The accord separated action from dispute, and action did in fact become: the northern-section reinforcement proceeded by the terms, the first installment of compensation arrived on schedule, and the dispute went on as before in the council — the order that Chapter 22 brought to a close now, for the first time, let the river valley hold both things at once. Two months later, the cross-version verification demanded in Chapter 20 turned up an anomaly in routine work: an old levee-foundation survey record the committee had once submitted to the base disagreed in its figures with the versions kept at the other two entrances; review, following the procedure of Chapter 21, confirmed that the committee's version had, in transcription, copied two bedrock bearing parameters for the northern section too high — and the two side walls of the reinforcement design happened to rest on precisely those two numbers. The action still succeeded; only, the action that took shape was built on a mis-transcribed figure.
Once the error was confirmed, the shape of the problem changed. The review had answered "whether the numbers were wrong"; it could not answer the three questions that came next: who has the authority to say "we were wrong" on behalf of the institution; what form the admission must take so that it reaches the people who have already absorbed the error; and once it has been said, what relation the admission bears to responsibility, to punishment, and to exemption. This chapter handles these three questions, and its object is not the committee alone — the compilers of public bulletins, professional systems such as surveying and engineering, any position whose output the public takes as a basis, all occupy the same position of admission. At the same time it accepts two stern corollaries head-on: public admission can be performed into a public-relations technique, and a culture of admission can punish honesty in reverse; and it holds the boundary on the other side — genuine public correction really does rebuild trust, and to judge every admission a performance and to take every admission as a virtue are the two directions of the same laziness.
The Authority to Admit Error
Those who discovered the anomaly were the verification clerks and a volunteer at the skeptics' archive, not the committee itself. This is not a coincidence of the valley but a structural constant: those who hold decision-making power simultaneously bear the full cost of discovering error — discovery means that work they handled or endorsed must be redone. If an institution transmits bad news only within the chain of power, the message is discounted once at every level it ascends; the protection group of Chapter 17 kept its misjudgment records in its own hands in exactly this way, refusing disclosure in the name of preventing schism. The right of first report of an error should therefore belong to the position that discovers it: once a verification report is registered, the admission procedure should start, awaiting no one's approval.
Between confirmation of an error and its public admission lies a stretch that can be lengthened without limit. Silent revision — quietly replacing the wrong numbers without saying they were replaced; delay — letting the flood season pass and letting the water close the case on the error; dilution — writing the two specific parameters down as "some data pending further verification." Each is cheaper than public admission, and each leaves no handle. Chapter 8 already established that destroying an entrance is more serious than changing text; silent revision sits precisely between the two: the text is changed, the provenance is erased. Chapter 20 fixed the interval between the error-confirmation date and the current-version update date as the hardest indicator of entrance integrity; this chapter advances it to the institutional level: the interval between the day an institution knows of an error and the day the institution admits the error is an equally hard indicator, equally beyond repair by narrative.
To understand public admission of error as personal courage is to leave it in the moral domain, waiting for a brave chair to appear. What the valley needs is not a hero but a procedure: once an error is confirmed by review, the duty to admit arises automatically, not at anyone's discretion to initiate. This is isomorphic with Chapter 7's establishment of stopping rules for doubt — not demanding that skeptics be more virtuous, but giving doubt an institutional outlet; admission likewise needs a position that does not have to be won anew each time. Virtue may be absent; the procedure must be present; a community that counts on virtue will obtain error correction at the frequency with which heroes appear.
Why the Right of Admission Cannot Rest Only at the Top
If the qualification to admit error belongs only to the top of the institution, every admission must queue for the top level's agenda, wording, and timing. The queue itself is the procedural attrition analyzed in Chapter 21: not refusal, merely permanent placement behind more urgent items. The deeper consequence is the markup — an admission at the top means the error has reached the summit, and the wording of the admission must be weighed together with the institution's overall image, so the list of admittable things grows ever shorter, until only what can no longer be concealed remains. The height of the right of admission is inversely proportional to the frequency of admission; this is not a law of psychology but a consequence of queue structure.
