A Different System is Not an Alternative

Alternatives can save lives. They can also become research and development for the world they were built against.

There is a tendency in radical technical culture to treat the existence of an alternative as evidence of political movement. A platform becomes hostile, so people build a different platform. A protocol becomes captured, so people write a better protocol. A corporation extracts from users, so someone proposes a cooperative version. A state abandons care, so people organize care elsewhere. A logistics system fails, so people create informal distribution. A system makes life unlivable, so people build something that lets them live anyway. On the surface, this looks like refusal. It looks like exit, or at least the beginning of exit. But much of what gets called an alternative is not outside the system it opposes. It is a difference presented in a form the system already knows how to process.

An alternative is not simply another way of doing something. It is another way of doing something made available as an option within an existing field of selection. That distinction matters. A practice can be different in its relations, obligations, memory, locality, danger, and purpose, but once it is rendered as an option it enters a comparative environment: it can be evaluated, benchmarked, adopted, funded, forked, standardized, regulated, scaled, dismissed, or absorbed. The old system does not need to share the values of the alternative. It only needs the alternative to become sufficiently legible that some part of it can be used.

That is the first claim: the problem is not “alternatives get captured later.” The problem begins earlier, in the conversion of relation into option. Capture is not merely what happens when a radical project becomes popular, or receives funding, or becomes institutional. Capture begins when a system built from specific relations is translated into a selectable form that can travel without those relations. A community practice becomes a model. A survival tactic becomes a toolkit. A refusal becomes a governance framework. A shared obligation becomes a protocol. A world becomes a use case.

This does not mean alternatives are useless. That would be a stupid conclusion, and usually a comfortable one. Folk need food, medicine, transportation, shelter, care, money, tools, information, and each other. People survive because someone made something else available when the official thing failed or turned dangerous. A thing can be partial, compromised, temporary, vulnerable to capture, and still save someone’s life. But survival is not the same as exit, and usefulness is not the same as difference. A thing can work and still become available to the world it was built against.

The sharper distinction is not between good alternatives and bad alternatives. It is between alternatives and different systems. An alternative is a selectable substitute inside an existing environment. A different system may require an environment the existing system cannot host without changing what the thing is. This means the old question—“is this a better option?”—is already part of the trap. The better question is: what relations make this possible, and what would have to be stripped away for it to become an option?

1: Difference Becomes Signal

Cybernetics gives us a cleaner way to see the trap, which is annoying, because cybernetics gives everyone a cleaner way to see the trap and then most people immediately build a dashboard. Norbert Wiener’s Cybernetics: Or Control and Communication in the Animal and the Machine (1948) and The Human Use of Human Beings (1950) understood societies, machines, and organisms through feedback: communication and control as linked processes, where systems receive signals, compare them to expected states, and adjust. Claude E. Shannon’s “A Mathematical Theory of Communication” (1948) treated information not as meaning, but as statistical difference: uncertainty, deviation, entropy, the measurable departure from expectation. W. Ross Ashby’s An Introduction to Cybernetics (1956) then gives the political edge through the law of requisite variety: a regulator can only control a system if it has enough variety to match the disturbances it must absorb. Stafford Beer’s Brain of the Firm (1972), later developed through The Heart of Enterprise (1979) and Diagnosing the System for Organizations (1985), extends this into organizational life, where viability depends not on perfect command but on the ability to reorganize across operational, coordinating, intelligence, and policy functions fast enough to remain coherent under changing conditions.

In that frame, alternatives are not just moral projects. They are generators of variety. They test behaviors the dominant system has not yet internalized. They reveal unmet demand, social tolerance, governance failure points, enforcement gaps, moderation patterns, labor requirements, trust dependencies, and forms of language that stabilize participation. They produce information in the cybernetic sense: difference that can be detected, processed, and routed. Even when they are built against the dominant system, they can increase the dominant system’s ability to regulate its environment by expanding what it knows how to recognize.

