Ye spread the first five files across the same table and found that in each of them someone had said "you may be wrong," yet only a few responses specified which judgment was wrong, on what grounds, and how it should be revised. Only from this did the capacity that Act One truly seeks to establish come into view: the corrigible self is not a person who never errs, nor a prudent observer perpetually ready to overturn himself. It is able to form tentative judgments, to know which question such a judgment answers, to allow relevant materials to change it, and, when no new materials are forthcoming, to stop re-examining and return to action and to life.
The preceding five chapters took up, in turn, confirmation by others, fact and interpretation, language, emotion, and dependence. Together they show that self-judgment is completed neither in an inner citadel sealed off from the world, nor by delegation to relationships, records, or authorities; the subject needs external materials and feedback from others, while retaining a position of participation in how materials enter a conclusion and how the conclusion accepts correction.
This chapter adds no further thought experiment of the same kind; instead, it sets the fictional scenarios of the preceding chapters side by side for comparison. The window dispute in the shared studio, the memory disagreement over the meeting notice, the alteration of words in the project retrospective, the emotional evaluation at the supply station, and the heating and dependence at the temporary lodging appear to belong to different domains of life. Yet they repeatedly expose the same difficulty: when a concrete judgment is expanded into an assessment of the whole person, correction can pass from an epistemic capacity into a form of self-expropriation.
Tentative Judgment and the Object of Correction
A person who must wait for absolute certainty before closing a window, attending a meeting, raising an objection, or requesting a repair can scarcely act at all. Everyday judgment is always formed under limited materials. Corrigibility requires, first, that the subject be able to say: "On the materials currently available, I judge thus, tentatively."
"Tentative" is not an excuse for carelessness. It includes the current grounds, the scope of application, and the conditions under which the conclusion might change. The studio member who says he closed the window when he last left can at the same time state that he saw the latch fall, that he did not re-enter, and that someone may have opened the window afterward. Judgment thereby has degrees of strength, rather than reducing to either belief or complete ignorance.
If a companion supplies access-control records with timestamps, the judgment may change; if he merely repeats "you often misremember," no material has been added. An evaluation of personality cannot substitute for specific counter-evidence. Correction requires new relations to enter the original proposition, not an indefinite decline of conviction under pressure.
The subject may also invoke "this is my experience" as a warrant exempt from examination. Feeling cold is an experience that genuinely occurred, but it does not directly prove a heating failure in the whole building; having seen the notice does not entail that the notice was never altered. Acknowledging a judgment as tentative both protects experience from erasure and prevents it from bearing conclusions beyond its own scope.
A demand that is easily exploited rewrites tentativeness as permanent suspension. The person who holds the power of interpretation need not offer counter-evidence; it suffices to keep reminding the other, "you too may be wrong." Each time the party concerned approaches a conclusion, he is required to re-examine his motives, his memory, his emotions, and his wording. The process appears open, yet in fact no material whatever can bring the re-examination to an end.
Infinite openness is not a higher degree of corrigibility. It cancels the point in time at which a judgment acquires tentative force, and it assigns the ongoing labor of knowing to those with fewer resources; the controller may keep his original decision, while the affected party must continually prove that he is entitled to raise the question at all. When doubt is distributed asymmetrically, correction becomes delay.
Stopping Re-examination Is Also an Epistemic Capacity
Stopping re-examination does not assert that the conclusion is forever correct, nor does it require that truth and life first be converted into a single balance of benefits. That no new materials are currently obtainable, that the conclusion already suffices to support a limited action, or that the party concerned needs rest can each ground a temporary stop. An individual's ceasing to invest further and an institution's completing its investigation are two different matters; that the person has exhausted his strength cannot serve as the institution's basis for declaring the facts fully ascertained.
Whether the meeting notice was altered at a given time can be handled by agreeing to examine version records, sending logs, and the retained copies of two independent recipients. If these materials suffice to answer the disputed question, and no substantive contradiction remains pending, the participants may state the current conclusion and close the case for the time being; the number of sources is not itself a criterion for closure. Should the original document later surface, the conclusion can still be reopened. Conditions for stopping and conditions for reopening can coexist.
Institutions without stopping conditions often mistake endurance for truth. Those who can keep appealing leave more records, while those who are exhausted and withdraw are treated as having acquiesced by default. In the end, the archive preserves only who could still keep speaking, not how the dispute arose.
Conversely, closure that comes too quickly can also suppress correction. A single oral explanation cannot override contrary version records. Rules for stopping must be proportionate to the stakes: an ordinary misunderstanding may end fairly quickly, whereas matters involving housing, income, health, or long-term assessment of identity require more complete materials and accessible channels of re-examination.