The right of admission should be distributed along the level at which the error resides: an error in a base entry is revised and announced by the entry's custodian; an error in a public decision is admitted by the council or committee that made the decision; a procedural error — such as survey transcription long having no review step — is admitted by the level that holds the procedure. Each level admits only the errors of its own level, without proxying across levels. This is not modesty but precision: an admission that reaches across levels dilutes the specific error into a general confession, and Chapter 22 has already shown that a totalizing narrative, whether toward guilt or toward gesture, preserves no structure.
Along with the devolution of the right of admission there appears a cheap deformation: every level admits error, and what each admits is its own "poor communication," while the levels that truly made the error possible — procedure design, budget allocation, personnel incentives — go unclaimed. The layering of admission must align with the layering of responsibility — whoever holds the right to decide holds the corresponding object of admission; a position with only executive power cannot be forced to issue apologies for the decision level's errors. Otherwise the public-correction system becomes a sewer for responsibility: honesty flows down the gradient of power and pools at the position least able to defend itself.
Cognitive Immunity Does Not Mean Not Erring
Chapter 17 treated cognitive immunity: the community's resistance to manipulation, its recognition of imposture, its defense of common facts. But the word immunity carries a built-in displacement — the work of an immune system is to recognize and exclude invaders, whereas most errors an institution must admit are not invasions but products of its own metabolism: transcription mistakes, crude procedures, skewed incentives. To understand correction as immunity naturally recasts every error as external cause — misled, attacked, undone by weather — and Chapter 17 has already seen how the protection group refused to disclose its own misjudgment in the name of defense. The object of immunity is incoming falsehood; the object of admission is one's own truth; in institutions the two jobs are often carried by the same people, and the confusion is correspondingly easy.
An institution airtight against external manipulation will still quietly produce errors internally; the better the immunity, the harder it becomes for internal errors to find an exit — "our procedures are error-proof" has become part of the institution's self-understanding, and to admit an error is to overturn this layer of understanding. Cognitive immunity and public correction are therefore two capacities, each valid in itself, neither a substitute for the other: the success of the former constitutes no exemption from the latter. Institutions that confuse the two develop a stable symptom: high vigilance toward external falsehood, high tolerance toward internal error, two standards sharing one defensive vocabulary.
For an individual, admitting error is difficult in dignity; for an institution, difficult in structure: the institution that corrects the error is the institution that committed it; judge and party are organizationally coincident. Chapter 14's conclusion about intermediaries applies here — no role is trustworthy by nature; what matters is whether the error can be constrained by other positions. Institutional admission therefore cannot rest on self-inspection alone: the review maintains distance of interest under the four conditions of Chapter 21; the text of the admission may come from the institution, but the validity of the admission must be proven by externally verifiable revision and notification. Self-inspection is one source of material, not proof of validity.
How Public Admission Reaches Consequences
The first form is changing the record itself: the current version of the erroneous entry is updated, and the old version, together with the grounds for revision, the confirmation date, and the scope of impact, passes into the version chain, preserved under Chapter 8's revision trail and Chapter 20's current-version discipline. This is the mechanism the previous two chapters have already erected; this chapter adds only a test function: the institution's attitude toward record revision is the first test of an admission's authenticity. Performed admission almost always skips this step — the statement is full of feeling, while the entries at the base and archive layers do not move by a hair.
The second form is delivering the error into the hands of those who have already absorbed it. Chapter 19 required that corrections reach the point of use, and Chapter 14 showed that a correction left on the archive page is no correction; the institution's position toward the public is an extension of this set of requirements: the error spread through bulletins, rulings, and design drawings, and the admission must return along the same chain — the two compensation rulings based on the erroneous parameters are re-evaluated, the issued construction drawings are annotated with the revision, and the correction bulletin is published from the same level's release position as the original. To hold a press conference for the statement and lock the correction into the archive page is a repair completed in form while the public impact is preserved intact — Chapter 14 has already delivered this criticism, and the institutional level is not exempt.
The third form is the easiest to omit: explaining how the error happened — at which link it arose, through which positions it could have been intercepted, and why the interception did not occur. An account of responsibility is not naming-and-shaming but structural accounting: if the transcription procedure had no review step, that is the procedure's responsibility; if there was a step and no one executed it, that is execution's responsibility; if someone reported it and it was suppressed, that is the suppression's responsibility. The consequences of the three cases are entirely different, and a vague "deep regret for lapses in our work" precisely smooths them into the same thing — wronging the compliant and letting off the responsible. This chapter offers no template for the text of an account; it sets only the requirement: the account must be verifiable, and the verifier should be able to return from the account to the records it cites.