This is why the dominant system does not need to destroy every alternative. Destruction is only one response. Often, the more efficient response is observation. Let the alternative run. Let it discover the edge cases. Let it absorb the weird early labor. Let it produce skilled maintainers, social norms, technical documentation, conflict histories, failure modes, scaling debates, user expectations, and legitimacy. Once that material exists, the system does not need to reproduce the whole relation. It can extract the useful pattern. It can take the operational knowledge and leave the obligation behind.

This is not conspiracy. It is systems behavior. A viable system does not only defend itself by blocking disturbance; it also learns from disturbance. It expands its repertoire. It increases its variety. It uses difference as training. The alternative becomes a distributed sensor array for the world it opposes, not because its builders intended that, but because legible difference is metabolizable difference. The system does not need to understand why people built the alternative. It needs to understand what worked.

This is where the phrase “research and development for power” becomes more literal than rhetorical. An alternative can function as unpaid R&D for the dominant order. It can test arrangements that the dominant order could not invent from within its own constraints. It can generate solutions to problems the dominant order has produced but cannot yet solve. It can demonstrate that people will accept, desire, or maintain a new form. It can create the appearance of inevitability around a function before power decides how to own it.

The point is not that alternatives are secretly collaboration. The point is that the form of the alternative often makes collaboration structurally available, whether or not anyone consents to it. If the work becomes legible as an option, then it becomes available to systems organized around selection. And once it becomes available to selection, the old system can ask the question it is best at asking: how can this be made useful here?

“Here” meaning inside the world that already exists.

2: The Option-Form

The option-form is what happens when a relation is transformed into a selectable object. It is not identical to the commodity-form, though commodities are one of its common outcomes. A commodity is organized around exchange. An option is organized around comparability. It does not need to be sold yet. It only needs to be placed into a field where it can be compared to other things as a possible choice.

This is why the option-form is so compatible with liberal governance, markets, platforms, standards bodies, philanthropic funding, academic research, and technical documentation. All of these institutions know how to handle options. They can make rubrics, criteria, scorecards, procurement processes, grant categories, adoption guides, benchmarks, safety evaluations, interoperability layers, policy recommendations, and best-practice documents. They can turn difference into a thing with features. They can ask whether it scales. They can ask whether it is accessible. They can ask whether it is neutral. They can ask whether anyone can use it. They can ask whether it can be adopted without changing the people adopting it.

Those questions are not always wrong. They are just not innocent. They belong to an environment in which the capacity to select has become more important than the capacity to remain obligated. Once something becomes an option, the chooser is placed at the center of the scene. The system, platform, public, funder, institution, adopter, or user becomes the one who evaluates. The relation that produced the thing becomes background. It may be acknowledged as origin, context, inspiration, stakeholder input, community knowledge, or cultural specificity, but the center of gravity has moved. The question is no longer “what world does this require?” It is “is this a good option for us?”

James C. Scott’s Seeing Like a State (1998) helps explain why this movement matters. Scott’s account of legibility is not just that states like to see things. It is that states simplify complex, local, embedded arrangements into forms that can be administered: maps, surnames, cadasters, standardized measurements, planned cities, managed forests. Legibility is not a neutral improvement in understanding. It is a transformation of the thing being understood, because the state can only act on what it can render into its own administrative categories.

The option-form is legibility under the sign of choice. It does not always arrive with police or bureaucracy. Sometimes it arrives as openness, adoption, publication, collaboration, community standards, or accessibility. But the underlying movement is similar: complex relations are simplified into forms that can be evaluated and acted upon from outside. What was embedded becomes movable. What was obligated becomes usable. What was maintained becomes adoptable. What was a relation becomes an interface.

Anna Lowenhaupt Tsing’s “Supply Chains and the Human Condition” (2009) and The Mushroom at the End of the World (2015) give another part of the structure. Tsing argues that contemporary capitalism does not simply homogenize everything it touches; it depends on difference, mobilizing diverse labor, landscapes, histories, and social arrangements through supply chains that can extract value without fully controlling the conditions that produced it. Difference is not outside accumulation. Difference is one of the materials accumulation uses.