The corrigible self is therefore not perpetual doubt; it is the capacity to manage the opening, the provisional settling, the re-examination, and the closing of judgment. It possesses a temporal boundary. Without that boundary, epistemic openness becomes a continuing requisition of the subject's life.
Correction Must Know What It Is Correcting
The preceding chapters have repeatedly distinguished experience, fact, interpretation, and responsibility. These four strata can influence one another, yet no one of them can directly absorb the rest.
In the fictional dispute at the supply station, the resident Lin feared that she would not receive supplies — this is a report of experience; whether the application was submitted and whether it entered the list are questions of fact; whether a staff member deliberately deleted the application is a judgment about intention; and who is to explain and remedy the matter once an error has appeared is a question of responsibility. Even if it were established that the application was never completed and submitted, this would not prove that she did not feel fear; and acknowledging the fear would not prove that the staff member harbored hostility.
When correction lacks an object, it often slides from a local error into a global assessment of the subject. A person who misremembers a meeting time is required to concede that his memory is unreliable in general; a single over-strong inference about motive is taken to require abandoning the description of all consequences. Conversely, one recording error by an institution does not directly prove that the entire procedure is manipulation.
Once the scope is clear, a person can revise an interpretation while retaining the experience, correct a fact without accepting a denigration of character, and acknowledge that the other party meant no harm while continuing to demand that the consequences be addressed. Such correction does not force the subject to choose between "I am entirely right" and "I am not to be trusted."
Four Kinds of Revision Require Different Materials
Revising a report of experience may consist in changing "I was angry then" to "more precisely, I was afraid then." The party concerned may also acknowledge that he no longer remembers clearly what he felt. As to whether the cold was caused by the room's position, by bodily condition, or by the equipment, that belongs to the revision of a causal explanation. The two tasks must not be conflated: an experience does not automatically vanish because its cause is re-adjudicated, yet the memory and formulation of an experience are not infallible either.
The stratum of fact requires objects, times, and comparable records. Who sent which version of the notice, when the heating equipment operated, whether a given sentence entered the minutes — each can be supported by the appropriate materials. The materials may still be incomplete, but the dispute at least knows what is missing.
The stratum of interpretation requires the reasoning to be made explicit. The same facts can support several causes; one must compare which explanation covers more materials and depends on fewer additional assumptions, and state what counter-examples would weaken it. Packaging an interpretation as "common sense" conceals the burden of inference.
The stratum of responsibility does not simply follow intention. A coordinator may in good faith have relied on an erroneous record and still be obliged to complete the correction and remedy; an affected party may have expressed herself excessively and does not thereby bear the whole responsibility for a defect in the facilities. Authority, foreseeable consequences, and capacity to correct jointly enter the allocation of responsibility.
Dark systems of interpretation habitually mix the four strata by design: they first use the party's emotions to negate facts, then use a single factual error to negate character, and finally use an assessment of character to cancel the appeal. If the subject wishes to correct some detail, he is required to accept the entire narrative about himself.
Yet criticizing this mixture cannot establish an exemption in the opposite direction. The importance of damaged experience does not mean that every factual judgment is correct, and inequality of power does not prove that every adverse outcome is intentional. Stratification makes critique more precise, because it allows real harm to be seen without having to depend on inflated conclusions.
Access to Materials and Feedback from Others
Corrigibility cannot be merely an inner virtue. A person may be willing to revise, yet if he cannot reach records, alternative sources, or fellow appellants, an open posture will not automatically yield reliable judgment.
In the version of the studio where permissions were concentrated, the person in charge decided whether records were made public; in the dispute over the meeting notice, the person in charge held the recipient list and the sending records; at the temporary lodging, the coordinator held both the maintenance files and the assessment of renewed stays. In such situations, demanding that the party concerned "remain open" is incomplete. One must also examine who controls the materials capable of changing the judgment, and whether proposing a different conclusion brings unrelated consequences.
Access to materials is not a matter of the more the better. Dozens of forwarded copies may derive from one and the same original message, and repeated appearance adds no independence; a large mass of unsorted documents likewise transfers the cost of verification onto those with less time. What corrigibility requires is that the relations among sources be visible, that key records be traceable, and that sufficiently independent paths of return exist outside the original interpreter.
Records Can Both Protect and Discipline
The preceding chapters affirmed records more than once: they allow memory disagreements to be settled otherwise than by authority or volume, and they give rule changes a timeline. But a record is not reality itself. The camera's field, the definition of fields, the authority to fill them in, and the retention period all select what becomes visible.
