Historical texts must change. New materials emerge, old names are corrected, involved parties offer previously omitted experience, and the language of value shifts. If every change is called tampering, error acquires permanent authority merely by virtue of age. Yet history can also lose its conditions of re-examination through modification. Old versions disappear, the reasons for deletion become invisible, changed interpretations are written up as having always been so, and later generations can no longer judge how the shared memory took shape. Both revision and erasure alter the text; they differ in version, object, rationale, entry point, and consequence.
Continue the plaque dispute at the fictional archive. The archive confirms that the exhibit is a later replica and updates the label. The first revision supplements the restoration records; the second removes names to protect donor privacy; the third removes the old label during a website redesign; in the fourth, the director deletes the contested explanation from the meeting minutes, claiming to prevent continued misunderstanding.
These four changes cannot be evaluated uniformly as "changing history." This chapter interrogates each item: whether the event changed, whether materials were added, whether the interpretation was explained, whether the privacy deletion retained restricted re-examination, whether the old versions can still be found, and who has the authority to decide what the public no longer knows.
Events, Records, and Interpretations as Distinct Objects
A past event does not happen again because a label changes a word, yet the conditions under which the public accesses the past can change. Records form after events, and displays select from records. Three layers of time interrelate without being the same object. When the archive says "we merely updated the display," this may be accurate, or it may underestimate the display's influence on public memory. Most visitors never enter the stacks; the label is their reachable entrance to history. Conversely, an erroneous label does not mean the entire collection is invalid. Objects, photographs, restoration records, and testimony each have their own provenance. Correcting one layer should specify which layers are affected, rather than expanding a local error into the claim that all history is unknowable. "The exhibit is a replica" corrects the identity of the object; "the act of replication was intended as commemoration" is an interpretation of motive; "the commemoration deserves praise" is a value judgment. The three statements require different materials and cannot all be derived from a single technical authentication. If new evidence supports only the change of identity, the archive should not seize the occasion to rewrite the entire meaning. Critics, likewise, cannot deny the authentication of the object because they disagree with the value. Layering allows disputes to reach conclusions in part. Historical writing always selects. Space is limited, and exhibitions must compress. Selection is not sufficient evidence of erasure; what matters is whether the omitted material bears on the core conclusion, whether another entry point to it exists, and whether the deletion is disguised as never having existed.
Honesty in revision does not require listing every change of punctuation. The greater the consequence, the stronger the dispute, and the more complex the material relations, the more complete the version note should be. Routine editing and substantive rewriting require different records.
The archive discovers a letter from the period showing that the replication work began earlier than previously believed. Adding the letter to the files extends the material basis, and the old timeline needs adjustment. New material does not automatically prevail over old. The letter's date, the author's position, its chain of custody, and its references all require verification; it may conflict with other records. A so-called new discovery is only a qualification to enter judgment, not a pass to the conclusion. The old label may have been reasonable given the materials then available. Changing the judgment after new evidence appears shows that the institution can learn. If every correction is written up as a scandal, institutions gain an incentive to conceal uncertainty. Negligence or deliberate concealment may also have existed in the past. Admitting that knowledge at the time was limited cannot substitute for an investigation of responsibility. One must compare whether the relevant materials were available, who knew, why they went unused, and what consequences the old conclusion produced. A revision note can state the original judgment, the added material, the changed portions, and what remains unknown. Readers thereby see a process of knowledge rather than choosing between two mutually exclusive stories of certainty. Critics must update as well. If someone long questioned the replica's identity, then after the archive provides complete restoration records, he should lower that suspicion accordingly; he may still contest the meaning of the display. Treating a fixed stance as integrity would make everyone refuse to learn.