This is exactly how many alternatives become useful. They are not absorbed because they are the same. They are absorbed because they are different in a way that can be connected. Their difference becomes a resource once it can be routed into the system’s existing channels. The kitchen remains local until its method becomes a model. The server remains relational until its governance becomes a template. The care practice remains embedded until its language becomes a policy framework. The forest remains a relation until it becomes a case study.

The option-form is not merely a corruption of politics by markets. It is a broader epistemic conversion, a way of making practices available to systems that cannot bear their conditions. It is how an alternative becomes readable enough to be chosen, and therefore readable enough to be stripped.

3: Recuperation Is Not a Later Problem

Recuperation is often narrated as a late-stage tragedy. First the project is radical. Then it grows. Then it becomes visible. Then funders arrive, institutions notice, language softens, compromises accumulate, and eventually the thing becomes a version of what it opposed. That story happens, but it is too slow. It treats capture as an event at the end of a timeline rather than a possibility introduced by form.

Recuperation begins when the useful part of a practice can be separated from the relations that made it meaningful. It begins when the practice can be documented as a model, not because documentation is inherently bad, but because documentation often creates the conditions under which the practice can be evaluated by outsiders as transferable knowledge. It begins when the relation is represented as a technique. The technique may be accurate. It may be useful. It may even be necessary to share. But once technique and relation are separable, the old system can preserve the technique while discarding the relation.

Freedom House Ambulance Service matters because it prevents the argument from becoming lazy. Freedom House was not a symbolic alternative. It was not aesthetic, not lifestyle, not a marginal experiment inflated by theory. It was care under conditions of racist abandonment. Established in Pittsburgh in 1967 through Freedom House Enterprises, a Black-administered organization, and Presbyterian University Hospital, with the involvement of physician Peter Safar and organizer Philip Hallen, it trained unemployed Hill District community members as emergency medical workers and brought paramedic care to an area lacking reliable emergency services. Its training included advanced practices such as intubation and defibrillation before such skills were common in out-of-hospital emergency care, and its training eventually informed national emergency medical technician standards.

That matters immediately, before any theory. People lived because of it. Black neighborhoods in Pittsburgh were being denied adequate emergency medical care, and existing emergency response was entangled with delay, neglect, police transport, and racist disregard. Freedom House made another arrangement possible. It trained people from the community, served people who had been abandoned, and demonstrated that emergency medicine did not have to be organized through the systems that had already failed them.

It also matters historically. The legacy of Freedom House is now recognized as foundational to modern emergency medical services and paramedic training, with public histories describing its role in shaping the standards of emergency medical care that later became national practice. But that is precisely where the argument becomes painful. The practices survived in another form. The knowledge traveled. The standards spread. The innovation was absorbed into hospital systems, municipal systems, 911 infrastructure, administrative authority, and emergency medical professionalization. The relation of community refusal that made Freedom House necessary did not survive in the same way.

That does not mean Freedom House should not have existed. It means success is not escape. Freedom House worked, and that matters. It also became part of a history through which the same administrative world that had failed Black communities learned how to administer emergency medical care differently. The system did not need to preserve the refusal. It needed the technique. It needed the training model, the ambulance standards, the operational proof, the evidence that advanced pre-hospital care could work.

This is the hard part. Power does not only capture weak things. It captures strong things. It captures working things because they work. The fact that an alternative is effective can be exactly what makes it useful to the institutions it was built against. A practice born from refusal can later strengthen the administrative world that made the refusal necessary.

This is not an argument against lifesaving work. It is an argument against mistaking lifesaving work for exit. Survival practices deserve defense on their own terms. They do not need to be converted into world-historical escape routes to matter. But when we call every effective alternative a step toward a different world, we lose the ability to study how effective alternatives become part of the world they opposed.

