On Recursive Freedom: A Theory of the Freedom to Revise
Most arguments about freedom are arguments about obstacles. Remove the chain and the person inside is presumed free, because the thing in the way is gone. Xue Zhixin’s Recursive Freedom: A Theory of the Freedom to Revise begins from the observation that the decisive constraints of this century do not work that way. A recommendation system forbids you nothing; it quietly redraws the space of what occurs to you to want. A gene edit coerces no choice; it shifts the affective baseline from which choices are afterwards made. A decision taken this decade about the climate, nuclear risk, or contact with another civilisation restricts no one alive — it removes options from people who have not yet been born. The book’s name for that last case is the irreversible collapse of possibility. Its claim is that all three share a structure that falls outside the reach of every major theory of freedom we have. The thesis “Freedom is the stewardship of possibility.” It stands on the first page and governs every page after. Possibility here means something precise: the set of states an agent can actually reach from its situation, not the logical or physical possibilities that hold in the abstract. Stewardship is deliberately the weaker verb. One protects what one owns, consumes what one has bought, develops what one has been given; one keeps watch over what must be handed on. Behind the motto stands the proposition the book exists to defend: Freedom is not merely the capacity to choose within a given value structure; more fundamentally it is the capacity to examine and revise that value structure itself. I call this capacity recursive freedom. The working definition is compact: the capacity of an agent to examine and revise its own value function, each word load-bearing. Agent is not restricted to human beings: a firm, a legislature, a machine with a value function and some mechanism for altering it, a civilisation, all qualify. Value function replaces the word “values” with a term covering preference orderings, normative judgements, life projects, moral principles, aesthetic standards, and the weights among them. Examine and revise are held apart on purpose: examination is a cognitive condition, revision a capacity condition, and either without the other is inert. Capacity means stable availability, not a single lucky act, and it depends on institutions, information, material resources, and education. Recursive freedom, the book insists, is not an inner state but a social and material capability. Three levels keep the concept from collapsing into second-order desire. First-order recursion changes the content of a value function — career before family, or the reverse. Second-order recursion changes the structure: it asks why one thinks of value as a ranking at all. Meta-recursion changes the rules by which recursion operates, including whether recursion needs a stopping point. Most of the history of philosophy, on this account, reached the second level occasionally and the third almost never. Recursive freedom is not a new species to add to a list that already holds negative liberty, self-legislation, non-domination, and capability. It is the dimension that makes the others count as freedom; they are its projections under different constraints. The introduction puts it best: negative liberty gives you room, positive liberty gives you the wheel, republican liberty checks whether there are bandits on the road, the capability approach tells you what equipment is on board — recursive freedom asks whether you can still change cars. The argument in three movements The first volume reconstructs what the book calls recursive freedom’s genealogy, analytically rather than reverentially: each thinker is restated in the vocabulary of his or her own period before a parts list and a diagnosis of limits follow. The organising distinction is between the structural level — did this thinker ever reach the question of evaluating the evaluating framework? — and the naming level — was that move called freedom? Almost everywhere the first is present and the second missing, which is why the tradition looks like a scatter of insights rather than a school. The conclusion is stated without modesty: recursive freedom was not invented, it was recognised. The second movement is ontological. Three presuppositions are diagnosed as the inheritance of the standard picture: the agent as substance, its values as static, itself as standing outside the world it evaluates. In their place: a relational ontology, a process ontology, a participatory cosmology, integrated into what it calls the relational–processual–participatory node. The subject is then redrawn as a continuum rather than a threshold, measured along six dimensions including integrative capacity, counterfactual richness, metacognitive accuracy, and social-relational depth. There is no subjective cliff; there are degrees of subjecthood, degrees of freedom, degrees of moral standing. The third movement is normative, and it is where the book takes its largest risk. Two bridge principles are offered. The first: if a property is constitutive of agency — without it there is no full agency — every agent has normative reason to maintain it. The second: if a property is a necessary condition of making any value judgement at all, any being that judges has reason to maintain it. With the premise that deep uncertainty makes revision a condition of rational agency, the argument derives that recursive freedom ought to be respected, not merely promoted. Respected first: it works as a side constraint that may not be irreversibly taken away. The author reports his confidence in the bridges — roughly 85 and 80 percent, 95 for the combination — and concedes that they do not dissolve Hume’s gap but relocate it. That sort of bookkeeping is rare, and it sets the tone throughout. Two results follow. Recursive freedom is non-alienable: an act of giving it up destroys the premise that makes the act voluntary. Nor does it stand alone: it is coordinated with a principle of negentropy responsibility and a principle of reversibility, whose priority ordering shifts with context — reversibility first under deep uncertainty, rough parity in ordinary conditions, reversibility and negentropy before recursion in existential ones. The machinery The formal part of the second volume builds a multi-typed language, eight axioms, and a measure. The axioms divide into existence claims, normative claims, and constraints, among them the requirement that the exercise of recursive freedom not be permitted to extinguish itself. The measure is recursive freedom degree: the proportion of distinct value functions reachable from a given one by non-trivial steps within its closure, decomposed into cognitive, volitional, structural, and relational components. Proofs are machine-checked; the specifications are written so a recursive audit could be engineered. The limits are stated with equal care: a formal system captures only what has already been conceptualised, its axioms are chosen rather than found, and it is closed by nature — hence the requirement that it ship versioned and replaceable, subject to the theory it formalises. What it is for Four pressure domains organise the applied chapters, each a mechanism by which value functions get rewritten: algorithms that reshape preference, neural interfaces that intervene in the formation of the will, genetic editing that locks the biological substrate, long-horizon decisions that foreclose the future. The method is a four-step cycle of principle, diagnosis, design, constraint. The question driving it is never “may this be done?” but “when this is done, is the channel for revision left open?” On AI alignment the book makes a sharp, practical proposal. If the target of alignment is human preference, then the conditions under which human preference may revise itself must be defined in the same breath; meta-preferences — preferences about how preferences are to be changed — belong inside the objective, and systems carry a duty to disclose the paths they open and close. The framing throughout is condition protection, not prohibition. A book that obeys itself Recursive Freedom runs to two volumes, six parts, thirty-two chapters, some seven hundred and sixty pages. The first volume carries the genealogy, the philosophical foundation, and the dialogue with received theories; the second, the formal construction, the contemporary controversies, and the interstellar horizon. Philosophers can begin with the foundation and the dialogue, engineers with the formalism and its applications, general readers with the introduction and the conclusion, which stand on their own. It also submits to its own standard. The text is licensed for rewriting with attribution, the formal code under permissive licences, and the version number serves as the distribution mechanism of recursive freedom, each revision carrying a change log. The justification is blunt: a book that forbids others to rewrite it gives its author no right to speak about recursive freedom. Five limitations are listed at the end — historical coverage, the cross-cultural dialogue, the losses inherent in formalisation, the lack of empirical testing, the author’s vantage point — followed by ten open questions across ontology, epistemology, normativity, practice, and the possibility of procedural agreement among civilisations. The last page does not close. It leaves the space of possibility open, in a sentence that is less a conclusion than a hand extended: not a full stop, an opening.
Authors
- Zhixin Xue (ORCID: https://orcid.org/0009-0007-8131-6531)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-14
- DOI
- https://doi.org/10.5281/zenodo.24217875
- Primary Topic
- Philosophical Ethics and Theory
- Type
- preprint