influx of resources but the introduction of CityScope, an MIT Media Lab platform that combined vast data modelling with participatory design.

In Hamburg’s case, the problem was familiar to anyone who has tried to balance realism and playability in a game set in a real place: competing demands, finite space, entrenched rules. The city needed to accommodate tens of thousands of new residents without fracturing existing communities or overloading fragile infrastructure. CityScope’s answer was not a single master plan but a simulation environment where stakeholders could manipulate parameters, test “what if” scenarios, and see the consequences unfold in real time.

Workshops replaced town halls. LEGO-like models stood in for buildings and transit lines. Augmented reality layered performance metrics — environmental impact, energy efficiency, transit accessibility — over the map as participants shifted pieces around. The brilliance was not in showing people a finished plan, but in making them co-authors of the city’s future, using a design language they could intuitively grasp.

For game designers, the parallels are obvious. This is environmental storytelling in a live, iterative setting: mechanics and rules are transparent, player agency is real but bounded, and feedback loops are immediate. In historical game design — whether reconstructing 18th-century Edo or a speculative near future — these principles hold. Players (or citizens) understand the constraints not as arbitrary obstacles but as part of the world’s logic. They experiment, fail, adjust, and negotiate shared space.

For historians, CityScope’s process is a reminder that archives are not just about preservation but about reassembly. When we reconstruct the past, we work with incomplete, unevenly distributed evidence; we simulate possibilities, weigh trade-offs, and — whether we admit it or not — create models that persuade as much as they inform. Hamburg’s workshops resemble what happens when a research seminar takes the leap from interpretation to counterfactual modelling: “If we change this variable, what else must shift for the system to hold together?”

The point is that AI-assisted consensus-building is not simply a governance tool. It’s part of a broader design methodology — one that blends system modelling, historical context, and participatory engagement. Whether you’re planning housing in a real city, designing a historically plausible game space, or constructing an argument about the past, the same questions apply: What rules are negotiable? Who gets to move the pieces? And how do you visualise the consequences clearly enough that people can argue productively about what comes next?

Hamburg’s experience suggests that the future of decision-making — in both civic life and digital culture — may belong to those who can merge the historian’s archive with the designer’s toolkit, turning data into worlds people can explore, inhabit, and change together.

References

Bitton, Mathis, and Elizabeth Haas. “How AI Can Help Tackle Collective Decision-Making.” Harvard Business Review, August 12, 2025. https://hbr.org/2025/08/how-ai-can-help-tackle-collective-decision-making

MIT Media Lab. “CityScope: An Urban Modeling and Simulation Platform.” MIT Media Lab. Last modified May 13, 2022. https://www.media.mit.edu/publications/cityscope/

OECD Observatory of Public Sector Innovation. “CityScope FindingPlaces: HCI Platform for Public Participation in Refugees’ Accommodation Process.” OECD OPSI. https://oecd-opsi.org/innovations/cityscope-findingplaces-hci-platform-for-public-participation-in-refugees-accommodation-process/

MIT Media Lab. “FindingPlaces.” MIT Media Lab: City Science. https://www.media.mit.edu/projects/finding-places/overview/

MIT Media Lab. “Shifting Priorities, Finding Places.” MIT Media Lab. https://www.media.mit.edu/posts/shifting-priorities-finding-places-1/

MIT Media Lab. “City Science Lab @ Hamburg.” MIT Media Lab: City Science. https://www.media.mit.edu/projects/cityscope-hamburg/overview/

Noyman, Ariel, et al. “Finding Places: HCI Platform for Public Participation in Refugees’ Accommodation Process.” arXiv, November 26, 2018. https://arxiv.org/abs/1811.10123

Revel, Manon, and Théophile Pénigaud. “AI-Facilitated Collective Judgements.” arXiv, March 6, 2025. https://arxiv.org/abs/2503.05830

Zhang, Angie, et al. “Deliberating with AI: Improving Decision-Making in Graduate Admissions.” arXiv, February 22, 2023. https://arxiv.org/abs/2302.11623

Karen Hao's Empire of AI is a deeply researched investigation into the rise of OpenAI and, by extension, the political, technological, and social stakes embedded in contemporary AI development.

