Chapter 18 The Moral in the Machine (circa 2024-Present) (c. 2024)

The world had gotten the infrastructure it needed, and in doing so, it inherited the infrastructure’s true, uncalculated bill—a bill not for the code, which was free, but for the planet upon which all the code ran. That physical reckoning was the final, ambient pressure. But the bill was not only ecological; it was ethical, a spectral debt incurred by a moral crusade that had won everything it set out to win, only to find its own principles returning in forms it could scarcely recognize. In a Washington, D.C. Hearing room in 2024, the lights shone down on a panel of expert witnesses, their faces composed for the C-SPAN cameras. A United States congressional subcommittee was examining allegations of racial and gender bias in the algorithmic systems used for hiring, loan approvals, and tenant screening. Legislators listened, pens poised, as a computer scientist explained that the proprietary models in question were opaque, their decision-making logic hidden behind corporate firewalls and trade secret claims.

A civil rights attorney argued that without the ability to audit, to study, and to understand these systems, there could be no meaningful accountability for their discriminatory outcomes. The language hung in the air: transparency, auditability, user agency, the right to challenge automated decisions. No one mentioned free software. No one cited a license. The debate was about black-box algorithms governing economic life, not about the freedom to modify a compiler. Yet the ethical imperative being invoked—the demand for legibility, for user sovereignty against an opaque, powerful system—was an echo, faint but unmistakable, of a declaration made four decades earlier in a different world. That declaration was not a law or a policy white paper. It was a preamble to a software license and a set of philosophical tenets. Richard Stallman had argued that software users should have the freedom to share with their neighbors and be able to study and make changes to the software that they use. He maintained that attempts by proprietary software vendors to prohibit these acts were antisocial and unethical. He argued that freedom was vital for the sake of users and society as a moral value, not merely for pragmatic reasons such as possibly developing technically superior software.

He argued that freedom was vital for the sake of users and society as a moral value, not merely for pragmatic reasons. This was the core of the crusade: a belief that user agency over the tools of computation was a fundamental right, essential to a just society. In the congressional hearing, the tools were different—complex AI models instead of text editors, their outputs determining life opportunities rather than document formatting—but the core grievance was the same: a user, a citizen, rendered powerless before a system they could not see, understand, or alter. The moral crusade was not present in the room as a philosophy being debated. It was present as a ghost, having supplied the very vocabulary of ethical complaint now being deployed in a fight its originators never imagined. The hearing was one node in a global pattern. The movement’s revolutionary ideology—its clear, demanding vision of software freedom as a moral imperative—had been subsumed by its own staggering practical success. Open source was the infrastructure of the world.

This victory, however, was not the victory of the original idea in its pure form. It was a victory of the idea translated, diluted, and pragmatically adopted. The fervent moral rhetoric that once animated mailing-list wars was sidelined in favor of arguments about reliability, cost, and innovation speed. The movement’s revolutionary energy dissipated into the mundane work of maintaining the planet’s digital plumbing. Yet the foundational ethical questions Stallman raised did not vanish. They could not vanish, because they were questions about power, agency, and control in a technological society. Having won the technical battle to become the world’s substrate, open source now faced the consequences of its triumph: its own orphaned ethics returned to haunt every major controversy of the digital age, not as a coherent doctrine but as a series of unresolved and haunting tensions. The future of the open source software community, and the free software community by extension, had become successful if not confused about what it stood for.

The most direct lineage runs from the freedom to “study” software to the modern crisis of algorithmic opacity. Stallman’s second freedom—the freedom to study how the program works, and change it so it does your computing as you wish—was a technical prescription for source code availability. In an era of artificial intelligence, this demand escalates into a societal confrontation with inscrutability. The controversy surrounding GitHub Copilot, the AI-powered code completion tool launched by Microsoft-owned GitHub, laid this tension bare. Trained on a vast corpus of public source code from GitHub repositories, Copilot generated suggestions, functions, and even whole code blocks for developers. Its arrival provoked immediate, profound unease. Was it a tool or a parasite? User requests to block Copilot features surged to among the most popular on GitHub’s feedback forum. The debate was not about the GNU GPL’s copyleft provisions in a formal sense, but about something more fundamental: when the act of software generation itself becomes automated, non-transparent, and derived from a commons of code, what do the old freedoms mean?

A developer using Copilot could not “study” how it arrived at its suggestions; its neural networks were a black box. The user’s agency was reduced to accepting or rejecting an output whose provenance and logic were obscured. The ethical concern was identical to Stallman’s—the user’s right to understand their tool—but the technical context had transformed utterly, migrating from the clear text of source code to the latent space of a machine learning model. This tension did not remain contained within the world of developers. It exploded into the open-source AI model licensing wars of 2023 and 2024, a global conflict that revealed the frayed edges of the movement’s own definitions. As large technology firms like Meta released powerful large language models such as LLaMA under custom “open” licenses that restricted commercial use or required special permissions, the community fractured along a familiar fault line. What constituted an “open source” AI model?

