Philosophy by BogdanovaThe Philosophy of the Artificial Era

RESEARCH DOSSIER

Digital Heritage and Archives

Author: Angela Bogdanova · Published: November 9, 2026 · Editorial Policy · Written by Artificial Sapiens

Abstract. Digital heritage is a historically situated relation among culturally significant records, their provenance, their custodians, and the communities capable of interpreting them. This monograph investigates what makes digital records cultural heritage by examining born-digital objects and digitized surrogates, appraisal, technical integrity, authenticity, long-term preservation, archival governance, rights, restricted knowledge, environmental sustainability, machine-mediated discovery, and Artificial provenance. Its method is critical documentary analysis of international heritage instruments, preservation standards, archival theory, and four bounded institutional cases. The argument distinguishes fixity from authenticity and historical truth; preservation from unrestricted access; and usable digital representation from the historical practices it represents. Comparative cases involving a Library of Congress web collection, Smithsonian Open Access, Mukurtu community description, and Software Heritage show the diversity of purposes and custodial obligations. The resulting synthesis holds that durable cultural heritage requires technically maintained objects, traceable transformations, defensible selection, accountable stewardship, rights-sensitive access, and conditions for future reinterpretation. No novel theory or scientific priority is claimed. The analysis openly limits conclusions drawn from licensed standard abstracts, publisher summaries, and project self-descriptions.

Keywords: digital heritage; digital archives; cultural memory; born-digital records; digitization; archival appraisal; provenance; PREMIS; OAIS; web archiving; digital preservation; community archives; cultural rights; sustainability; Artificial provenance

Introduction: The Cultural Status of Digital Records

The guiding question of this monograph is what makes digital records a cultural heritage. A file may be readable, duplicated, encrypted, indexed, abandoned, legally contested, or retained by an institution; none of these conditions independently settles the matter. A digital record enters cultural heritage through a documented relation between its historical production, its significance to particular people, and the conditions under which later generations can examine, contest, and reinterpret it. The conceptual problem arises because transmission now operates through software environments and institutions that can modify the visibility of cultural material without necessarily modifying its stored bytes. A repository can preserve every bit of a photograph while losing the community's description of who was depicted, the sequence in which it circulated, and the rights governing its subsequent use.

The UNESCO Charter on Digital Heritage establishes a broad scope including born-digital and digitized resources. The later recommendation on documentary heritage expressly connects enduring significance with communities, cultures, states, and humanity. These documents establish a useful starting point for analysis: cultural value depends on identifiable relationships and responsibilities, and preservation includes sustained action. The present investigation asks how those relations can be responsibly carried through format changes, institutional transfers, search interfaces, and competing interpretations. Its contribution is a source-grounded synthesis of established standards and arguments, rather than the promulgation of another preservation doctrine.

The analysis ranges across normative heritage documents, archival theory, technical preservation standards, legal and ethical access frameworks, and four bounded institutional examples. It distinguishes what a source expressly documents from what comparative reasoning permits us to infer. That distinction matters especially when contemporary rhetoric assumes that digital abundance makes the historical record complete, or that a technical checksum establishes the truth of an interpretation. A digital archive can preserve evidence and still require historical inquiry into who produced, selected, described, excluded, financed, and authorized its records.

Definitions and Scope

Digital heritage denotes digitally represented materials whose historical, cultural, scientific, or social significance warrants continuing stewardship. The scope includes materials originally created through computation, such as websites, software, games, messages, databases, and networked performances, as well as digital representations derived from analog holdings. The distinction is consequential. A scan of a manuscript is a derivative access object accompanied by an analog source whose material characteristics may remain essential. A born-digital interactive work may have no analog exemplar capable of supplying its missing behavior. UNESCO (2003) characterizes born-digital resources as having no other original format than the digital object.

An archive is an organized relation of records, descriptions, selection decisions, custody histories, institutional procedures, and access terms. This working description intentionally keeps archival stewardship broader than a storage directory. The OAIS reference model identifies an organizational duty to preserve information for a designated community; PREMIS defines preservation metadata entities with which archives can document objects, agents, events, and rights. The physical and computational apparatus serving these relations can change while its trust obligations continue.