Drawing on hundreds of interviews and a significant body of internal correspondence and documentation, Hao reconstructs the organizational transformation of OpenAI from a nonprofit research collective to a well-funded, secretive corporate actor at the center of what she calls a new imperialism of artificial intelligence.

The book excels in three interrelated domains. First, as a study of OpenAI as an industry leader, it offers granular insight into the internal operations, shifting ethos, and strategic realignments of OpenAI from its founding through the ChatGPT era.

Second, as a psychological portrait, it casts OpenAI as an expression of Sam Altman's personality, whose oscillation between techno-idealism and market opportunism mirrors the broader contradictions of Silicon Valley.

Third, and most critically, Hao draws upon the history of imperialism and colonialism to situate OpenAI's trajectory within a global frame: as the vanguard of an emerging techno-empire that extracts labor, data, and ecological resources from a planetary periphery to build models whose risks and benefits are managed by a small, remote, powerful billionaire elite.

Listen to Karen Hao’s Empire of AI at Audible

Hao brings to this analysis a rare breadth and depth of understanding of the intersection between technology and society. Trained as an MIT scientist and experienced as a tech journalist, she writes with clarity and precision about both machine learning architectures and their political implications.

Most notably, she advances a historical analogy between AI development and colonialism. The comparison is not rhetorical flourish but a structural argument: the concentration of technological power, the subordination of marginalized labor, and the appropriation of common resources serve as pillars of both systems. Hao does not merely diagnose this dynamic but offers an alternative: a discrete, labor-enhancing vision of AI that supports rather than supplants human capacities.

The book’s empirical detail and narrative structure are compelling, but its normative ambition is what sets it apart. Hao does not assume technological determinism. Rather, she insists that AI systems, like empires, are built through political choices. Her critique of OpenAI is not reducible to personal antagonism or institutional mismanagement. It is a claim about the incompatibility of democratic accountability with the current mode of AI production.

The Audible edition, which Hao narrates herself, reinforces this message with precision and conviction, her delivery augmenting the book’s clarity and urgency of purpose. Empire of AI thus contributes not only to public understanding of OpenAI but to a wider reckoning with the infrastructures of digital power.

Hao's book is not just an exposé of a single firm but a field report from a contested future. It is valuable not because it answers all questions, but because it frames the right ones: Who builds AI? For whom? With what resources? And at what cost? The book has provoked fierce opposition, most notably from OpenAI's leadership.

Sam Altman has publicly criticized Hao's portrayal of the company and its mission, while other figures within the AI industry have also sought to discredit her reporting. These responses underscore the stakes of Hao's challenge to not only technological practices but the authority of those who currently shape the agenda. This makes the book all the more significant and well worth reading.

Ghost of Tsushima (2020) is more than a stylish samurai adventure – it is a case study in how video games construct and circulate a vision of Japanese history for global audiences. Developed by the American studio Sucker Punch and set in 13th-century Japan, the game follows Jin Sakai, a samurai defending Tsushima Island from Kublai Khan’s Mongol invasion in 1274. Ghost presents a highly stylised jidaigeki-inspired narrative that plays fast and loose with facts in the service of drama (Townsend, 2024). In Martin Roth’s terms, the game exemplifies the “multilayered spatialisation” of modern media: its content reflects both local Japanese lore and global cinematic tropes, and its production and distribution span cultures to deliver a packaged notion of “Japaneseness” worldwide. Using in-game visuals as historiographic evidence, this review critically examines Ghost of Tsushima’s narrative fiction, gameplay systems, cultural representations (and omissions), and its reception, all to explore how a Western-made game became a visual history of Japan that players keenly embrace.

Game Narrative as Period-Drama Fiction

Screenshot 1: Mongol invasion fleet at Komoda Beach, as seen in the opening cinematic. A mythic tableau that sets the tone for the game’s epic, romanticised framing of history. Credit: Sucker Punch