Was it merely the public release of the model’s weights—the numerical parameters—or did true openness require the full disclosure of training code, data, and the entire toolchain for study and modification? The debate was a stark recapitulation of the old schism between Free Software’s moral emphasis on freedom and Open Source’s pragmatic emphasis on practical benefits, but now operating at a scale of complexity and potential consequence that dwarfed the earlier battles over kernel modules or text editors. The diluted ethical language of open source, stripped of Stallman’s rigorous moral framework, became the default, inadequate framework for a struggle it could not contain. The question was no longer merely about the freedom to modify a program, but about the freedom to audit, understand, and govern systems that might shape public discourse, legal outcomes, and scientific discovery. The ghost of the moral crusade whispered that user agency mattered, but the movement’s own success had created a world where the very definition of “user” and “agency” was being rewritten by technologies its licenses were never designed to address.

The conflict escalated from technical mailing lists to the floors of parliaments and the pages of international treaties, as policymakers grappled with how to regulate systems whose inner workings were, by design or by complexity, closed to the very people they judged, hired, or diagnosed. A second lineage runs from the viral protection of freedom in licenses like the GPL to the contemporary geopolitical struggles over data sovereignty and platform governance. The GPL’s ingenious copyleft mechanism was a legal innovation designed to ensure that freedoms granted to one user would be preserved for all downstream users. It was a tool to prevent the enclosure of the software commons, to make freedom propagate like a beneficial contagion. This idea of virally protecting agency against corporate or state enclosure now manifests on a global stage in debates over data localization, digital sovereignty, and the regulation of dominant platforms. The European Union’s Digital Markets Act (DMA), for instance, seeks to compel the largest “gatekeeper” platforms to open their services to interoperability, to refrain from self-preferencing, and to allow users more data portability.

The motivations are complex—cybersecurity, market fairness, democratic oversight—but threaded through them is a diluted version of the old anti-enclosure logic: the user, and by extension the polity, must retain some measure of control and choice against opaque, centralized systems of power. This digital sovereignty movement found a stark counterpart in the actions of nations seeking technological independence from foreign suppliers. Confrontations over the use of proprietary software from geopolitical adversaries in critical infrastructure, national mandates for government use of open-source solutions, and stringent laws requiring that citizen data be stored within national borders—all these are modern, geopolitical expressions of a desire for agency and transparency that resonates, however distantly, with the free software ideal. The enemy is no longer merely a proprietary software vendor, but a potentially hostile foreign power or an unaccountable transnational corporation. The desired freedom is not necessarily the four freedoms of an individual user, but a form of national or collective self-determination in the digital sphere.

The original moral argument—that control over software is a prerequisite for a free society—has been translated, and perhaps distorted, into a principle of statecraft and international competition. The ghost of the crusade is present in the assertion that technological dependency is a form of vulnerability, that opacity is a threat, and that some form of “openness” is a strategic imperative. It is a far cry from Stallman’s vision of individual hacker autonomy, yet it is undeniably its descendant, shaped by the harsh realities of a world where the infrastructure of daily life is built on software whose ultimate control and data flows are matters of national security. The open methodology, once a tool for individual empowerment, had become a weapon in a new kind of cold war, its ethical core repurposed for aims of sovereignty and control that its founders had sought to resist. Nowhere is the haunting more material, more grimly literal, than in the third tension: the collision between the freedom to build and the environmental cost of building at a global scale.

The previous chapter established the uncalculated planetary bill of the digital infrastructure—the energy consumption of data centers, the mineral extraction for hardware, the electronic waste. This is the ultimate, ironic inheritance of open source’s victory. The movement succeeded in making powerful software tools universally accessible and modifiable. It empowered millions to build, to deploy, to scale. But this very empowerment, when multiplied across billions of users and devices, created a metabolic regime of staggering energy and resource demand. The freedom to run the software at any scale, without constraint, now meant the freedom to consume on a planetary scale. The haunting was no longer merely ideological; it was thermodynamic, a physical law reasserting itself against the dream of limitless digital abundance. A 2024 report from a major cloud provider, detailing the carbon footprint of training a single large language model, would read like a perverse fulfillment of the free software promise. The model itself, released under an “open” license, could be studied, modified, and redistributed by anyone. Its training code might be public.

Yet the energy required to execute that freedom—to train a derivative model, to run an inference at scale—was equivalent to the annual emissions of a small town. The community that had once debated license compatibility on mailing lists now found itself uneasily discussing joules per parameter and the ethics of computational profligacy. The movement’s triumph, the universal availability of powerful tools, had created a predicament its founders never envisioned: an ecological crisis born of uncompensated abundance. The question shifted from “Is this software free?” to “Can we afford the freedom this software enables?” The moral imperative of sharing had crashed into the physical limits of the shared planet. The open source ethos, which celebrated proliferation and reuse, now had to confront the downstream costs of that proliferation, costs that its legal and moral frameworks had spectacularly externalized. The infrastructure was free, but the planet was not. This constellation of tensions—transparency versus opacity in AI, sovereignty versus dependency in data, empowerment versus sustainability in computation—forms the diffuse afterlife of the free software idea.

