We should not have to preserve the conditions that shaped us

The politics of necessity

A contingent arrangement becomes the way things are through a sequence. People organize life around it, adapt, invest effort and identity, and accumulate dependencies. Alternatives become more costly to pursue. The arrangement starts to look necessary, and its accumulated consequences become arguments for preserving it.

The sharpest version is manufactured validation. An institution doesn't only present a choice as destiny. It can make the world confirm its assumptions. A platform defines what counts as engagement, shapes the behavior it measures, and cites the measurement as evidence that its design works. A hospital raises productivity expectations until staff adapt, then treats the adapted performance as evidence that the expectations were reasonable. The institution's success at producing confirming evidence can reduce the availability of independent evidence against its assumptions.

The archive's measurement essays take this apart: The Meter and the Whip asks which instruments survive being turned into targets, and The Invisible Denominator shows what a measure hides before anyone games it. How I Learned to Laugh at 'Inevitable' has the shorter version: inevitability has to manufacture its own proof.

The question is how an institution can distinguish discovering that it was right from acquiring enough power to make itself right.

Adaptive foreclosure

People adapt. A nurse becomes expert at managing understaffing. A worker becomes fast at compensating for defective software. A community builds informal systems to survive inadequate public services. Their competence is real, and it is what makes the underlying failure less visible and less urgent to fix.

The archive calls the result institutional insulation. The better people become at living under a constraint, the easier it is for the institution to keep the constraint. The adaptation obligation follows: developing a capability should not create a legitimate obligation to keep exercising it. The value of a person's adaptation should survive the disappearance of the conditions that required it.

There is a counterweight. A person shouldn't have to justify their particularity through usefulness, productivity, or conformity. The program is not another optimization; it is an account of what revisability is for. Rank Means Never Having to Adapt is the positional case, and The Right to Get Tired is the practice.

Effective revisability

Freedom is not the number of options on offer. A person can choose employers without influencing the organization of work, choose an accommodation without contesting the norm, or formally refuse while bearing a cost that makes refusal unavailable. Freedom requires the effective power to contest and revise the conditions that structure the options.

The test is whether the opportunity to change an arrangement survives the dependencies people accumulate while living under it. That is second-order freedom: not the freedom to choose among available options, but the power to participate in determining which options are made available.

Revisability is not restitution. A system can roll back a record, correct a decision, repair the consequences, restore the losses, or revise the arrangement that permitted the failure. It can succeed at the first two and fail at the rest. The legitimacy of automation depends partly on whether its operators can bear responsibility for harms that cannot simply be undone.

The archive's cases are Speak Up and Become Disloyal, where the protective move is structurally foreclosed in practice, and You Were Free to Choose, where the option exists and using it costs too much.

The legitimate obligation

The hard question is not whether institutions should be contestable. It is how people can rely on one another without creating obligations that constrain someone's freedom. A promise that can be withdrawn whenever it becomes inconvenient is not a promise. A decision that can never settle anything cannot govern.

The strongest objection to the program is: how much revisability is enough? The answer has to distinguish legitimate stability from domination. It has to consider legitimate expectations, compensation, the distribution of adjustment costs, collective authority, and the protections owed to people affected by change.

The morality of sacrifice belongs here. A person can derive pride, knowledge, solidarity, or identity from having endured hardship without believing that eliminating the hardship would undermine the significance of what they endured. The value of the experience does not require perpetuating the conditions that produced it.

The test

The conventional fear about capable AI is that humans lose control of the machines. The more likely problem is that humans keep technical control while losing political and practical control over the institutions that deploy them.

The test is whether the people governed can participate meaningfully in changing the conditions under which the institution's legitimacy, success, and necessity are established. Progress should be judged not only by the possibilities a society creates, but by whether people gain the effective freedom to recognize, pursue, contest, and revise those possibilities.

The hardest question it leaves open is whether institutions can preserve that freedom without making durable cooperation, reliable obligations, and long-term collective projects impossible. That is where the critique has to become a theory of institutional design.