The Form of Arrival Must Be Proportionate to the Diffusion of the Error
A survey parameter is not a number lying in a storeroom; it became design drawings, schedule clauses, three bulletins, and two rulings. The radius of the duty to admit equals the radius of the error's diffusion: to correct only the survey report while failing to notify the residents affected by the rulings is to admit an error that does not exist — the error that never left the archive page. The proportionality principle accepts no bargaining in reverse: one cannot, on the ground that "impact at most points of use is limited," correct only the most visible instance; the judgment that impact is limited must itself rest on an assessment, and no such assessment exists before the re-evaluation is complete.
Notification operates under the proportionality principle on the premise that the institution knows where its outputs went. Which villages and notice boards the public bulletins reached, which households the ruling documents were served on, which work sections the construction drawings were issued to — distribution records are the address book of notification. An institution without this address book cannot even produce the list of its own errors, and can only offer an undifferentiated apology to "the residents at large," and undifferentiated notification equals no notification. Distribution records are therefore not administrative housekeeping but a component of correction capacity, to be maintained in ordinary times rather than reconstructed after the incident.
This structure holds equally for three kinds of position: the deliberative bodies that make public decisions, the public media that issue bulletins, the professional systems that issue survey conclusions. The media of their errors' diffusion differ — decisions diffuse along procedure, bulletins along transmission, conclusions along citation — but the structure of admission is the same: confirmation, revision, notification, account of responsibility, none of which may be waived on grounds of professional particularity. The defense a professional system most easily raises is "technical judgment always carries error"; the defense conflates two things: the existence of error does not exempt the registration of error, and the difficulty of judgment does not exempt the notification of the grounds of judgment.
Chapter 20 set a time limit for current-version updates; institutional admission likewise needs one: within how many days after confirmation by review the revision, the notification, and the account of responsibility must be completed; overdue, the grounds must be registered, and the overdue record itself is public. A duty of admission without a time limit degenerates into an agenda item that can queue forever — and the queue, as noted above, is the first attrition an error encounters when the right of admission is raised so high that only the top authority can exercise it. The length of the limit can be argued about; the existence of the limit cannot.
Bearing Responsibility, Performance, and the Punishment of Honesty
The first boundary: admitting error is not an indulgence. That the institution has publicly admitted its error does not mean the affected residents "ought to move on"; the re-evaluation of the two rulings based on erroneous parameters is not a favor to the residents but a component of correction. To treat admission as a ritual for closing the topic — one admission, in exchange for never raising it again — is the settlement maneuver this chapter will confront directly below; here the structure is first set in place: exemption is the conclusion of a determination of responsibility, not an effect of the act of admission itself; only after layered determination is there anything whose exemption can be discussed.
The second boundary must equally be held: failure to admit is not commission of a crime. Chapter 3 already warned against directly assigning motives to actors from the shape of records; an institution's failure to admit promptly may be malicious delay, procedural blockage, or genuine ignorance. Error, negligence, and cover-up are three levels of conduct, each with its own evidentiary material. Convicting "non-admission" outright will force out more refined cover-ups — since silence is a crime, then perform the admission; the cost of performance has always been lower than the cost of confrontation.
The usable distinction is consistent with Chapter 14's handling of intermediary error: deliberate falsification, gross negligence, and misjudgment under reasonable conditions correspond respectively to three kinds of consequence — accountability, rectification, and exemption. The key institutional design is the exemption zone: for errors that arose despite operating by procedure within one's own authority, the one who admits is not punished for admitting — otherwise no one admits; deliberate falsification and knowing concealment are not exempted for the earnestness of the words — otherwise admission becomes a whitewashing technique. Exemption targets the type of conduct, not identity, and not the degree of earnestness in attitude; a system that couples exemption to earnestness trains acting skill.