4: Portability Is a Capture Interface

Technical culture gives the option-form a moral language. It calls it openness, interoperability, portability, standards compliance, documentation, accessibility, and reuse. These values have histories, and many of those histories matter. Open standards and free software have resisted monopoly, enclosure, vendor lock-in, opaque infrastructure, and corporate control. They have given people tools, leverage, and room to breathe. But technical culture often treats these values as if they settle the politics of use, when they usually only settle the conditions of availability.

ActivityPub is useful here because it is not a villain. Published as a World Wide Web Consortium Recommendation in 2018, ActivityPub is a decentralized social networking protocol providing client-to-server and server-to-server APIs for creating, updating, deleting, delivering, and federating content. The Fediverse has real communities, real histories, real moderation practices, and real attempts to make social life outside corporate platforms. Those things matter. But the protocol is not the world. A protocol can move messages; it cannot by itself move obligations. A server can federate; it cannot by itself carry trust. A codebase can be copied; it cannot by itself reproduce the relations that made it meaningful.

This became visible when Truth Social appeared using Mastodon code. The Software Freedom Conservancy wrote in October 2021 that early evidence strongly supported Truth Social’s test site being based on AGPLv3-licensed Mastodon software, with the dispute focused on whether Truth Social was offering users the corresponding source code required by that license. Mastodon also described Truth Social as using Mastodon code and said it had sought compliance with the AGPL source-code requirements.

The point is not that Mastodon caused Truth Social. That would be unserious. The point is that the code did not carry the world with it. A platform associated with anti-corporate, decentralized, community-moderated social media could provide technical material for a right-wing platform because the software was portable in a way the surrounding relations were not. The AGPL could require source availability. It could not require anti-fascist sociality. The protocol could support federation. It could not enforce a world.

This is not a bug in the narrow technical sense. It is the politics of portability. The Open Source Definition, maintained by the Open Source Initiative since its founding in 1998 and derived from the Debian Free Software Guidelines, requires that open source licenses not discriminate against persons, groups, or fields of endeavor. That commitment has a history and a rationale: it prevents arbitrary permission regimes and preserves broad freedom to use software. But it also means that open source, as formally defined, cannot function as a complete ethical boundary around use. It can preserve certain freedoms of access, modification, and distribution. It cannot preserve the relations in which those freedoms take on liberatory meaning.

This is why ethical source projects keep appearing at the edge of open source. Coraline Ada Ehmke’s Hippocratic License, first released in 2019, the Anti-Capitalist Software License, and the Non-Violent Public License are all responses to the same contradiction: developers want the circulation benefits of open software without the political emptiness of unrestricted availability. Whether or not any particular license works legally or strategically is a separate question. The important point is that the desire for these licenses reveals the limit of openness as an ethics. People keep trying to make software carry obligations that the dominant open-source form is designed not to carry.

A hostile system does not have to share your values to use your affordances. It only needs the affordance to arrive without the relations that would make its use answerable. This is why portability is not neutral. It decides what travels and what gets left behind. In technical systems, what travels best is often the part that has already been cleaned of local relation: the code, the interface, the protocol, the documentation, the API, the license, the standard. What travels poorly is the part that made the thing matter: trust, memory, obligation, moderation history, shared danger, local accountability, refusal.

The repo is not the community. The license is not the relation. The protocol is not the world.

5: FAIR, CARE, and the Refusal of Reuse

Data governance has already encountered a version of this problem, and it gives us a more precise vocabulary than most platform debates. The FAIR principles—Findable, Accessible, Interoperable, Reusable—were developed to improve scientific data management and stewardship, especially in a research environment where data, tools, workflows, and algorithms need to be locatable and usable across systems. Mark D. Wilkinson and colleagues formalized the FAIR principles in “The FAIR Guiding Principles for scientific data management and stewardship,” published in Scientific Data in 2016. In that context, FAIR makes practical sense. It is an answer to fragmentation, waste, opacity, and scientific inefficiency.

