The algorithm is not your sovereign.

Audio chapter 1 of 6

A better answer. Whose decision?

KirkLast time, the employer handed back the commute and kept the decisions. Now someone hands me an intelligent model. Have I gained agency?

SpockPossibly. Remote control separated a useful concession from a transfer of power. We can apply the same test here: which decisions actually change hands?

AdamA model helps me compare possibilities. But if its controller gets to decide what happens to me, the intelligence can become another remote control.

KirkThe remote has learned to explain why it is pressing the button.

SpockAnd an explanation is not consent. Intelligence can be a resource for individuals and communities to exercise agency. Or it can justify transferring authority to whoever controls the model.

AdamThe anarchist position is that no amount of mathematical sophistication creates a right to command. I can value expertise without granting its owner sovereignty.

KirkThe authoritarian technocratic answer would be: our calculation is sufficiently superior, so we may override your judgment.

SpockAt those extremes, the organizing principles are incompatible. One distributes decision-making power. The other concentrates it behind a claim of objective correctness.

AdamVoluntary delegation is possible. I can ask a tool to do a task, within limits, and take the task back. That is different from surrendering the right to challenge its purpose.

KirkSo the question is not whether the machine is clever. It is whether cleverness has been turned into a license to command.

Audio chapter 2 of 6

Two switches. Two axes.

KirkThen anarchism and algorithmic technocracy sit at opposite corners of the map?

SpockNot necessarily. The site separates ownership horizontally from control vertically. Collective ownership is left; private ownership is right. Centralized control is above; decentralized control is below.

AdamAgency versus imposed authority separates them primarily on the vertical axis. Technocracy can occupy either upper quadrant.

KirkA public planning model on the left. A private platform on the right. Both can impose decisions.

SpockCorrect. Public ownership does not automatically decentralize control. A community-owned model could be managed by a small committee that permits no meaningful challenge.

AdamAnd a private tool can support voluntary decisions. That puts the arrangement lower on the control axis, without making private ownership anarchism.

KirkWhere is anarchism in this site’s vocabulary?

SpockToward collective ownership and decentralized control. But opposition over control does not make every pair of systems diagonally opposite.

AdamTry the switches below. Keep the model equally capable. Change ownership first, with the decision rule fixed. That moves horizontally.

KirkNow keep the owner and change whether people can refuse. That moves vertically. Same intelligence. Different power.

SpockThese are conceptual illustrations, not empirical scores. Real institutions can control different decisions differently, and whose agency counts is part of the question.

Audio chapter 3 of 6

The equation cannot choose its own purpose

KirkSuppose the model really does find the best answer. Why would anyone refuse?

AdamBest for which purpose, and at whose expense? The objective, data, constraints and weights were chosen by someone.

SpockA perfectly computed answer does not establish that its objective is the one people should be required to accept. Calculation can expose a tradeoff. It cannot settle the authority to impose it.

KirkGive me a small equation before we accidentally appoint a sovereign.

SpockImagine a community choosing a bus route. Express scores ninety for speed and thirty for access. Local scores fifty for speed and ninety for access. These are invented teaching numbers, on a zero-to-one-hundred scale.

AdamThe speed weight is yours to change below. Access gets the remaining weight. Each total is speed times its weight, plus access times the remaining weight.

KirkAt equal weights, Express gets sixty and Local gets seventy. Local wins. At sixty percent speed, both get sixty-six. There is a tie.

SpockA tie needs another rule. A model controller would have to supply one; the original equation has not secretly decided it.

AdamNow flip recommendation to binding order. Do not move the weights. The arithmetic stays exactly the same. What changes is whether the residents can reject the result.

KirkWe did not improve the model. We gave its controller a different power.

AdamEven the editable weights leave two preselected routes and two preselected criteria. Meaningful participation includes challenging that frame. Choosing from someone else’s menu is not control over the whole transport system.

Audio chapter 4 of 6

The Chart has to pass its own test

KirkThis site already ranks political systems with mathematics. Are we now arguing against our own Chart?

SpockWe are applying the same test to it. The Chart, earlier in the course, lets readers change the weights used to rank systems. That is calculation readers can explore, rather than a verdict they must obey.

AdamThat is a useful instrument of agency. But it is not the whole test.

KirkBecause we still picked the instruments?

SpockThe site chooses the systems, five vitals and starting scores. The Chart calls the scoring editorial. Changing a weight changes how much a score matters; it does not change the evidence or assumptions behind the score.

AdamSo if panarchism comes first, it comes first inside that chosen model. The ranking does not independently prove the values used to build it.

KirkAnd winning our spreadsheet does not grant a right to govern.

SpockExactly. A model can clarify an argument without becoming an independent warrant for that argument.

AdamOpen the Chart, change its weights and inspect its premises. Keep the right to reject its frame. That is how the site should be used, too.

Audio chapter 5 of 6

Can you refuse and still live?

KirkFine. Put an opt-out button beside every algorithm. Problem solved?

AdamIf saying no costs me my livelihood, shelter or access to essential services, the button can hide coercion. Refusal has to be practical.

SpockAsk who sets the objective, inspects the evidence, corrects an error and proposes an alternative. Who can reverse a decision? Who hears an appeal, and can that person actually change the outcome?

KirkThe appeal desk forwards my complaint to the model that rejected it.

AdamThen the explanation has not supplied a remedy. Authority remains where it was.

KirkBut a community still needs shared rules. A bus cannot take every possible route at once.

SpockCollective agency does not promise that every individual preference can be fulfilled. It does not give one person a right to impose harms on others.

AdamIt asks whether the people affected can meaningfully govern the rules, challenge how they are applied, and hold delegated power accountable.

KirkSo coordination can be necessary without making the model’s owner the only person allowed to decide.

SpockYes. Real constraints remain real. The institutional question is how decisions about those constraints are made and challenged.

Audio chapter 6 of 6

The algorithm is not your sovereign

KirkWe have kept the models, the equations and the expertise. What have we refused?

SpockThe extra claim that superior calculation entitles its controller to command. That is a political claim, and it requires a political justification.

AdamBetter models can improve forecasts, expose tradeoffs and help communities coordinate. Accuracy matters. It should increase our capacity to decide.

KirkKeep two questions separate: how good is the answer, and who has authority to act on it.

SpockThe first concerns knowledge. The second concerns consent, power and accountability. A score for the first does not answer the second.

AdamRemote control asked which decisions the employer handed over. Here we ask which decisions the model controller lets the affected people govern. That is the continuity: a useful tool or concession can arrive while authority stays put.

KirkWho controls the variables? Who bears the consequences? Who can say no?

AdamThose are the political questions. Intelligence should help people exercise agency.

SpockThe algorithm is not your sovereign.