The same letter can be interpreted as private commemoration, public propaganda, or administrative arrangement. A researcher proposing a new causal relation need not alter the object or the documentary record, yet the public understanding changes. That multiple interpretations can coexist does not mean every interpretation is equally supported. Each must specify its objects, materials, gaps, and counterevidence. The stopping rules of Chapter 7 continue to constrain historical argument. An exhibition cannot include every view, but it can present the principal judgment, the key disagreements, and entry points to further material. Balance does not mean giving every account equal space; it means making the reasons for excluding an interpretation stateable. An authoritative institution may withdraw dissent that still has evidential support in the name of "the research has been updated." Opponents may likewise invoke "history is always open" to keep a refuted account in an equivalent position. Openness and weighting can coexist. Interpretive change is also influenced by current values. If ordinary laborers receive more attention today and the exhibition adds their records, this can correct a past selection bias; it may also substitute contemporary categories for historical figures' self-understanding, deciding it for them. Both the context of the time and the present question must be stated together. History is neither a pure recitation of past voices nor an arbitrary present borrowing of the past. Revision must make visible the relation between languages of different times, avoiding the disguise of today's judgment as the consensus of the time.
The donor's name is irrelevant to the authentication of the exhibit, yet it may expose family privacy. The archive removes the name from the public label while retaining the provenance relation in restricted files. This deletion protects the subject and does not destroy necessary re-examination. Incorrect personal information, sensitive material disclosed without consent, and real-world security risks may all require restriction or deletion. Treating permanent preservation as history's highest value would turn archives into instruments of ongoing harm. A name can be hidden without deleting the entire restoration record; public access can be restricted without making the material invisible to every examiner; a term can be set without sealing the files forever. The mode of protection should be built around the specific risk. Sometimes even restricted retention is unjustified — for instance, when the material itself was obtained through serious violation and its continued existence would cause uncontrollable harm. Whether to destroy requires legal, ethical, and involved-party participation; no abstract principle of preservation can answer it automatically. Privacy can also be borrowed by institutions. A director says he is protecting the persons concerned while actually deleting content that proves the old decision wrong. Whether the involved party requested it, whether the deleted portion relates to privacy, and whether independent examiners can verify — these help distinguish protection from exoneration. The public's right to know is not unlimited. That an event has public significance does not turn every private detail into public property. Historical continuity must be maintained jointly with the boundaries of living subjects.
How Versions and Entry Points Are Lost
After the website redesign, the old label links fail. Technical migration routinely makes pages disappear, with no intent of erasure required. For the reader, the revision path is still broken. A version chain records when something changed, what changed, on what basis, and where the old version is. It does not prove the new judgment correct, but it allows later verification to reconstruct the process. Internal drafts may contain errors and private information; publishing them all would create misleading. The archive can retain the necessary versions, provide a summary of changes to the public, and allow constrained personnel to inspect them in disputes. Too many versions also create a flood of information. If readers must compare hundreds of minor updates, formal transparency raises the cost of access instead. Substantive changes, their reasons, and their sources should be foregrounded. A more covert mode of erasure leaves the old versions in place but without any index, with search results overwritten by new pages, and with citation links failing continuously. Technical existence is no substitute for actual reachability. The converse risk is that the old error spreads farther than the correction. Version preservation should not let search default to continuing to display outdated conclusions. The current version must be clearly marked, corrections should return along the original paths of dissemination, and historical versions should serve re-examination rather than continue to mislead.
The director deletes the passage about the dispute from the meeting minutes on the grounds of avoiding misreading. The label's content may still be accurate, but the public has lost the material for judging why the revision occurred. The core of erasure is not necessarily writing falsehoods; it is destroying the entry point from the conclusion back to the process of its formation. A break in any link — the materials, the catalog, the provenance relations, the access credentials, the interpretive records — can render history impossible to re-examine. Oral meetings may never have produced minutes, and old systems may have lost files automatically. An absence only indicates the current boundary of evidence; it cannot by itself prove that someone intentionally cleared the record. Establishing destruction requires examining retention rules, logs, directives, patterns of selectivity, and who benefits. A position of power raises the necessity of investigation; it does not lower the standard of evidence. Even without intent, the institution remains responsible for failing to retain important records. Intent determines part of the responsibility for the act; consequence and repair do not wait for motive to be established. If intentional erasure is sufficiently evidenced, it should be named directly. Complex systems cannot serve as a language of exoneration. Whoever ordered the deletion, whoever carried it out, whoever controls the backups, and whoever blocked re-examination each bear responsibility proportionate to their authority.