The Performance of Admission
Now the first stern corollary: public admission can be entirely performance. The statement can be written with more feeling than a genuine admission, the press conference can carry the well-timed pause and the well-measured self-reproach, while nothing changes at the record layer — the parameters were replaced, but in a silent revision; the rulings were not re-evaluated; the account of responsibility leaves only "inadequacies in the communication link." Identification of performative admission does not rely on inference of motive but on verifiable items: was the entry revised, was the notification issued, was the account of responsibility layered, were the affected rulings re-evaluated. Once those four questions have been asked, the rhetorical quality of the statement no longer matters.
Performance has a longer-lasting variant: one grand total admission, in exchange for permanent exemption thereafter. After "we bear responsibility for this accident" has been said, anyone who raises the matter again is seen as refusing to let go — admission is used as the key that closes discussion, not the door that opens revision. The method of discrimination is isomorphic with Chapter 22's identification of delaying negotiation: look at the acceptance of time constraints. After genuine correction, the relevant entries, procedures, and rulings remain reviewable and re-openable; after settlement-style admission, every re-raising of the same topic is recorded as untimely. The former requires error to be a structure; the latter needs error only to be an event.
The beneficiaries of performed admission are the positions in the institution that must answer for errors, and the superiors who take "moving on" as a virtue of governance; the cost-bearers are twofold — those affected, who receive a ceremony and no re-evaluation, and the basis pool of the next public decision: once the error has been digested by performance, the procedure is unchanged, and the same error will come again along the same path. When it happens a second time, the first performance turns from asset into liability, and all the institution's statements are marked down wholesale — this is the interest that performance finally collects even from the performer.
The dialectical other side must be equally hard: genuine public correction does exist, and it does rebuild trust. To read every admission as public relations is a mirrored cynicism, sharing its laziness with unconditional belief in statements — neither goes to check. The material for discrimination is ready to hand: whether the revision has landed, whether the notification has arrived, whether the exemption zone has been honored, whether the same error has recurred. An institution's admission credit rests not on the quality of its statements but on the historical distribution of these observable items; cynicism and credulity are both evasions of this thing that can be done.
How a Culture of Admission Punishes Honesty
The second stern corollary is deeper: even if the institutional position of admission is established, the incentives may still run in reverse. Members who admit error publicly find themselves marked as "people something happened to," bypassed at promotion, avoided on projects; while the colleagues who chose silent revision at the time have clean files and glossy evaluations. Observing this allocation once is enough — thereafter everyone in the institution knows that the price of honesty is a career and the cost of concealment is nearly zero. This is not a question of character: a system that places honest people in loss positions ends with only two kinds of people, those who can afford the loss and those who lie.
When the punishment of honesty has solidified, institutional honesty is no longer possible — not because no one is willing to be honest, but because the honest are systematically cleared out of the positions where decisions can be made. Chapter 20 analyzed the same drift in preservation budgets: no conspiracy is needed; the distribution shifts on its own. Identification relies not on any single personnel decision but on distributions: the correlation between admission records and promotion records, the ratio between exposed concealments and voluntary admissions, the rate at which the exemption zone is actually invoked. These indicators are of the same family as those Chapter 21 required the review system to publish — the cost of restoring them is merely to lay out the existing registers.
Under inverted incentives, silence develops its own public grammar: "for the institution's reputation," "so as not to damage morale," "now is not the time." Each of these reasons can be true locally, yet together they constitute the ideology of cover-up. Chapter 14's reminder applies here: one must not explain every act of silence as malice, nor let the institution explain every error as timing or overwork — analysis of the conditions of formation helps allocate responsibility; it does not make responsibility disappear. The grammar itself should be registered: when "not the time" becomes the standing reason, the time will never come.
Repair does not rely on moral mobilization but on changing the allocation. The exemption zone must be genuinely honored and publicly invoked — the first case of someone admitting by procedure and going unpunished is worth more than a hundred documents advocating honesty; in evaluation, an admission record must not count as a negative, and an exposed concealment must count as a negative; protection for internal reporting connects with Chapter 17's counter-evidence entrance, able to bypass the immediate superior and forbidding retaliation. The test is equally simple: observe whether, when the next error appears, the timing of the first report from inside the institution moves earlier. Whether incentives are repaired is seen not in documents but in timing.