From its opening cinematic, Ghost of Tsushima frames the Mongol invasion as a grand, romantic conflict. The story centres on Jin Sakai, one of the last samurai on Tsushima after the invading Mongols slaughter the island’s defenders, and his quest to save his homeland and his uncle, Lord Shimura, from Khotun Khan (a fictional cousin of Kublai Khan) (Wu, 2020). Major characters, such as Yuna (a thief-turned-ally), Sensei Ishikawa (an elderly archery instructor), and Lady Masako (a vengeful noblewoman), bring depth to a tale of honor, sacrifice, and guerrilla resistance. Yet this narrative is deliberately ahistorical. The developers chose to make an entertaining period drama, “not a history lesson”, ensuring the game “won’t make a game that Japanese people will feel is insulting”. In practice, Ghost freely fictionalises events. In reality, Tsushima’s Mongol attackers were ultimately repelled by a typhoon, not a lone samurai, and many picturesque locations in-game never existed on the island. By leaning into a Kurosawa-style legend rather than strict accuracy, the game “earns belief” through evocative storytelling rather than factual recounting. The result is a narrative steeped in samurai cinema tropes: lone heroes, noble sacrifices, and clear moral stakes. It’s a consciously mythic vision of 1274 Japan, one that has been crafted to resonate as authentic in spirit, even while taking liberties with history (Ivan, 2021).

Gameplay Mechanics: Combat, Exploration, and Style

Screenshot 2: Jin Sakai traverses Tsushima’s stylised wilderness. Gameplay merges exploration with cinematic spectacle, reinforcing the player’s identity as a wandering hero (Narin, 2022). Credit: Sucker Punch

Playing Ghost of Tsushima feels like stepping into a samurai film. The gameplay balances elegant sword-fighting mechanics with stealth and exploration, all presented through an artful lens. In combat, players can challenge foes honourably in one-on-one standoffs or adopt the “Ghost” approach, employing assassination, darts, and trickery that Jin’s samurai upbringing once forbade. The swordplay is grounded in what developers called “mud, blood and steel”, emphasising lethal precision; every swing and parry carries weight, and enemies fall in just a few strokes if the player’s timing is true. To make duels feel cinematic, the game even includes an old-fashioned duel mode (complete with tense pre-fight stances and wind-swept surroundings), reminiscent of chanbara films. When traversing the island, Ghost eludes intrusive HUD markers in favour of storyline guidance, a Guiding Wind system that literally blows in the direction of objectives, replacing the typical video game mini-map with a gust of nature. This design choice not only preserves immersion in the world but also reinforces the game’s theme of living in harmony with nature and intuition, as a wandering samurai might. Players can roam Tsushima on horseback, follow foxes to hidden Shinto shrines, compose poems at scenic spots, and engage in spontaneous sword duels under falling cherry blossoms. The developers included a robust Photo Mode to encourage players to capture these moments and even added a special “Kurosawa Mode”, a black-and-white film-grain filter with Japanese dialogue audio, to pay homage to classic samurai films (Cork, 2020). In short, every mechanic, from the stylised duels to the way maple leaves swirl on the breeze, works in concert to make the player feel like the hero of a lost Akira Kurosawa epic.

Cultural Representation and Historical Omissions

Screenshot 3: Jin composes a “haiku” near ancestral graves. A symbolic moment that evokes spiritual reverence, though historically anachronistic (Colucci, 2020). Credit: Sucker Punch

Ghost of Tsushima constructs a vision of Japanese identity that is at once loving and curated. On one hand, the game overflows with cultural markers meant to celebrate Japan’s heritage, the samurai code of honour (bushidō) that Jin struggles not to betray, the practice of bowing to fallen enemies or wiping one’s blade (the chiburui ritual) after battle and moments of zen like composing haiku in beautiful natural vistas. Players and critics have praised such details for painting “a true picture of Japanese culture”, elevating the game to a “love letter to the samurai genre”, as director Nate Fox put it (Weber, 2020). At the same time, these representations are selective. The game presents an idealized samurai ethos, emphasizing loyalty, honor, and courage, while quietly sidelining aspects of Kamakura-era society that don’t fit the romantic narrative. Notably, Ghost never acknowledges shudō, the tradition of erotic male mentorship bonds among samurai, which was a documented part of samurai life in feudal Japan. This omission aligns with the game’s overall pattern of sanitising or simplifying history to avoid discomfort.