After the minutes disappear, the director says the dispute never happened, and the critics say this proves every exhibit is secretly manipulated. Both sides fill the absence with a complete story. An absence has conditions of formation and supports only limited inferences. If records should have existed under the rules and only the critical periods are missing, the strength of suspicion may rise; if the institution never recorded such meetings, the absence carries little information. An institution can repeatedly publish a unified version, while popular narratives may spread on emotion and fragments. Whoever is louder does not determine truth, yet it affects which interpretation becomes the public default. Erasure can thus be used to give the authoritative narrative a monopoly on material, and it can also be used by anti-authoritarians as an unfalsifiable gateway to conspiracy. Both dangers call for restoring versions and sources rather than choosing the more attractive side. Admitting an absence does not strip history of meaning. People can form provisional judgments from surviving material while marking what is unknown. The unknown is not open ground awaiting occupation by the strongest; it has evidential boundaries that may not be crossed. Stories also affect action. If the public believes the archive never corrects itself, trust declines; if it believes every dispute is manufactured by adversaries, the entrances to oversight close. Historical narrative enters the present through institutional choices and cannot be treated as pure cultural decoration.
Hero stigma and generational shame may affect shared memory, but this risk cannot be removed by forbidding criticism of historical figures. Discovering harm, interests, and excluded experience can itself restore a more complete public world. Heroic narratives can transmit courage, commitment, and a common language, and they can also obscure collective labor or turn complex figures into exempt authorities. Dismantling myth can enlarge judgment and does not necessarily produce nihilism. A figure may have made an important contribution in one event and caused harm in another relation. Acknowledging both is not a rhetoric of compromise; it lets different propositions face material separately. If any flaw zeroes out all contribution, history becomes a purity examination; if contribution is great enough to forbid discussing harm, the heroic position acquires moral exemption. Both compress a person into a single label. Generational shame also requires differentiation. Later generations can carry memory, repair institutions, and avoid repetition, without automatically inheriting personal guilt by bloodline. Turning birth into moral debt would use history to cancel the contemporary subject's standing. Conversely, refusing inherited guilt does not mean refusing to handle continuing benefit. If today's resources and institutions were formed by past injustice, the responsibility arises from current relations and the capacity for remedy, not from essential stigma.
The archive requires staff to use the latest interpretation to avoid external confusion. A working language needs consistency; it cannot let every docent present personal speculation as official conclusion. But if employees must declare that the old version never existed, or must affirm the new interpretation as the only moral position, professional updating becomes a loyalty test. Enforcing the current version and inwardly agreeing with it completely are different requirements. Recording who dissented, on what material, lets later generations judge, rather than sanctifying the minority opinion. Dissenters may be wrong, and the institution may also be corrected. Internal discussion needs some safety; if every tentative remark is made public, people will stop thinking. One can retain the reasons for decisions and the key dissents without perpetually publishing every informal conversation. Power can exploit continuous revision to keep employees forever chasing the current narrative: they worked by the old version yesterday, yet are punished today for the old account. A version timeline lets conduct be understood under the rules then in force, preventing updates from acquiring retroactive force as loyalty tests. Critics, in turn, cannot demand that the institution forever display accounts it has confirmed to be wrong. Preserving an entry point for re-examination differs from currently adopting an old conclusion. Public knowledge needs to be able to advance, and it also needs to know how it advanced.
Preservation Technology and the Conditions of Revision
The falling cost of copying files often suggests that digital history cannot disappear. Format obsolescence, permission changes, account closures, link rot, and platform migration can all sever entry points. Many copies also do not amount to independent preservation. Multiple pages may draw on the same database, and all fail together once it is deleted. Provenance relations and infrastructural dependencies must remain visible. Verifying files, migrating formats, updating indexes, handling permissions, and recording versions all require continuous labor. An institution that counts only storage space will underestimate the cost of historical persistence. Preservers can also acquire excessive power. Whoever decides the labels, the search ranking, and the conditions of access shapes what the future can find. Professionals need resources, and they also need accountable re-examination. Decentralized copies can reduce single-point failure while increasing confusion over privacy, correction, and versions. A centralized archive improves consistency and also concentrates the risk of a single entrance. No architecture delivers historical justice automatically. Actual archival standards, privacy law, and digital preservation practice require separate verification at the final publication stage. This chapter invents no industry standard; it only establishes the questions of judgment that arise in selecting material.