If an organization records only "completed or not," and not the improvised substitutions, the waiting, and the objections, a stable surface will conceal the maintainers' additional labor; if it records only how often a person raised problems, and not whether those problems were later substantiated, the appellant may be rendered into data as a "high-conflict member."
Stronger control can require the subject to submit self-reports without end. Recording emotions, loyalty, errors, and improvements each day may appear to increase self-knowledge, while in fact subjecting private life to continuous external evaluation. Self-observation is refashioned into self-surveillance; the more complete the record, the harder it becomes to preserve any zone that owes no explanation to power.
Records therefore need a scope defined by the question, boundaries of access, and arrangements for retention and review proportionate to their use. Some records must be preserved long-term, and uniform deletion upon expiry cannot be made a condition of correction. Who may view them, to what decisions they may be applied, how errors are amended, and whether they continue to affect standing after exit — all belong to the conditions of corrigibility. Opposing records does not protect the subject, because oral authority may be even harder to verify; unlimited records do not protect the subject either, because visibility itself can become a resource of domination.
The corrigible self needs to possess records in part, not to become an appendage of records. It can correct itself with materials, and it can also call into question how the materials were produced. The critique of records is not a return to a pure interiority; it places the instruments of observation themselves back into the observable position.
Feedback from Others Must Not Become a Grant of Reality
Chapter One showed that confirmation by others supplies the interfaces of comparison and correction. Without others, the subject is easily trapped in a single perspective; with only others, the subject may hand over the very standing of experience to the relationship.
Reliable feedback involves at least three things: the other states what he observed, distinguishes observation from interpretation, and allows the reply to change his own judgment. If the other merely pronounces "everyone thinks so," feedback becomes social pressure; if the critic merely pronounces "you must believe in yourself," the subject loses the material for comparison.
In the fictional scene of the project retrospective, A proposed that "the standard had changed," while B summarized his conduct as "resistance to change." A shared language helps organize action, but when a label simultaneously fixes the meaning of events, character, and future standing, it is no longer merely an instrument of communication. Corrigibility requires that words can be taken back apart into behavior, time, and consequence.
Others may also acquire interpretive priority through care. When the helper says "I know better than you what is safe for you," there may be limited legitimacy in an emergency; but if this priority has no object and no term, care will gradually spread over the choices of a life. Should the person receiving help raise an objection, it is explained away as his not yet understanding his own situation.
The other end is equally dangerous. An anti-control group may require the party concerned to adopt a unified narrative as proof that he has awakened; continuing to trust certain old relationships, declining to make the experience public, or not acting for the time being are all described as the control still being in force. The subject passes from one system of interpretation into another; the words change, while the structure that grants reality does not.
Distributed Feedback Must Bear Difference
Multiplying opinions does not automatically produce reliability. A group may repeat a single source, or may exclude discordant experiences for the sake of solidarity. The value of distributed feedback lies in allowing difference to remain, and in making sources, interests, and the authority to revise less concentrated.
Some may believe the party's feelings while reserving judgment about intention; others may confirm the recording error while holding that certain misunderstandings in the communication also existed. Different supports need not be fused into one complete story. What cannot for now be unified should be retained, rather than forcibly converged by the position with the most authority or the most injury.
The subject, too, must bear materials that tell against him. Corrigibility is not the mere collection of supporters. Contrary evidence must be relevant, verifiable, and directed at a definite object; a person need not accept insult, blanket denial, or unlimited hypotheticals as counter-evidence, but neither may he refuse a material merely because it comes from someone he dislikes.
Observational convergence in RC describes how multi-level interactions form relatively stable structures. Applied to social judgment, it at most indicates that stability arises from continuing relations; it cannot prove that the majority, an institution, or an intimate holds the authority to grant reality. Social consensus itself must still submit to examination at the entry points of fact and the differences of power.
Self-Discipline and the Conditions of Body and Mind
After prolonged negation, a person may delete, before speaking, the parts he expects to be ridiculed; prove, before making a request, that he has no emotions; and retain, when keeping records, only the fields that fit the external frame. Such conduct can be a protective strategy and does not show that the subject has consented to the external evaluation.
When the strategy lasts too long, it may alter attention itself. The person no longer merely chooses how to express; as soon as an experience arises he asks, "am I permitted to feel this?" The external requirement has become an internal pre-review. Even when the original authority is absent, judgment still passes first through an imagined permission.
Where conditional recognition, unpredictable consequences, monopoly over materials, and difficulty of exit combine over a long period, the subject may reduce the visible expression of difference. To judge whether this change has occurred requires process records and materials across time spans; it cannot be inferred directly from one instance of silence or compliance, still less may adaptation be read as the subject's permanent re-engineering.