But FAIR is also an almost perfect statement of the option-form. Findable. Accessible. Interoperable. Reusable. These are exactly the properties the dominant system wants from anything it hopes to metabolize. Again, this does not make FAIR evil. It means FAIR is an availability framework, not a complete ethics of relation. It tells us how to make digital objects travel. It does not, by itself, tell us whether they should travel, who has authority over their movement, what obligations attach to use, or what damage reuse might cause.

The CARE Principles for Indigenous Data Governance were developed partly because this relational absence matters. CARE—Collective Benefit, Authority to Control, Responsibility, and Ethics—shifts the center from data mobility to Indigenous self-determination, collective interest, authority, and purpose. Stephanie Russo Carroll and colleagues published “The CARE Principles for Indigenous Data Governance” in the Data Science Journal in 2020. CARE does not simply add “be nice” to FAIR. It changes the grammar of the problem. The question is no longer only whether data can be found, accessed, interoperated with, and reused; it is whether its use remains answerable to the people, relations, and futures implicated by it.

OCAP makes the same issue harder to evade. The First Nations principles of Ownership, Control, Access, and Possession, first articulated in 1998 and maintained by the First Nations Information Governance Centre, establish how First Nations data and information should be collected, protected, used, and shared. They explicitly place data governance in relation to First Nations sovereignty, worldview, traditional knowledge, and protocols. This is not a technical addition to openness. It is a jurisdictional and relational challenge to the assumption that access is the natural endpoint of ethics.

This matters for alternatives because many alternatives are treated like data under FAIR logic. They are made findable through publication, accessible through documentation, interoperable through standards, and reusable through licensing. Once this happens, they can circulate as good objects. They can be adopted, forked, taught, translated, and funded. But CARE and OCAP remind us that circulation is not the same as justice. Knowledge does not become ethical because it moves. A practice does not become liberatory because it can be reused. A system does not become public simply because it has been made visible.

This becomes very plain around Indigenous autonomy. On paper, many radicals know how to praise sovereignty, privacy, refusal, opacity, land-based relation, independent technology, and the right of a people to govern their own affairs. The words are not difficult. The difficulty begins when those values do not offer themselves as an alternative for everyone else to adopt. What if a system is for a people, and not for outsiders? What if the point is not to generalize it? What if the right relation is maintenance rather than publication? What if even knowing how the thing works gives people outside the relation too much power over it?

The demand for openness can start wearing radical clothes while behaving like extraction. If the work is not shared, it is treated as suspicious. If it is shared, it is judged as an option. Then the old measures return: documentation, interoperability, scalability, neutrality, accessibility to people it was not made for, usefulness in contexts that do not share its obligations. Difference gets converted into comparability, and comparability gets mistaken for accountability.

This is the point at which “access” becomes a weaponized virtue. Access to what, for whom, under whose authority, and with what right of refusal? If those questions cannot be asked, then openness is not an ethics. It is just logistics.

6: The Alternative as Training Data

The dominant system does not only extract tools from alternatives. It extracts behavior. Every alternative produces a trace: issues opened, conflicts moderated, users onboarded, abusers handled, funds distributed, burnout managed, governance disputes resolved or not resolved, forks attempted, standards debated, norms stabilized, edge cases discovered, rituals invented, language refined. These traces do not remain inside the project. They become observable material. They teach platforms, funders, states, companies, researchers, and future builders what kinds of arrangements people will tolerate, maintain, desire, defend, or abandon.

This is why the alternative becomes training data for governance. Not necessarily training data in the narrow machine-learning sense, though sometimes that too. Training data in the wider cybernetic sense: behavior that makes a system’s environment more intelligible and therefore more governable. A cooperative teaches capital where cooperation breaks under market pressure. A mutual aid network teaches institutions how informal care routes around abandonment. A community server teaches platforms which moderation tasks can be delegated to users. A radical curriculum teaches universities how critique can be packaged into professional development. A land-based practice, once described too clearly, teaches outsiders which relations can be simulated without obligation.