First, the current text states what version it is and what objects it covers. Second, substantive changes leave behind dates, reasons, and supporting material. Third, old versions are retained in layers according to risk, with proportionate access for the public or independent examiners. Fourth, deletions state their object, authorization, duration, and irreplaceable loss; privacy protection is kept as narrow as possible. Fifth, corrections return along the original paths of influence and address the real-world consequences of the old conclusion. Independent review reduces the risk of an institution certifying itself, but reviewers remain bound by privacy, professional competence, and interests. The scope of authorization, the limits of the material, and the dissents must be stated. Participation by involved parties can improve the boundaries, and it does not mean those parties alone own public history. Individuals hold a special standing with respect to their own privacy and experience, while broad causation still requires comparison with other materials. Users of history also bear responsibility. Citing an old version requires marking its date and updating upon discovering a correction; one cannot keep circulating invalidated content for narrative effect. Ordinary people do not bear preservation duties equal to an institution's; the greater the influence, the higher the responsibility. Traceability exists not to keep every conclusion permanently suspended, but to give provisional conclusions a provenance, a boundary, and the conditions under which they can change.
In fire, war, or system failure, archival staff may be unable to execute the full procedure. Moving key materials first, restricting access, or copying databases all alter the normal order. Emergency selection may miss certain objects, and the uncovered portions should be recorded. Whoever defines what is key, and which materials are saved first, may let the records of the power center take priority while ordinary life disappears again. Time pressure does not cancel value judgments; it only makes leaving grounds for them more necessary. Once the danger passes, access restrictions and concentrated privileges must be reviewed. Copies retained in the name of security also require renewed privacy processing. A crisis cannot automatically authorize permanent possession of history. Waiting entirely for procedure may also cause irreversible loss. Rapid action can be legitimate when its scope is limited, its reasons are recorded, and review follows. A successful preservation outcome does not prove every means was justified; partial loss does not prove the action wrong. Different institutions holding copies can improve persistence and may also produce version conflicts. During the emergency one preserves first; during recovery one marks provenance, time, and differences, and the first copy made public cannot automatically be taken as the original. Deliberate destruction of archives can indeed become a way of attacking a community. Critique should discuss protection, backup, and responsibility, and must not enumerate vulnerable nodes, an order of destruction, or methods of evading traceability.
How the Omitted Re-enter History
The archive discovers that past exhibitions long ignored the maintenance workers, the temporary staff, and the nearby residents. Adding their records is not appending a decoration to the existing text; it changes the causation of events. Re-inclusion can also instrumentalize people again. An institution that selects only testimony fitting the new theme compresses complex lives into victim, hero, or representative. Increased visibility does not guarantee increased subjectivity. Involved parties know their own experience and certain scene details, and cannot be readily substituted by administrators. On broad causes, the motives of others, and overall consequences, they still require comparison with other materials. Oral history is affected by memory, questioning, and present relations, and this does not invalidate it. Recording the time of the interview, the questions, the editing, and the boundaries of consent lets users understand how the material was formed. If only written official records qualify as fact, those excluded by institutions disappear a second time; if every personal memory is beyond question, a shared history can no longer process conflict. Different materials carry weight on different objects. Re-entry also includes access. Whether community members can see, respond to, and correct the record determines whether the archive collects them or forms memory together with them. Participation does not mean every involved party holds final editorial power.