Screenshot 4: Khotun Khan in fire-lit framing before executing Lord Shimura. The game's villain is visually coded as absolute evil, reinforcing a mythic moral binary between barbaric invaders and noble samurai. Credit: Sucker Punch

Similarly, the Mongol invaders in Ghost are depicted as brutal enemies (raiding villages, burning fields), but are stripped of certain grim realities; for instance, historical chronicles of Mongol campaigns include reports of survival cannibalism during sieges, a dire detail the game pointedly leaves out. By omitting these darker or more complex truths, Ghost of Tsushima constructs a streamlined cultural myth: a pure clash of samurai valour versus barbarian invaders, with little to sully the archetypal heroism on display (Spartan The Conqueror, 2020). This mythmaking has a purpose; it delivers an emotionally resonant and easily digestible tale of triumphant “Japanese spirit”, even if that requires overlooking the messier nuances of 13th-century history. In the context of visual historiography, we might ask: whose vision of Japan is being foregrounded and whose is being quietly left out of frame?

Comparative Cultural Critique: Ghost vs. Assassin’s Creed: Shadows

The largely positive response to Ghost of Tsushima’s romanticised history stands in stark contrast to the scrutiny heaped on another recent portrayal of samurai-era Japan: Ubisoft’s Assassin’s Creed: Shadows. Ghost has been widely celebrated, especially in Japan, for its respectful tone and cinematic authenticity, “embodying the spirit of Japanese period dramas” in a way that even earned praise from local authorities. Japanese gamers voted it their Game of the Year 2020, and it received a rare perfect score in Famitsu, indicating that its American creators had succeeded in crafting a vision of Japan that resonated with Japanese audiences (Ashcraft, 2020). By contrast, Assassin’s Creed: Shadows (a 2024 title set in 16th-century Japan) became a lightning rod for controversy before it even hit shelves. Ubisoft’s game, developed by a French studio, was criticised for various cultural inaccuracies and insensitivities: a trailer showed a Shinto shrine being violently desecrated by the player-character, which Japanese commentators (up to the Prime Minister) condemned as disrespectful (SCMP Asia Desk, 2025). The backlash forced Ubisoft to alter content in a day-one patch amid accusations of “cultural insensitivity” (Carpenter, 2025; Chundawat, 2025; Shutler, 2025).

Why did Ghost of Tsushima escape such critique while Assassin’s Creed: Shadows ignited it?

One explanation lies in differing approaches to tone and meticulous consultation. Sucker Punch worked closely with cultural experts and modelled Ghost after beloved Japanese media (samurai cinema, period dramas), explicitly aiming not to “insult” Japanese audiences. Ubisoft, by comparison, waded into Japan’s history with a franchise known for sandbox mayhem and faced swift reproach when that freedom crossed cultural red lines (for example, allowing players to graffiti or destroy sacred sites without contextual sensitivity). There may also be a double standard at play regarding who is allowed to mythologise Japan. As one analyst notes, Ghost “romanticises bushidō ethics but avoids scrutiny, while Assassin’s Creed: Shadows faces outrage” for similar fictionalisation. Both games take liberties for the sake of fiction, yet only the Western-produced AC: Shadows was met with diplomatic backlash and online uproar. This suggests that Ghost of Tsushima earned a degree of trust, perhaps due to its clear admiration for Japanese culture, which insulated it from accusations of appropriation. It opened a conversation about representation in games: Who gets to tell Japanese history, and under what terms? The differing fates of Ghost and AC: Shadows illustrate how careful cultural positioning (and fan goodwill) can make or break a game’s reception when navigating the minefield of historical representation.

Metadata, Paratextual Analysis and Multilayered Spatialisation through Fan Discourse

Screenshot 5: Kurosawa Mode transforms a fiery battlefield into a cinematic spectacle, as Jin Sakai charges into combat amid black-and-white chaos, an homage to classic samurai cinema. Credit: Sucker Punch

Martin Roth’s metadata-based framework emphasises that video games must be understood within the multilayered spatial contexts of their production, distribution, and reception, not just through their in-game content (Roth, 2025, p. 112). In this sense, “metadata” refers not merely to back-end data but to attributes such as developer origin, platform, genre labels, and marketing networks, which profoundly shape how a game is culturally positioned and interpreted.