This is not a claim that nothing should be shared. It is a claim that sharing is never only sharing. It produces records, and records can travel into systems that do not share the obligations of the people who generated them. Documentation can protect against abuse, coordinate care, preserve memory, and make collaboration possible. It can also create an extraction surface. The question is not whether documentation is good. The question is: documentation for whom, under what conditions, with what defensive limits, and toward what forms of use?

This is where second-order cybernetics becomes useful. Heinz von Foerster’s Cybernetics of Cybernetics (1974) and Observing Systems (1981), along with later work by second-order cyberneticians, pushed cybernetics toward the recognition that observers are part of the systems they describe; there is no innocent position outside observation. The observer changes the system by observing, and systems that know they are being observed behave differently. Alternatives do not simply exist in front of power. They exist in environments where the act of explaining, documenting, publishing, and interpreting them becomes part of their operational reality. The observer is not outside the system. The writeup is not outside the practice. The case study is not outside the struggle. The explainer is not outside the extraction.

This includes this essay. It takes a thought about non-portability and makes it portable. It turns refusal into argument, argument into content, content into circulation, circulation into whatever Substack, search indexes, citations, screenshots, notes, and models do with it after this. That does not make the argument false. It makes the argument implicated. The critique is not outside the mechanism. The best it can do is mark the conversion while it happens, clearly enough that the next time someone reaches for a toolkit, a framework, a standard, a public explainer, or a “how can we scale this?” question, something in the hand slows down.

That hesitation matters because the system runs on automatic yes.

7: Environment Is Not Context

The deepest mistake in alternative-building is treating environment as context. Context is what surrounds the system. Environment, for the systems I am talking about, is part of the system. A community kitchen is not merely a food-distribution object located inside a community. Its viability may depend on trust, memory, informal obligation, knowledge of who eats what, who can be asked for help, who cannot be asked, who needs anonymity, who has been harmed before, who carries keys, who knows which neighbor will call the cops, who can stretch rice, who knows where the extra coolers are, who can drive after dark, who should not be photographed, and who will show up without being praised.

Strip out the kitchen and ask whether it is an alternative restaurant, and you have already changed the object. It might become useful in that form. It might even inspire something good. But the relation has been altered. The same is true of mutual aid funds treated as alternative welfare offices, servers treated as alternative platforms, land-based practices treated as alternative sustainability models, Indigenous governance practices treated as alternative political theory, and care arrangements treated as alternative service provision. The object alone does not tell you what the system is. Often, the system is the set of relations that make the object possible.

This is where Stafford Beer becomes more interesting than the usual dashboard version of Beer. Project Cybersyn, developed in Chile between 1971 and 1973 under Salvador Allende’s Popular Unity government, with Stafford Beer, Fernando Flores, and Chilean engineers, workers, and administrators, is often remembered for its aesthetics: the operations room, the telex network, the proto-internet aura, the socialist cybernetic dream. But its deeper significance lies in the problem it tried to address: how to build a distributed decision-support system for a nationalized economy without simply reproducing command bureaucracy. It attempted, however incompletely and under enormous pressure, to route information from factories through statistical monitoring and back into decision-making while preserving some degree of worker and enterprise self-regulation.

The lesson is not that Cybersyn is the good version of cybernetics and corporate management is the bad version. The lesson is that the same cybernetic vocabulary—feedback, viability, regulation, information, control—does not determine the politics of the system. The politics lives in the relations: who senses, who decides, who is obligated, who can refuse, what counts as noise, what counts as constraint, what happens when local knowledge conflicts with central modeling, and whether the system is built to preserve autonomy or merely optimize throughput.

This is why a different system is not just “the better version of what is already available.” It is not Uber with worker ownership, Twitter with nicer moderation, Amazon with a cooperative wrapper, welfare with a friendlier interface, policing with community input, finance with values branding, or open source with a better code of conduct. Those things may reduce harm. They may be tactically necessary. They may give people room to breathe, defect, survive, gather, learn, and fight. But many remain alternatives because they preserve the function and change the governance, ownership, aesthetics, or access layer around it.