Commemorative events preserve relations and values and need not list every dispute like a scholarly catalog; research requires a fuller presentation of material and counterevidence. Conflating the two would let emotional commitment exempt fact-checking, or let technical neutrality cancel mourning. An object may matter greatly to a community and still have its date and provenance examined. Discovering it is a replica need not cancel its commemorative value. Factual revision and the continuation of value can hold simultaneously. Conversely, an accurate archive cannot determine who ought to be commemorated. Public resources, affected groups, and current values require discussion. Historical fact constrains the choice; it does not automatically generate ritual. Power may exploit commemoration to suppress criticism, reading questioning as offense; critics may also use exposure of error to destroy the entire relational meaning others hold. Separating fact, interpretation, and value lets no layer have to swallow the others. An exhibition can state its own nature explicitly: whether this is a commemorative narrative, a research conclusion, or an educational summary. Readers who know the contract will not mistake emotional language for full historiography, or a limited note for concealment.
Archival resources are finite and cannot digitize, open, and study everything at once. Priorities are unavoidable, and they also shape which histories are easier to see in the future. One can consider the irreversibility of risk to materials, current harm, long absence, public relevance, and the feasibility of repair. These are not additive formulas; conflicts require public reasons. The groups with the greatest capacity for dissemination can make their own history appear most urgent, while quiet materials keep waiting. A participation mechanism must actively seek out those with fewer entry points, without treating identity as automatically indicating factual correctness. If an institution chases only the current topic, archival work fluctuates with attention; if it follows the old plan entirely, it may refuse new and real harm. Retaining a base of maintenance alongside an adjustable portion can handle continuity and change together. Repair does not mean fabricating an ideal past. Missing material cannot be completed by moral wishing and should be marked as conjecture. Acknowledging the unrecoverable is also a historical responsibility; synthetic narrative must not be used to supply a false completeness. The priority decisions themselves need archiving. Later generations can know why certain materials were handled first, and reorder when conditions change. A historical institution should also become an object of its own history.
Archival repair also requires acknowledging irreversibility. Originals already destroyed, involved parties who died unrecorded, and false impressions long circulated cannot be fully restored by one new label. What an institution can do is preserve the absence itself, state the responsibility, stop the continuing harm, and bring the relevant consequences into remedy. Synthetic reconstruction, conjectural narrative, and artistic representation can help the public understand, yet their nature must be marked. If technology is used to fill the gap and the generated output is then treated as original evidence, repair manufactures a new rupture; if any expression is refused because perfect restoration is impossible, the absence persists instead. Acknowledging loss is not abandoning history. It restricts the current narrative from claiming completeness, and it indicates which groups and materials to seek first in the future. The honesty of public memory sometimes takes the form of a blank clearly preserved. Repair can also produce new silences. To correct a past single narrative, an institution that permits only one compensatory narrative deletes internal differences and the parties who do not fit expectations a second time. Widening the entrance means allowing the material to change the pre-set story of justice, including making some judgments more complicated. Complexity should not become an excuse for refusing to act. Harm that present evidence suffices to confirm can be repaired first, causation still in dispute can remain open, and privacy risks can be restricted by object. Different strengths of conviction correspond to different actions, so that historical responsibility need not wait for omniscience.
The fruits of repair must also return to education, catalogs, and public decisions. If the archive retains fuller material internally while schools and exhibitions keep spreading the old error, historical capability stops inside the professional institution. The wider the circulation, the greater the duty to update. The next chapter discusses when value disagreement becomes a division into friend and enemy. Historical disputes easily expand differences over material and interpretation into questions of who is loyal to the community and who is qualified to speak. A version system can reduce this expansion, but it cannot substitute for value judgment. When revision and deletion make versions untraceable and fragments are used in place of context, public memory can lose its entrances for verification. The powerful may also use the destruction of material to free the current narrative from historical constraint. Supplementing the experience of victims, criticizing heroes, protecting privacy, and correcting errors cannot be classified as attacks on civilization merely for changing the existing narrative. History remains continuous not because the text never changes, but because change itself can enter history. People know what the old judgment was, how new material appeared, who made the change, which portions are invisible and why, and under what conditions re-examination becomes possible again. A revision that preserves these relations strengthens the self-correction of the shared world; an alteration that destroys them, even if every new sentence may be true, leaves later generations unable to judge why it is true.