In the case of Ghost of Tsushima, this framework allows us to analyse how the game’s transnational construction contributes to its perceived authenticity. Developed by Sucker Punch Productions (a U.S. videogame developer) and published by Sony Interactive Entertainment (a Japanese platform-holder), the game blends American production values with Japanese narrative and aesthetic traditions. Its genre tags, “Action”, “Historical”, “Open World”, “Samurai”, and features like Kurosawa Mode or original Japanese voice acting act as paratextual cues that signal cultural legitimacy (Ivan, 2021). These layered elements of metadata do not merely describe the game; they structure its reception and interpretation. As Roth explains, games are “spatially unstable objects” whose meaning shifts depending on the industrial and cultural contexts they traverse (Roth, 2025, p. 95).

The development team undertook at least two research trips, alongside personnel from Sony Japan, to Tsushima and other historical sites. There, they consulted local artisans, Japanese historians, and martial arts experts to inform the authentic cultural depiction (Takahashi, 2020). The result was well received; Japanese critics praised the game’s fidelity to samurai cinema, with some claiming it felt “more Japanese than Japanese games” (Ashcraft, 2020). Analysts such as Amber V and Koselke (2024) note that its lack of ideological imposition and focus on entertainment over instruction helped secure its positive domestic reception.

Building on this, Roth’s idea of spatialised metadata helps explain not only how games are produced and distributed, but also how their cultural significance emerges in circulation, especially through player communities and paratextual discourse. Fan commentary, reviews, forum posts, and user-generated media function as interpretive layers through which meaning is constantly negotiated. As Švelch (2020) notes, contemporary game paratextuality focuses on how players and fan cultures contribute to the game’s evolving cultural footprint.

Paratextual analysis of Ghost of Tsushima’s online reception, such as Steam reviews and Reddit threads, reveals three major interpretive patterns that emerge consistently across fan commentary:

• Reverent immersion: Players frequently praise Ghost of Tsushima’s landscapes, sound design, and cinematic flair, often describing the experience with phrases like “breathtaking vistas” or noting that it made them “feel like a samurai.” This sense of immersion is frequently attributed not only to major narrative or mechanical elements but to subtle environmental details, such as guiding foxes, rustling wind, or falling cherry blossoms, that collectively reinforce the game’s perceived cultural authenticity.

• Kurosawa-inspired romanticism: The optional Kurosawa Mode is widely celebrated in fan discourse. Players cite classic films like Seven Samurai or Yojimbo when reflecting on the game’s visual tone, framing it as a tribute rather than an appropriation. These comparisons position Ghost of Tsushima within a familiar, globally recognisable samurai media lineage (Sheridan, 2020).

• Historical nitpicking: A vocal subset of players critiques the game’s liberties with historical fact, such as mixing armour styles from different centuries or including haiku poetry centuries before it emerged. These critiques, while often respectful, signal a tension between aesthetic stylisation and historical fidelity.

Screenshot 6: Jin and a fellow warrior stand in ornate, anachronistic armour amid battlefield wreckage. This exaggerated aesthetic triggers fan debate about stylisation versus authenticity. Credit: Sucker Punch

Together, these responses illustrate how Ghost of Tsushima is not just interpreted through gameplay or visuals, but through a multilayered process of cultural evaluation shaped by its production metadata and its reception in fan discourse. This analytical framework, which combines metadata, paratextuality, and spatial theory, highlights how digital games function not merely as entertainment but as dynamic sites of cultural mediation and mythmaking.

In developing my framework for analyzing Ghost of Tsushima, I draw from three intersecting strands of theory: Roth’s (2022) metadata analysis and multilayered spatialization, and Švelch’s (2020) conception and deployment of paratextuality. Building on the idea of fan discourse as paratext, Švelch (2020) allows for an analysis of the cultural life of the game beyond the software itself. Švelch observes that contemporary game paratextuality “focuses on players, fans and game cultures,” treating their contributions as integral to the media ecosystem (Švelch, 2020). A researcher might therefore employ paratextual or discourse analysis methods to investigate how audiences negotiate and reshape the game’s meaning. For example, one could collect a sample of Steam reviews and Reddit posts, then qualitatively code user utterances into themes such as immersion, cinematic language, historical accuracy, or cultural critique. This approach positions fan commentary not as peripheral, but as part of the multilayered meaning-making process that defines a game like Ghost of Tsushima.

