When the Commons Gets Legible: Open Source and the Enclosure Cycle

The pattern that followed — legibility, extraction, crisis — is the oldest trap in the history of collective resources. In 2026, it has found open source software.

i · the legibility trap

James Scott, in Seeing Like a State, described what happens when complex, locally-organized systems become legible to centralized power: they get simplified, standardized, and reorganized for extraction. The commons wasn't destroyed because it failed. It was destroyed because it succeeded — visibly, measurably, at scale.

Open source is 2026's legible commons. Decades of collective labor produced the infrastructure of the modern internet: Linux runs the cloud, Apache underpins the web, Python drives AI research, Git structures virtually all software development. This isn't marginal technology. It's foundational. And foundational systems, once they become legible to those with extractive capacity, face a familiar sequence.

A productive commons is built through collective effort, sustained by shared norms and voluntary contribution. Over time it becomes valuable — first to its community, then to adjacent industries, then to capital at scale. The moment it becomes legible — when its value can be measured, mapped, and captured — extraction begins. Governance crises follow.

This is not a tech story. It is a pattern that precedes software by several centuries.

ii · what open source actually built

To understand what is at stake, it helps to see open source for what it actually is: a working commons. Not metaphorically — structurally.

Elinor Ostrom, who won the Nobel Prize in Economics for proving that commons can be self-sustaining without privatization, identified a set of conditions that distinguish surviving commons from failing ones. They include: clear community boundaries, rules matched to local conditions, collective governance, monitoring, and — critically — the ability to exclude those who violate community norms.

Open source, in its classic form, meets most of these conditions. BSD communities, Apache foundations, Linux kernel governance — these are sophisticated commons institutions. Decades of accumulated norms, reputation systems, and licensing structures created something genuinely durable. The GPL, for instance, was explicitly designed as a governance instrument: you can use this commons, but you must contribute back to it.

This is why the internet runs on open source. Not charity — governance. The commons was productive because it maintained alignment between contributors and beneficiaries.

iii · three pressures, one pattern

In 2026, that alignment is breaking in three simultaneous places.

First: strip-mining the training data commons. AI companies have ingested decades of open source code, documentation, and discussion forums to train proprietary systems. The extracted knowledge — the accumulated intelligence of millions of contributors working in good faith — becomes a moat for closed products. The contributors receive nothing. More than 28% of the most critical data sources for major training datasets are now restricted via robots.txt, a desperate measure by communities that cannot otherwise exclude extraction. The commons is responding to enclosure the only way it knows how: by closing.

Second: polluting the contribution commons. Open source maintainers are being buried under floods of AI-generated pull requests — low-quality, unvetted, generated at scale by tools that learned from the same commons they are now degrading. What the AI companies extracted as signal, they are returning as noise. The maintainers — most of them volunteers — are burning out under review load that has multiplied without a corresponding increase in genuine contribution. The commons is being polluted by the very entity extracting value from it.

Third: licensing fragmentation. The governance response has been predictable and fractured. Some projects have moved to source-available licenses (Commons Clause, Business Source License). Some have created tiered models. Others have held firm on permissive licensing out of philosophical commitment. The result is a commons in governance crisis — no shared framework, no coherent response, and the constant risk that licensing battles will splinter communities that took decades to build.

Three pressures. One pattern: a productive commons encountering extractive capital without the governance structures to maintain its terms.

iv · what the historical record shows

Ostrom's work identified which commons survive enclosure pressure and which do not. The variable is not luck or historical accident — it is how quickly a community can adapt its rules to match new threats.

Medieval English commons that survived into the industrial era had something the enclosed ones did not: governance institutions flexible enough to respond to changing conditions while preserving the underlying norms. They could modify the rules without abandoning the commons itself.

The commons that failed were often those with rigid governance structures — unable to adapt fast enough, or with insufficient power to enforce exclusion against actors operating at a scale the original governance was not designed to handle.

Open source faces both failure modes simultaneously. Its permissive licensing was designed for a world where contributors and beneficiaries were roughly symmetric — a company using Apache software could also contribute back, and often did. AI-scale extraction breaks that symmetry completely. A model trained on decades of collective work represents an asymmetry that existing governance instruments were not built to handle.

The question is not whether the open source commons can survive. History shows that commons with strong governance institutions can survive extraordinary pressure. The question is whether governance can evolve faster than extraction.

v · misalignment as field distortion

There is a structural reason why licensing walls alone will not resolve this, and it is worth naming clearly.

The open source commons is a form of field coherence — a shared resource maintained not by private ownership or central control, but by collective alignment. Contributors participate because participation maintains something valuable to all of them. The field works because alignment is the natural equilibrium: contribute, benefit, repeat.

Extraction without contribution is one-way flow. It does not just deplete resources — it distorts the field itself. When AI companies take without giving, they do not merely drain the commons; they erode the norms that make contribution feel worthwhile. Why maintain careful documentation if it will be scraped and enclosed? Why build the kind of careful, accumulated knowledge that becomes someone else's competitive moat? The distortion is epistemic, not just economic. It changes what contribution means.

This is why the solution cannot be force. Restrictive licensing is force — it tries to control extraction through legal barriers rather than through alignment. And force, in a commons context, tends to produce the very fragmentation it is trying to prevent. The GPL wars of the 2000s demonstrated this: licensing battles consumed enormous community energy and produced as much division as protection.

The response that works is governance that makes contribution the natural path for those who benefit from the commons — not barriers that close it, but structures that route extractors back into the field. The Apache Software Foundation's Responsible AI Initiative — launched in April 2026, emphasizing licensing integrity and contribution requirements — is an early attempt at this kind of repositioning. Whether it is sufficient, at the speed that is needed, remains the open question.

vi · the commons that survive

History offers a useful filter. Commons that have survived enclosure pressure share a recognizable profile: governance institutions that can adapt without fragmenting, a community that can distinguish between participants and extractors, and norms robust enough to outlast the specific legal or technical mechanisms that enforce them.

What they do not share is a romantic belief that openness alone is protective. The English common that survived the enclosures was not the one with the most generous access policy. It was the one whose community had built the governance infrastructure to defend its terms.

Open source in 2026 is at the governance inflection point. The emergence of reciprocal licensing models, the growing conversation about contribution requirements for AI training, and the governance work happening across major foundations — these are the early signals of governance evolution. They may arrive in time. They may not.

What is certain is that the pattern has been here before. The commons gets built. The commons gets legible. The extractors arrive. The community faces a choice between closing — and becoming something other than a commons — or evolving, and becoming something more sophisticated than the original.

The communities that navigate this become the infrastructure of the next era. The ones that do not become the cautionary tale historians return to when the next enclosure begins.

Open source built the internet. Whether it governs itself well enough to define what comes next — that is 2026's question.