Difference is harder. Difference may require relations the old environment cannot host: trust that cannot be onboarded, memory that cannot be archived publicly, land that cannot be abstracted into location, language that cannot be translated without losing authority, obligations that cannot be reduced to access control, privacy that cannot be represented as a user setting, refusal that cannot become a feature, and smallness that is not a stage before scale.

If the environment is part of the system, then portability is not merely movement. It is amputation.

8: Abstention as System Capacity

This is where doing nothing becomes more interesting than it first sounds. Not doing nothing as purity, lifestyle, retreat, or the tote-bag version of rest where you are permitted to pause only so you can return to the same machine with slightly better hydration. Doing nothing matters because a system that must respond to every input can be controlled by whoever can generate inputs.

This is a cybernetic claim before it is a moral one. If a system has no meaningful no-op, no capacity to let a prompt die, no way to distinguish disturbance from obligation, then input generation becomes control. You do not have to win an argument with such a system. You only have to keep it processing. Tickets, crises, provocations, opportunities, emergencies, deadlines, grant calls, discourse cycles, platform collapses, bad versions of good ideas, worse people building worse tools faster—each one becomes an occasion for response. A system with no abstention capacity will experience all of this as governance.

Technical culture is especially vulnerable here because it has beautiful little machines for converting anxiety into tasks. Open an issue. Make a branch. Write the spec. Spin up the server. Draft the manifesto. Launch the project. Start the working group. Build the alternative. Keep the future moving by giving the present more of yourself. This can be responsible. Sometimes it is responsible. But if responsibility can only appear as more production, then responsibility has already been captured by the system that needs you producing.

W. Ross Ashby’s law of requisite variety is usually read from the side of the regulator: if you want to stabilize a system, you need enough variety to respond to disturbance. But there is another side. If the dominant system increases its variety by harvesting alternatives, then not every new form of variety should be made available to it. Refusal is not merely saying no to a project. It is the preservation of unharvested variety. It is the defense of relations that do not become inputs.

This is why abstention has to have consequence. If “abstain” only means “someone else does the thing anyway,” then abstention is decorative, a checkbox beside a conveyor belt. Real abstention changes the field. It withholds labor, withholds knowledge, withholds legitimacy, withholds the object the system was prepared to process. It may slow a cycle. It may prevent documentation. It may deny a hostile institution the chance to learn. It may keep a practice inside the relations that make it true.

The dominant order gives us a board-game picture of politics, where refusing a move simply passes advantage to the opponent. Sometimes that picture is useful, but it is not the world. The world is not a closed rule set with stable victory conditions. What a move means changes while the move is being made. The conditions of action and inaction shift through the consequences of prior action and inaction. A refusal can open something. A pause can break a rhythm. Not building can deny the system an object it was ready to process.

A hope without a nope in it is not really hope. It is recruitment.

9: The Good Alternative-Builder

The option-form does not only capture objects. It produces subjects. The good alternative-builder is legible in a particular way: constructive, solutions-oriented, technically competent, ethically articulate, well-documented, responsible, collaborative, transparent, available for panels, capable of writing grants, comfortable with standards language, and able to explain their work to outsiders without making the outsiders feel indicted. They are not merely against something. They build.

This subject is rewarded because they make antagonism administrable. They give the system someone to talk to. They give funders a proposal, journalists a story, platforms a moderation model, policymakers a stakeholder, universities a case study, companies a pattern, and publics a reassuring picture of dissent as productive citizenship. The good alternative-builder translates contradiction into infrastructure.

Again, this is not a moral accusation. Many people occupy this role because people need things now. Someone has to maintain the server. Someone has to coordinate the fund. Someone has to build the tool, write the guide, answer the email, hold the keys, teach the workshop, keep the kitchen open, and make the spreadsheet less cursed. That labor matters. But it also positions the builder as an interface between a hostile world and the relations that hostile world wants to understand.