Beyond qualitative themes, one might also quantify patterns in the discourse, such as how often players reference Kurosawa, or how expressions of approval compare to critiques. These data can help reveal how Tsushima’s visuals and historical framing are received and evaluated by its audience. Through such analysis, it becomes possible to trace how players negotiate the game’s authenticity claims: some may embrace cultural stylisation as being “true to spirit”, while others may reject specific elements as “inaccurate”! This framework is also transferable to other titles; for example, one could examine how Assassin’s Creed fans debate historical fidelity or how forum communities engage with cultural representation in any historically themed game. In each case, players’ remarks form a paratextual archive that sheds light on the reception, contestation, and co-construction of cultural meaning in games.

Conclusion

Ghost of Tsushima exemplifies a richly immersive exercise in visual historiography and cultural mythmaking, marrying a stylised narrative, stunning visual design, and smooth gameplay mechanics to deepen player engagement. This chapter’s analysis, drawing on Roth’s metadata framework and Genette’s notion of paratextuality, revealed how the game’s respectful use of samurai-era aesthetics and multilayered spatial design choices work in concert to create a compelling reimagining of 13th-century Japan. At the same time, Ghost of Tsushima’s curated historical vision consciously omits certain historical realities (such as the practice of shudō and the full brutality of the Mongol invasions) in service of a romanticised, transnationally palatable vision of Japan’s past. These selective omissions, while sanitising the past, enhance the game’s mythic narrative of honour and courage, reinforcing a heroic samurai ethos that resonates widely (Hashimoto, 2020). The cultural significance of Ghost of Tsushima is evident not only in its critical acclaim and enthusiastic reception by players worldwide, including praise from Japanese observers who lauded its authenticity (one industry veteran remarked the Western-made title felt “even more Japanese” than local works) (Video Games Chronicle, 2020), but also, in its real-world impact on heritage engagement, as seen when international fans helped fund the restoration of a Tsushima Island shrine (Video Games Chronicle, 2021). Ultimately, Ghost of Tsushima stands as a testament to how modern video games can serve as both entertainment and an interactive historical narrative, utilizing creative license and layered design to invite a global audience into a stylized yet culturally resonant past.

References

Amber V (2024) Ghost of Tsushima’s historical inaccuracies were accepted by the Japanese because of the devs’ non-imposing, entertainment-focused approach, says Japanese analyst. Automaton Media, 30 October. Available at: https://automaton-media.com/en/news/ghost-of-tsushimas-historical-inaccuracies-were-accepted-by-the-japanese-because-of-the-devs-non-imposing-approach/

Ashcraft, B. (2020) Ghost of Tsushima is being praised by Japanese critics. Kotaku, 15 July. Available at: https://kotaku.com/ghost-of-tsushima-is-being-praised-by-japanese-critics-1844387298

Carpenter, N. (2025) Assassin’s Creed Shadows day-one patch seemingly addresses Japanese government concern over destructible shrines. Polygon, 21 March. Available at: https://www.polygon.com/assassins-creed/543360/assassins-creed-shadows-temple-destruction-day-one-patch

Chundawat, B.S. (2025) Japanese PM Ishiba slams Assassin’s Creed for ‘cultural disrespect’ over shrine depiction. Udaipur Kiran, 21 March. Available at: https://udaipurkiran.com/japanese-pm-ishiba-slams-assassins-creed-for-cultural-disrespect-over-shrine-depiction/

Colucci, M. (2020) Find haiku locations in Ghost of Tsushima. Screen Rant, 25 July. Available at: https://screenrant.com/find-haiku-locations-ghost-tsushima/

Cork, J. (2020) I love Ghost of Tsushima’s ‘Kurosawa Mode,’ but it’s not for everybody. Game Informer, 16 July. Available at: https://gameinformer.com/2020/07/16/i-love-ghost-of-tsushimas-kurosawa-mode-but-its-not-for-everybody

Consalvo, M. (2007) Cheating: gaining advantage in videogames. Cambridge, MA: The MIT Press.

Genette, G. (1980) Narrative discourse: an essay in method. Translated by J.E. Lewin. Ithaca, NY: Cornell University Press.