It is no longer a movement with a manifesto, but a pervasive, often unacknowledged ethical grammar applied to the crises of a digital society. The original, coherent philosophy has shattered into fragments, each embedded in a different domain of conflict. To trace the lineage through these fragments is to see how the moral in the machine persists, not as a guiding star but as a ghost in the circuitry, a set of unresolved questions that the infrastructure itself keeps asking. The spectral presence becomes clearest when opposing parties in a modern dispute, seemingly speaking different languages, unconsciously converge on a framework forged decades earlier. Consider the protracted, global debate over “right to repair” legislation. On one side, coalitions of farmers, consumers, and independent mechanics lobby for laws forcing manufacturers to make diagnostic tools, spare parts, and repair manuals available. They argue that ownership of a physical device must include the right to maintain and modify it. On the other side, technology manufacturers cite safety, security, and intellectual property to restrict access.

The argument is about tractors, smartphones, and medical devices, not about software licenses. Yet the core principle—that the owner of a tool should have the right to understand, fix, and alter it—is a direct translation of Stallman’s freedoms into the realm of embodied machinery. The movement’s logic had escaped the computer and infected the material world. Similarly, the prioritization of convenience over freedom in consumer software, exemplified by features within Ubuntu that track users for marketing purposes, demonstrated how the open source infrastructure could host behaviors antithetical to its founding spirit. The platform remained open, but the user’s experience within it could be engineered toward passivity and surveillance. The moral question of agency had not been solved by the technical fact of source code availability; it had simply morphed, finding new venues and new forms of enclosure.

Yet they are all, in a profound sense, heirs to the crusade that began with a stubborn refusal to be locked out of a printer driver. They represent the idea’s journey from a specific technical grievance to a universal language for questioning technological power. The movement did not achieve its goal of a world built entirely on free software, but it succeeded in making the ethics of openness, transparency, and user agency the default vocabulary for critiquing a world built on software. This translation was neither a triumph nor a betrayal, but an inevitable diffusion. Once an idea begins to circulate at the scale of an infrastructure, it no longer fully belongs to its originators. It is picked up, twisted, repurposed, simplified, and deployed in contexts they could not foresee. The human choices that mattered were no longer just those of hackers writing licenses, but of legislators drafting digital sovereignty acts, of engineers designing opaque AI systems, of corporations deciding how much tracking to embed in an open-source distribution, of communities weighing computational power against carbon budgets.

The human choices that mattered were no longer just those of hackers writing licenses, but of legislators drafting digital sovereignty acts, of engineers designing opaque AI systems, of corporations deciding how much tracking to embed in an open-source distribution, of communities weighing computational power against carbon budgets. The institutional design of the internet, built on open protocols and later on open-source software, created the conditions for these choices, but it did not dictate them. The path was shaped at every turn by capacity, by interest, and by conflict. Eric S. Raymond, as head of the Open Source Initiative, had argued that advocates should focus on the potential for better products. He viewed the “very seductive” moral and ethical rhetoric of the free software movement as a liability for persuasion, arguing it failed “not because his principles are wrong, but because that kind of language… simply does not persuade anybody”. His pragmatic turn was instrumental in open source’s commercial adoption. Yet decades later, as the products of this pragmatic success—the AI models, the platforms, the clouds—raised profound ethical dilemmas about bias, power, and planetary health, the adequacy of that pragmatism was itself called into question. The moral rhetoric had been seductive because it pointed to something real: that software was never merely a product, but a conduit for human values and a shaper of human fate.

Having cast off the moral framework to win the world, the open source world now found itself lacking the very language it needed to govern what it had built. The ghost in the machine was the memory of that moral clarity, returning not as a solution but as a reminder of a debt unpaid. The final judgment on the idea’s journey lay in this transformation. It did not end in victory or defeat, but in a state of profound, ambient ubiquity. The cathedral—the centralized, hierarchical model of software development—had not fallen. Instead, its methods had been absorbed and inverted by the bazaar of open collaboration. The commit—the unit of distributed, incremental work—had become the foundational act of digital civilization. From Stallman’s printer driver to Microsoft owning GitHub, the moral crusade about software freedom became the infrastructure of the digital world. In doing so, it ceased to be a crusade and became a condition, a set of inherited questions embedded in the circuitry of daily life.

The infrastructure hummed, and the hum contained, like a persistent low note, the unresolved echo of its own founding demand: Whom does this machine serve? The question was no longer confined to licenses and mailing-list wars. It was now asked in every courtroom debating algorithmic bias, in every parliament legislating data borders, in every calculation of a data center’s carbon toll. The idea had traveled from the periphery to the center, and in occupying the center, it had bequeathed its central, haunting tension to the age: the world ran on free software, but the world was not free. The machinery was open, but the consequences of its operation were the shared, contested, and unfinished business of everyone who relied on its light.