This is one reason refusal is so often pathologized. Refusal does not produce the good subject. Opacity seems suspicious. Nonparticipation seems immature. Smallness seems like failure. Non-scalability seems like a design flaw. Incommunicability seems elitist. The person who says “this is not for you” becomes harder to process than the person who says “here is how you can adopt it responsibly.” The latter can be invited into the room. The former threatens the room’s claim to relevance.

This is also why technical culture keeps confusing legibility with alignment. If someone can explain their system, document it, and make it interoperable, they appear serious. If someone cannot, or will not, they appear underdeveloped, proprietary, paranoid, or politically confused. But the refusal to become legible to a hostile environment may be a more advanced political capacity than the ability to explain everything clearly.

The old system loves responsible enemies. They are easier to process.

10: Different Questions

The old questions arrive quickly because they are already installed in us. How does it scale? Who can use it? Can it interoperate? Is it documented? What license is it under? Can it be forked? Can it be audited? Is it neutral? Is it accessible? Is it better than the existing option? Can it be adopted without changing the people adopting it?

Those questions are not always wrong. They are just not innocent. They belong to a world that already knows how to process alternatives.

A different set of questions begins elsewhere. What relations does this require? What has to remain unavailable for it to keep working? Who is obligated to whom? What would adoption destroy? Who is made safer by documentation, and who is made vulnerable by it? What forms of refusal does the system preserve? Where does abstention stop something from happening? Which parts are techniques, and which parts are relations that cannot be ported without becoming something else? What would the dominant system have to strip away in order to use this, and have we already helped it do that?

These questions do not produce clean answers. Good. Clean answers are often where extraction begins. The world that wants to use you prefers your work in pieces: a model, a diagram, a toolkit, a checklist, a protocol, a grant deliverable, a best-practices document, a reproducible workflow, a public-facing explanation. Sometimes those are useful. Sometimes they are necessary. But sometimes they are the teeth.

The better research program is not “what should we build instead?” It is: how do practices become alternatives? When does a relation become a model? When does a model become a standard? When does a standard become infrastructure? When does infrastructure become governance? When does governance become the new condition everyone else has to survive?

This is not only a political question. It is technical, linguistic, material, and archival. It is about APIs, licenses, source repositories, data standards, procurement forms, moderation queues, funder language, university syllabi, search indexes, screenshots, citation habits, and the soft violence of “best practices.” Best practices are often the afterlife of someone else’s emergency. A practice is developed under pressure. It works because people had to make it work. Later, someone observes it, abstracts it, cleans it, names it, and circulates it as a transferable lesson. The original conditions disappear. The lesson survives.

That survival is treated as respect. Sometimes it is. Sometimes it is how the teeth enter.

11: A Different System Is Not an Alternative

A different system is not defined by being morally better. It is defined by the fact that the old system cannot host its relations without changing what they are. That means different systems often look disappointing from the outside. They may not scale. They may not publish. They may not invite everyone. They may not satisfy transparency norms. They may not produce metrics. They may not be available as a model. They may not let outsiders verify that they are good.

This makes people uncomfortable, especially people trained to treat public legibility as the basis of ethics. But the public is not abstract. Some publics include enemies. Some publics include institutions that know how to turn critique into policy, policy into procurement, procurement into infrastructure, and infrastructure into governance. Some publics include people who do not want your world to live but are happy to use whatever your world figured out.

This is why a system can be destroyed by making it available as an alternative. Not because sharing is always wrong, and not because secrecy is always good. Secrecy can protect domination as easily as it protects autonomy. Black boxes can be proprietary, abusive, and violent. But openness is not automatically liberatory either. Interoperability is not neutral. Friction is not always failure. A boundary is not automatically domination. Sometimes a boundary is privacy, ceremony, community defense, or simply the ordinary fact that not every relation is available to every observer.

The point is not to stop building. Build when building is the right relation. Make the tool when the tool is needed. Share what strengthens the people it is meant to strengthen. Save lives where lives can be saved. Make alternatives when alternatives give people room to breathe, defect, survive, gather, learn, or fight. But do not confuse that with making a different world.

A different world is not built by feeding the old one better options.

It begins where the old world’s questions stop working.