Hashimoto, K. (2020) Ghost of Tsushima, Kurosawa, and the political myth of the samurai: unpacking the baggage of well-intentioned homage. Polygon, 23 July. Available at: https://www.polygon.com/2020/7/23/21333631/ghost-of-tsushima-kurosawa-films-samurai-japan-abe-politics

Ivan, T. (2021) Sucker Punch discusses Ghost of Tsushima’s approach to cultural authenticity. Video Games Chronicle, 15 March. Available at: https://www.videogameschronicle.com/news/sucker-punch-discusses-ghost-of-tsushimas-approach-to-cultural-authenticity/

Koselke, A. (2024) Ghost of Tsushima didn't need to be historically accurate because it didn't "impose any ideology" and placed "top priority on entertainment value". GamesRadar, 30 October. Available at: https://www.gamesradar.com/games/action/ghost-of-tsushima-didnt-need-to-be-historically-accurate-because-it-didnt-impose-any-ideology-and-placed-top-priority-on-entertainment-value/

Narin, C. (2022) Ghost of Tsushima screenshots. Blog at WordPress.com, 4 May. Available at: https://cihannarin.com/2022/05/04/ghost-of-tsushima-screenshots/

Roth, M. (2025) Unboxing Japanese videogames: a metadata-based approach to the production and distribution of spatial instability. Cambridge, MA: The MIT Press.

SCMP Asia Desk (2025) Japanese PM Ishiba condemns Assassin’s Creed video game over cultural disrespect of shrine. South China Morning Post, 20 March. Available at: https://www.scmp.com/week-asia/politics/article/3303172/japanese-pm-ishiba-condemns-assassins-creed-video-game-over-cultural-disrespect-shrine

Sheridan, C. (2020) Ghost of Tsushima Kurosawa Mode explained: how it changes the game and why. GamesRadar, 10 July. Available at: https://www.gamesradar.com/ghost-of-tsushima-kurosawa-mode-explained-how-it-changes-the-game-and-why/

Shutler, A. (2025) ‘Assassin’s Creed Shadows’ removes controversial feature after backlash from Japan’s prime minister. NME, 20 March. Available at: https://www.nme.com/news/gaming-news/assassins-creed-shadows-day-one-patch-shrines-temples-japan

Spartan The Conqueror (2020) Khotun Khan WAR CRIMES – Burning Enemies ALIVE in Ghost of Tsushima!. YouTube, 15 July. Available at: https://www.youtube.com/watch?v=wZFU1ixfV_

Švelch, J. (2020) Paratextuality in game studies: a theoretical review and citation analysis. Game Studies: The International Journal of Computer Game Research, 20(2). Available at: https://gamestudies.org/2002/articles/jan_svelch

Takahashi, M. (2020) Balancing history and fun in Ghost of Tsushima: an interview with Sucker Punch co-founder Brian Fleming. Polygon, 16 July. Available at: https://www.polygon.com/interviews/2020/7/15/21324263/ghost-of-tsushima-research-interview-mari-takahashi.com

Townsend, V. (2024) Ghost of Tsushima’s Japanese localization producer responds to Assassin’s Creed Shadows. Automaton Media, 22 May. Available at: https://automaton-media.com/en/news/ghost-of-tsushima-japanese-localization-producer-responds-to-assassins-creed-shadows/

Video Games Chronicle (2020) ‘Japan should have made Ghost of Tsushima’, says Yakuza creator: Toshihiro Nagoshi praises Sucker Punch’s Kurosawa homage. 30 July. Available at: https://www.videogameschronicle.com/news/japan-should-have-made-amazing-ghost-of-tsushima-says-yakuza-creator/

Video Games Chronicle (2021) Ghost of Tsushima fans have helped raise $260k for repairs on the real island: shrine thanks fans for helping repair damage. 11 January. Available at: https://www.videogameschronicle.com/news/ghost-of-tsushima-fans-have-helped-raise-260k-for-repairs-on-the-real-island/

Weber, R. (2020) Ghost of Tsushima’s samurai rabbit roots, fox petting, and commitment to authenticity. GamesRadar, 20 May. Available at: https://www.gamesradar.com/ghost-of-tsushima-interview-nate-fox/

Wu, K. (2020) Ghost of Tsushima: Video Game Review. Japan Nakama, 16 September. Available at: https://www.japannakama.co.uk/gaming/ghost-of-tsushima-review/

Read the full story at: https://asia.nikkei.com/Business/Media-Entertainment/Wanted-30-000-animators.-Japan-s-anime-future-at-risk

Read the full article at: https://tinyurl.com/38mhs7y8

In the autumn of 2022, our team launched a website for a digital heritage initiative called the Hashima XR Project. At first glance, the site resembled a conventional prototype showcase: teaser trailers, interface mockups, scripted voice samples. It promised an immersive reconstruction of Gunkanjima—also known as Hashima Island—a coal-mining community turned post-industrial ruin turned UNESCO World Heritage site.