Self-reflection here has a double role: it can help a person discover that he has inferred too strongly, and it can also become the language of internal censure. The difference lies not in whether one reflects, but in whether the reflection has a symmetrical object, stopping conditions, and actual choices. When only the affected reflect, while the records and rules of the powerful remain exempt from examination, openness is being used in one direction only.
A Freely Made Commitment Cannot Replace Continuing Consent
A person who publicly supports a group, a project, or a relationship may later find that some of the conditions have changed. An earlier commitment carries moral weight, yet it cannot swallow the whole of future judgment. If every revision is called betrayal, sunk costs will let identity answer in place of concrete reasons.
If the group treats the revision of a public statement as betrayal, changing one's words may carry costs in reputation or relationship. The party concerned may maintain the old formulation to keep his place, but this does not prove that he still believes it inwardly. Public expression, actual judgment, and the consequences the group prescribes must be distinguished; altered cognition cannot be inferred directly from consistent statements.
This does not mean commitment is worthless. Stable cooperation depends on people not withdrawing their responsibilities at every passing discomfort. A corrigible commitment should state its term, its core obligations, its modifiable parts, and the remedies after exit; it holds one answerable for change, rather than defining change itself as moral failure.
A darker structure has the subject take part in evaluating others, to prove that he has understood the rules. The executor derives his position from obedience, and begins to maintain the very institution that once restricted him. This is no ground for writing every intermediate executor as a perpetrator: they may be constrained and consequential at once, and responsibility must be distributed according to authority, choice, and capacity to correct.
The corrigible self also includes acknowledging that one has taken part in harm, without letting the acknowledgment expand into a permanent loss of standing as a subject. If only thorough self-negation counts as taking responsibility, responsibility once more becomes a dispossession of reality. Repair requires concrete actions, consequences, and change; it cannot be substituted by a confession of identity.
Body and Time Provide the Boundaries of Correction
Judgment needs a body. Sleep deprivation can affect learning, attention, memory, and decision-making; Chapter Twenty will set out its scope drawing on materials from the NHLBI. Fatigue, pain, and other forms of distress should likewise be recorded in their concrete circumstances — an entire testimony cannot be canceled by the mere name of a state. Bodily condition needs to enter the examination, not to serve as a shortcut to a verdict on character.
Sustained checking itself consumes the conditions of judgment. Chapter Seven discusses unpredictable demands and Chapter Twenty the domination of body and time; here let one boundary be retained in advance: an arrangement that requires the subject to keep taking positions, or to make major commitments, under acute distress must justify its necessity, and must be examined for whether conditions of pause and adjustment exist.
Changing one's view after rest may show that the earlier judgment was affected by fatigue, or may merely mean that new materials arrived; not changing it does not prove the conditions of the time were irrelevant. Physiological condition and the truth of a conclusion do not map one-to-one; time, task, and reproducible materials must be compared.
One arrangement calls for vigilance: an organization requires continuous waiting or compresses rest, yet when the party concerned expresses confusion, attributes it solely to his personal capacity. What can be checked first here is whether the demands, the conditions of rest, and the criteria of evaluation are consistent. Whether a given performance was caused by sleep deprivation, and whether the arranger deliberately exploited that effect, each still requires separate proof. Chapter Twenty will further distinguish bodily state, task performance, and responsibility within specific relationships.
A critic, too, may decide a person's entire life for him on the ground that "you are not in a good state right now." Deferring high-stakes decisions is sometimes protection — provided its scope is limited, its reasons stateable, the subject able to take part in the review, and his position of decision restored upon recovery. Protection must not become guardianship without term.
The arrangement of time is a condition that admits concrete improvement: one can reply after a full night's sleep, take the question apart, examine materials when no immediate punishment impends, suspend a self-justification that has no terminus. Time is not an auxiliary resource; it decides whether correction still belongs to the subject.
A Finite Baseline for the Corrigible Self
First, experience obtains a position in description. A person can begin by stating what was felt and what followed, without first having to prove an entire causal account; to acknowledge experience is not to accept every interpretation.
Second, propositions have scope. A dispute answers to definite objects, periods, and relations; it does not leap from one error to a global assessment of character, nor from one injury to the whole of intention.
Third, the entry of materials is stateable. Where the records come from, whether the sources are independent, who can modify them, and what an absence signifies all enter the judgment; the quantity of material does not substitute for the relations among sources.
Fourth, counter-evidence has a path of change. The participants state what would raise, lower, or suspend conviction; an interpretation that absorbs every outcome is not deeper, but closed.
