@mc5punk 24 November 2025

Short story: Lúmen in Silence (Chapters VII and VIII)

Chapter VII — Ambivalence, Surveillance, and Markets The technological transformation that marked the early decades of the century was not merely a sequence of technical advances; it left in its wake a complex web of social, political, and economic effects that revealed, starkly, the ambivalence of innovation. Regulatory reforms and public programs designed to mitigate risks did not fully eliminate a persistent problem: mass surveillance, now renewed and made sophisticated by seemingly benign digital tools. On the one hand, the widespread incorporation of smart cameras, urban sensors, and predictive behavior models produced tangible benefits. In numerous neighborhoods—especially those with precarious infrastructure or high crime rates—the combination of computer vision, temporal analysis, and predictive algorithms made it possible to anticipate crime hotspots and improve the allocation of patrols and social resources. Official statistics showed, for several years, marked declines in certain street crimes and faster incident resolution. For many citizens these technologies meant a sense of security and a perception—sometimes justified—of greater state effectiveness in everyday protection. But that progress came with invisible costs and subtle modes of control that reshaped public life. Cameras with facial recognition and behavioral scoring systems, integrated with private and public databases, opened channels for ubiquitous oversight that operated more at the margins than with ostentatious displays of power. The probabilistic nature of models—their biases, errors, and opacities—introduced dynamics of exclusion: already vulnerable groups found themselves labeled by risk patterns, with real consequences for access to employment, housing, and services. Scandals erupted when it became clear the phenomenon was not merely state-driven. Private behavioral analytics firms began commercializing extremely detailed population segments: from purchasing patterns and mobility trajectories to inferences about political beliefs or health conditions. These microsegmentations, sold to both advertisers and governments with authoritarian agendas, multiplied the possibilities for manipulation, discrimination, and repression. Media outlets and civil society organizations published contracts and data flows showing, for example, how political campaigns tailored messages to microaudiences to polarize or mobilize; or how certain regimes contracted predictive models to anticipate and neutralize dissent. Regulatory reaction was swift. The Algorithmic Supervisory Authority (ASA)—created in response to the combination of private power and technical opacity—had to intervene with a battery of strict measures: limits on the retention and sharing of sensitive data, independent model audits, partial bans on certain uses of biometric recognition and, crucially, requirements for transparency in the marketing chains of population analytics. These measures sought to rein in opaque markets and restore basic safeguards, but they faced legal challenges and corporate resistance that exposed the fragility of the regulatory framework against transnational actors with the resources to evade controls. Citizen pressure played a decisive role. Social movements, coalitions of academics and journalists, and digital rights platforms demanded clauses that went beyond mere technical transparency: they called for open access to the models that affected fundamental rights—for example, tools used in administrative sanctioning or employment selection—so that they could be audited, replicated, and corrected by the scientific community and rights defenders. That demand for “open access” was articulated as an ethical and political requirement: if an algorithmic decision can condition a person’s life, its logic could not remain encapsulated in a corporate lab. In some countries these pressures became law: public model repositories, documentation requirements, and obligations to provide human appeal mechanisms against automated decisions. In electoral politics new ideological and programmatic splits emerged. Movements that called themselves defenders of the “ethics of abundance” argued that automation should not concentrate gains in the hands of a few but should be redistributed to guarantee universal welfare. Their proposals combined progressive automation taxes, basic income or compensation for job loss, public education oriented to new skills, and democratic, citizen control over essential infrastructures—from data networks to public service platforms. These groups maintained that technology should expand rights and leisure time, not deepen precarity. Opposing them, libertarian and pro-market sectors doubled down on defending broad entrepreneurial freedom: they argued that international competitiveness and innovation capacity depended on flexible markets, light regulation, and strong intellectual property rights. For these actors, strict rules on AI deployment or limits on the trade of models were brakes on development that would drive away capital and delay the adoption of beneficial technologies. The disputes were not merely doctrinal: they played out in election campaigns where the central axis ceased to be only the traditional economy and instead became how to distribute free time (shorter workweeks enabled by automation), how to manage personal and collective data, and who would govern the large technology platforms. Promises of guaranteed employment, greater leisure, or unfettered markets became contrasting arguments in debates, rallies, and legislative proposals. At the international level, the race for leadership in artificial intelligence added another layer of complexity. States and regional blocs pursued divergent strategies: some opted for rigorous regulatory frameworks with strong export controls on models, ethical standards, and human-rights safeguards; others offered more permissive environments and accelerated approvals so corporations could deploy technologies at scale. The resulting competition spawned strategic alliances between governments and firms, investment flows directed toward “friendly” jurisdictions, and diplomatic tensions over security, technology transfer, and digital sovereignty. Global AI governance, far from progressing uniformly, advanced in fits and starts. There were partial agreements—sectoral protocols for health care or transport, interoperability standards, and governance frameworks for specific military applications—but the absence of a universal normative body produced divergent regulations that companies and states learned to exploit. Regulatory fragmentation fostered, on one hand, a race in which jurisdictions competed to attract technological capital by offering advantageous regimes; on the other, it complicated the coordination of common responses to transnational risks such as the spread of authoritarian surveillance tools or the outsourcing of critical services to foreign providers. In sum, the landscape left by the expansion of AI and automation was ambiguous: concrete advances in safety and efficiency coexisted with new forms of control and concentrations of power. Regulatory and civic responses managed to curb some excesses and open deliberative spaces, but they also revealed structural limits—the speed of markets, resource asymmetries, and the difficulty of crafting global rules—that kept alive the tension between the promise of technological abundance and the dangers of ever more invisible and distributed power. Chapter VIII — The Mesh and Its Ethics At the heart of the Aurora Center, community life found its epicenter in a network of practical initiatives that, without fanfare, redefined the meaning of the public. The “mesh” —the term the community adopted to name the web of technical, normative, and affective collaboration— was not a didactic metaphor but a project built day by day: meetings, whiteboards full of diagrams, midday coding sessions, and evening workshops where both security protocols and poems about machines were discussed. The first layer of the mesh took the form of collective learning. Algorithmic literacy courses were organized especially for older people, who had been the most disoriented by automated decisions affecting health, pensions, or access to services. The classes were deliberately practical: it was not enough to explain what a neural network was; they showed concrete examples —why a medical priority order could change according to a parameter, how to identify bias in an access form, how to read minimal model documentation— and, above all, how to lodge appeals. Sessions combined analog activities —paper flowcharts, role-playing administrative hearings— with exercises at screens where decisions were simulated and correction paths traced. At the same time, robotic repair workshops became multifaceted spaces: repairing domestic assistants, disassembling community drones to understand their sensors, and building prototypes that prioritized transparency over maximal efficiency. Those workshops taught not only how to fix hardware but how to interrogate firmware and social firmware: what data a device collects, how often, who has access, and under what conditions. The idea was that the capacity to repair should also be the capacity for democratic control over infrastructures that would otherwise remain sealed in devices and incomprehensible terms of service. The citizen laboratories, for their part, did not replicate the academic lab model or that of private startups. They were hybrid spaces where models were trained according to explicit criteria of equity and accountability. Projects prioritized carefully curated datasets to avoid reproducing historical discrimination; preprocessing choices were documented and model versions recorded to enable public audits. Irene —coordinator of one such laboratory— spearheaded the creation of a public archive of local models: a library of algorithms with clear, accessible documentation not only for technicians but for any neighbor. The repository included everything from the model’s purpose and scope to the fairness metrics used for evaluation and instructions for filing a human appeal when an automated decision affected health, social benefits, or housing access. At a weekly assembly, Noa spoke up to correct misunderstandings among those who saw technology either as an enemy or as a panacea. “It’s not about sabotaging technology,” she said calmly. “It’s about subjecting it to collective deliberation: deciding which values we want it to encode.” The phrase resonated because it distilled a practical ethic: technology was not neutral, but neither was it immutable; it should be the result of democratic processes where priorities between safety, privacy, equity, and shared well-being were negotiated. Marcos, whose deep voice had gained a reputation for cutting through unproductive bureaucracy without losing patience, added a technically grounded political point: “And learning to build redundancies. Any system that concentrates can fail catastrophically. Decentralization is not romanticism; it is security.” Marcos explained with examples: replicating critical services in community nodes, designing alternative routes for data provision, and maintaining verified copies of models and datasets to restore services when a centralized failure occurred. The proposal was both technical and organizational: institutionalize guard committees, emergency protocols, and regular resilience drills so the mesh would not depend on a single point of failure. Kim introduced an intergenerational dimension: formal education as a vector for technological democratization. She proposed incorporating school modules that teach girls and boys to audit models, interpret datasets, and exercise empathy toward automated decisions. Those modules included age‑adapted activities: games illustrating bias mechanics, exercises to build community datasets, debates on when an algorithmic decision merits human intervention, and labs that simulated appeals and repair processes. The proposal prompted an ambivalent response from attendees. There was enthusiasm for empowering new generations with critical tools —the conviction that informed citizens would be less vulnerable to manipulation and better able to participate in public decisions— but also concern about the project’s material sustainability. Maintaining resources —tutors, equipment, physical spaces, and access to clean data— required funding and institutional stability that were not always guaranteed. They discussed mixed funding models —public grants conditional on community governance, sponsorships without ceding control, neighborhood micro‑memberships— and agreed any source should be subject to strict independence and transparency criteria. The mesh, overall, responded to a dual challenge: how to make technology serve daily life without reproducing hierarchies, and how to convert technical knowledge into democratic practices. This also required rethinking notions of expertise. Laboratories adopted co‑production formats: technical experts worked alongside social workers, lawyers, educators, and neighborhood representatives. Project evaluation criteria now included technical indicators (accuracy, robustness) and socio‑political metrics (differential impact across groups, documentation accessibility, existence of appeal channels). Operationally, concrete guiding principles were defined for the mesh: responsible openness (access to models with privacy safeguards), proportionality in application (avoid automation where the cost of error is high), periodic public accountability (community reports on deployments), and reversibility (mechanisms to undo harmful applications). Cultural practices were also implemented: review rituals —quarterly public sessions to evaluate projects and decide continuations— and mentorship policies to incorporate new participants without reproducing hierarchies. Tensions were inevitable. Some local companies, pressured by commercial interests, tried to influence priorities; lab members disagreed over when a model should be published openly and when access should be restricted due to risks. Those discussions clarified ethical boundaries: the principle of transparency had to coexist with prudence regarding potential malicious uses. The mesh learned to distinguish between total openness and responsible openness, defining criteria for differentiated publication according to risks and possible harms. Over time, the Aurora Center became a reference point: a node of the mesh demonstrating that technological democracy is not a slogan but a chain of practices. Its projects inspired replications in other neighborhoods: local repositories, adapted school programs, repair workshops, and community governance agreements for digital infrastructures. Some of those nodes formed a wider network sharing standards, tools, and emergency protocols. Ultimately, the mesh represented a lived ethic: a pragmatic proposal for technology to be subject to deliberation, repaired when it failed, limited when it harmed, and always linked to the construction of collective capacities. It promised no definitive solutions but offered a conceivable strategy: design institutions and practices that, with humility and rigor, keep technology in the service of the community and of the values the community chooses. See previous chapters: https://read.cash/@mc5punk/short-story-lumen-in-silence-chapters-v-and-vi-19bdcfe7 Source of the images. Image created with Bing.