But unlike most digital heritage projects, this one never became playable. Visitors could preview a possible experience, but they could not enter it. The simulation remained incomplete. No build was released. And the virtualized island mining community—rendered in geometry and placeholders—stayed eerily silent.

This was not a technical failure. It was an ethical decision.

The Hashima XR Project set out to document the visual layers of the island and complex cohabitation of laboring populations: Japanese supervisors, Korean colonial subjects, and Korean and Chinese wartime mobilised labor (possibly some Chinese prisoners of war, though this is unconfirmed and requires further research)—all living and working within a space roughly 480 meters by 160 meters (0.063 square kilometers).

As the project progressed, however, institutional pressure mounted not to engage with the topic of non-Japanese labor before 1945. This has been the response of Japanese heritage practitioners since Hashima Island became a UNESCO World Heritage site in 2015, when the South Korean government criticized the island’s designation on the grounds that Korean and Chinese forced laborers were used there prior to and during World War II.

Local heritage partners pushed for simplification. Sponsors discouraged emphasis on colonial or coercive labor systems. A potential key stakeholder proposed a non-disparagement clause that would have rendered large portions of the narrative off-limits. At that point, the question was no longer how to proceed, but whether proceeding was possible without compromise.

We stopped building the game. But we didn’t abandon the project.

From Simulation to Counter-Archive

When we launched the project website, it presented a vivid vision: sample narratives, trailer footage, and preview imagery of what an immersive experience might look like. The prototype itself was minimal—little more than a navigable shell—but it gestured toward an ambitious framework. Visitors encountered a virtual world in the process of formation.

What they could not see was that development had already begun to unravel.

As the narrative scope expanded to include the complex work-life configurations that defined Hashima during its industrial peak—across civilian, colonial, and wartime mobilised labor categories—disagreements emerged. Local collaborators voiced increasing discomfort with the project’s attempt to include a labour dimension due to the professional and social ostracization they were facing. Ultimately, funding lapsed, support dissolved, and the build froze.

We did not choose to stop in protest. We were stopped—gradually, materially—through a series of refusals external to the design itself.

Yet rather than conceal this suspension, we began to reframe it. The project’s unfinished status became an object of analysis in its own right. Were the fragments of the project—the unused scenarios, the narrative logics never coded, the historical questions left unanswered—the archaeological remains of the material history of Hashima Island that has been decoupled from its status as UNESCO heritage site, tourist attraction, and political lighting rod par excellence in Korean–Japanese relations?

Refusal Is a Form of Design

In heritage practice, incompletion is often seen as a failure: a technical hurdle, a funding shortfall, a logistical misfire. But the Hashima XR invites a different reading. Its incompleteness is testimony to a history that has become unrenderable because people, both in Japan and Korea, are not ready to listen to alternative voices.

The remains of our efforts are a fragmented record of competing pressures, withdrawn commitments, and unfinished work. It is the archaeological remains of an attempt to record via XR technology the heterogeneity and multiplicity (plurality) of a social reality that has been decoupled from the Hashima UNESCO heritage site because Japanese narratives under-contextualise the issue of labor while Korean narratives over-contextualise Korean victimhood (See Jie-Hyun Lim, Global Easts. Remembering, Imagining, Mobilizing, 2022).

What Comes Next?

The Hashima XR Project is not over. Its form has changed.

We are now exploring how best to present the project’s unfinished components—not as a playable simulation, but as a curated evidentiary structure. This may take the form of an annotated design dossier, a scholarly article, or a public-facing web archive. Each mode would document not only what we attempted to build, but also what we encountered: institutional constraints, historiographical tensions, and the infrastructural limits of heritage production.

What began as an immersive heritage experience has become something else entirely: a study in obstruction, an index of refusal, and a methodological provocation. We do not offer completion. We offer the trace.

If you want to understand why the island is silent, do not look for what was said. Look for what has been made unspeakable.