Fifth, re-examination can end. Conditions for stopping and for reopening coexist, and the labor of knowing is not requisitioned without limit; high-stakes questions receive time proportionate to them, and low-stakes disagreements need not occupy a life forever.
Sixth, revision does not cancel standing as a subject. A person can acknowledge a factual error, an over-strong interpretation, an omission of responsibility, or past participation in harm, without being required to hand the whole self over to an external narrative; relationships and institutions, too, must bear correction, and must not impose openness in one direction upon those with fewer resources.
These six conditions are not a scale for measuring personality. To score them, and then to decide on that basis who is entitled to believe himself, would re-enact the very structure this book criticizes. They are questions for examining relationships and processes, and their concrete form changes with the scene.
A person may command ample records in one dispute and depend on authority in another; an organization may permit factual correction yet permit no inquiry into responsibility. Corrigibility is not a fixed attribute, but a capacity tentatively formed among subject, materials, time, relationships, and consequences.
RC's "processual completeness" emphasizes that cognition forms tentative judgments through sustained trial and error. Translated into the question of the subject, it cannot be understood as the perpetual dismantling of the self. The aim of revision is not to leave the person without conclusions, but to let judgment act while granting itself no permanent exemption, and to draw on relationships without surrendering to them the ownership of reality.
Retaining Local Capacities Is More Truthful Than Declaring Full Recovery
A subject may already be able to check work records yet still find it hard to speak within an intimate relationship; he may be able to refuse a demand yet still have to live dependent on the same institution. Corrigibility can be recovered locally; it need not wait until every fear and every dependence has vanished before it is acknowledged.
Writing recovery as a single awakening manufactures a new standard of failure: if the person later doubts again, yields again, or needs help, he is deemed never truly to have become a subject. The more realistic judgment is whether concrete interfaces have increased — one more source, one stretch of time free of hurrying, one decision that can be withdrawn, or one supportive relationship not purchased by obedience.
Local recovery may also relapse. Being able today to give an account of the events and falling temporarily silent tomorrow because of new consequences does not make the earlier progress false. Institutions should protect paths of return, rather than requiring the subject to perform steadfastness without interruption.
Recovery must also allow judgments of differing strength. The subject can confirm that a notice was indeed altered while still not knowing whether the alteration was intentional; he can be certain that a relationship keeps him in constant tension while deferring the decision whether to leave. Compressing these distinct convictions into a single "whose side do you really believe" re-manufactures wholesale alignment. The more local conclusions can be preserved independently, the less any later material need overturn the complete self in one stroke.
The supporter's task, likewise, is not to complete the story on the subject's behalf. He can accompany the verification, record what remains unknown, help bear the consequences in reality, and accept that the party concerned finally arrives at an interpretation different from his own. If assistance continues only so long as the subject accepts the helper's judgment, the relationship of recovery still trades obedience for resources.
From the Baseline of Judgment to the Changing of Rules
At this point the distinctions among the conditions of judgment formed across the first six chapters are complete. Act One does not close here, for these normal dependences are about to become the interfaces a hijacker can occupy. Experience needs comparison, yet does not apply to others for its existence; fact, interpretation, emotion, and responsibility are interrelated, yet each has its own evidential scope; language can connect experiences, and can also set up checkpoints; dependence can extend capacity, and can also concentrate materials and consequences.
This chapter gathers these distinctions into a baseline of judgment: tentative conclusions, definite objects, traceable materials, genuine counter-evidence, stopping conditions, and revision that does not cancel the subject. In the discussions of control that follow, one should not proceed merely from a relationship's being uncomfortable or a person's having judged wrongly, but from whether these conditions are being curtailed continuously and systematically.
A rule that changes occasionally may be a reasonable adjustment; it is rules repeatedly modified retroactively, exceptions interpreted by one side alone, and past conduct continually punished under new standards, that gradually manufacture a debt of predictability. The next chapter enters this vertical process, where the corrigible self is no longer only an inner capacity but must face an environment that keeps altering the conditions of judgment.
If the environment permits recording the versions of rules, stating the reasons for change, limiting retroactive consequences, and receiving objections, the subject can revise along with the change. If the environment requires the subject to prove himself anew each time while leaving the rules no verifiable history, correction is inverted into obedience.
The corrigible self is therefore not a solitary fortress newly reinforced. It is more like a set of interfaces that still work: feelings can enter description, materials can enter judgment, counter-examples can change conclusions, relationships can offer support, body and time can set boundaries, and after error there remains a place from which to act. What reality hijacking truly takes away is seldom some particular correct answer, but the conditions under which these interfaces can coexist. Recovery, likewise, is not completed by the declaration "believe in yourself," but begins with returning the conditions to the subject, one item at a time.