The Cost of Correction and the Conditions for Rebuilding Common Facts
Chapter 21 has already shown that confirmation by review does not mean trust recovers automatically; the same holds for institutional admission. The rebuilding of trust occurs not on the day of the press conference but in the time after it: whether the revised entries are cited as usual by all parties, whether the re-evaluated rulings are carried out, whether the affected residents actually receive the difference, whether the same error fails to reappear before the next flood season. Chapter 22 required that compliance records register only the completion of terms — the counterpart of institutional credit is precisely a sequence of observable facts of this kind.
For an institution that has never publicly admitted an error, its all-correct record carries no information — one cannot know whether there were no errors or no admissions; an institution that has carried out three genuine corrections makes its fourth statement, on the contrary, more credible. This is the deepest benefit of a public-correction system: it exchanges the unguaranteeable question "whether the institution errs" for the verifiable question "what the institution does after erring." The former demands infinite trust; the latter requires only finite bookkeeping.
From this the statement can be given an exact position: it is the first cell of the sequence of facts, not the sequence itself. The function of the statement is to initiate — to confirm, to commit, to set the time limit; the proof is given cell by cell afterward by the revision, the notification, the re-evaluation, and the account of responsibility. A community satisfied with the statement receives public relations; a community that takes the sequence as its standard receives commitments that can be honored and can be falsified. Falsifiability may be precisely where the value lies: only an institutional commitment that can be falsified can be spoken of as trustworthy or untrustworthy.
The Apportionment of the Cost of Admission
Public correction has real costs: redoing the design, re-evaluating rulings, paying the difference, bearing the depreciation of reputation. How the costs are apportioned decides the system's sustainability. If all of it falls on the individual who admits, honesty will be liquidated out; if the re-evaluation is delayed in the name of "avoiding damage to morale," the victims are made to continue paying interest on the institution's error. The proportionate apportionment is: re-evaluation and the difference are borne by the institutional budget, procedural responsibility by the level that holds the procedure, the execution layer within the exemption zone. The schedule of cost apportionment should itself be public — who paid how much, out of which budget column, visible.
When the institution turns admission into an individual apology by front-line staff — making the transcription technician face the residents alone — it is doing two things: absorbing top-level responsibility at the lowest-level position, and incidentally degrading admission into individual lapse. The layered discipline of the account of responsibility applies here in reverse: an individual may account for what happened at his own link, with neither the right nor the duty to admit error on behalf of procedure, budget, and incentive structure. An apology that cannot trace responsibility upward is merely an interception of responsibility.
Signs of Failure and the Conditions for Rebuilding Common Facts
The signs are observable: statements multiply while entry revisions diminish; admission and non-admission no longer differ in evaluation, or admission becomes a negative; total admissions appear at regular intervals while the same error recurs; the queue time for re-evaluation applications by affected rulings rises year by year; the timing of the first internal report is pushed later year by year. A single item can have other explanations; when several coexist, public admission has retained only its rhetorical function — it is used to name everything except revision.
Recovery is isomorphic with the previous two chapters and starts from the cheapest point: first publish the statistics of the interval between confirmation date and admission date — this requires only laying out the existing registers; then honor the first exemption case and publish it; then restore the time limits and queue distributions of re-evaluation applications; only then the rearrangement of the right of admission and the modification of the evaluation structure. The first steps cost so little that no budgetary grounds can justify refusal, and at that point refusal itself is evidence. Institutions in the failure period will still err; the minimum of correction is that of Chapter 19: the revision of the individual error plus one honest notification, without masquerading as systemic rectification.
The boundary is thereby clear. Public admission of error cannot guarantee that the institution will not err again, cannot substitute for the pursuit of deliberate falsification, and cannot supply answers to disagreements of value — what it accomplishes is one limited thing: turning "we were wrong" from a one-time moral event into an institutional position that can be distributed, verified, and accumulated as credit. A layered right of admission means that admission need not queue to the very top; the three forms — revision, notification, account of responsibility — complete, admission reaches those already affected; the boundary between exemption and the bearing of responsibility distinct, honesty need not be purchased at the price of self-destruction. The two failures — performance and the punishment of honesty — each have signs, the signs can be registered, and registration is itself the starting point of repair. A community does not need an institution that never errs — such institutions exist only in propaganda beyond verification — it needs positions where error can be admitted, admission can be verified, and verification can be remembered. Hold this position, and error is an accident, not the daily secretion of a mode of governance.