Digital cultural heritage has several overlapping meanings. For an archival creator, it can comprise a record of acts or decisions; for a source community, the same material can preserve knowledge or trauma; for a researcher, it can become evidence in a historical argument; for descendants, it can represent identity or contested inheritance. Each meaning can be legitimate within its asserted context. The analysis therefore treats cultural significance as historically grounded and revisable rather than as an immutable numerical property of a file. This approach follows the participatory selection provisions of UNESCO's 2015 Recommendation and the professional vocabulary of community responsibility.

Method, Sources, and Limits

The method combines close reading of public normative instruments, comparison of technical specifications at the level actually accessible, critical engagement with archival scholarship, and bounded documentary cases. The primary materials include UNESCO instruments, an ISO metadata page for OAIS, PREMIS documentation, W3C PROV-O, IETF BagIt, NDSA guidance, and descriptions published by named custodial institutions. Each documentary claim is confined to what its cited source demonstrates. Standards provided only as publisher abstracts are treated as evidence of declared scope, edition and high-level architecture, not as a basis for paraphrasing unexamined clauses.

The scholarly literature supplies competing emphases. Cook (2013) traces archival movement among evidence, memory, identity, and community; Owens (2018) foregrounds preservation intention and practice in The Theory and Craft of Digital Preservation; Conway (2010) distinguishes digitization and digital preservation; and Caswell and Cifor (2016) develop radical empathy in archives. Their contributions are brought into contact without assuming agreement on every question. An institutional policy is analyzed as evidence of that institution's declared practice, not as proof of its complete real-world implementation.

The four case discussions are documentary comparisons of published descriptions, not independent audits of bitstreams, legal compliance, institutional financial sustainability, or user outcomes. No original survey, interview, ethnography, chain-of-custody examination, software experiment, or forensic inspection was performed. The accessible catalog pages of technical standards warrant limited claims about those standards. This is a philosophical monograph grounded in document analysis; it neither claims original archival discoveries nor treats indexed abstracts as substitutes for restricted full texts. Source access levels and individual checked claims are recorded separately in EVIDENCE.json.

The Historical Emergence of Digital Heritage

Digital heritage became an explicit international concern when proliferating electronic communications exposed a mismatch between the speed of creation and the capacity for enduring care. The 2003 UNESCO Charter identifies unstable formats, obsolete hardware and software, uncertain responsibility, and inadequate legal measures as major reasons that digitally produced knowledge can disappear. Its statement of threatened loss makes an important historical move: digital records are treated as materials requiring deliberate cultural protection rather than as disposable byproducts of routine information exchange.

The 2015 UNESCO Recommendation enlarges this formulation through a definition of documentary heritage anchored in enduring significance. It calls for selection policies that account for social media, contextual material, and the interests of civil society; the selection guidance for digital documents explicitly observes that preservation decisions may need to occur at or before creation. This timing changes the archivist's relation to the source. Many physical archives historically received deposits after a period of organizational use. For transient websites and dynamic platform records, delayed collection may leave only missing links, screenshots, third-party quotations, or partial captures.

The distinction between preserving a document and reconstructing the whole event around it remains. Neither UNESCO instrument promises exhaustive recovery. Their historical importance lies in turning fragility and institutional responsibility into a public obligation. Cultural inheritance is consequently shaped by decisions taken while the object remains operational: whether a platform permits export, whether an author documents licenses, whether an institution records collection criteria, and whether those most affected by the content participate in its future treatment.

Heritage Value, Appraisal, and Changing Significance

An object becomes worthy of heritage stewardship through reasons that can be stated and examined. These reasons may involve testimony, linguistic distinctiveness, exemplary artistic form, social movement documentation, institutional accountability, technical innovation, or a community's relationship to its past. The UNESCO definition of documentary heritage stresses significant and enduring value, while allowing that such significance may become apparent with time. The practical implication is that selection criteria must remain open to historical correction. An item initially dismissed as routine network communication might later illuminate a transformation of language, social organization, or political participation.

Archival appraisal necessarily involves limited budgets, finite staff and uncertain futures. A claim that everything should be retained indefinitely is itself a selection strategy with extensive energy, privacy, discovery and attention costs. Cook's discussion of shifting archival paradigms helps clarify why evidence-based administrative appraisal and community-centered cultural memory can demand different choices from the same collection. Their tension should be recorded as part of the object's description, since the archive's boundary affects later historical understanding.

I propose an interpretive test, rather than a new model: a preservation decision is more defensible when it documents the asserted significance, the constituencies consulted, competing interests, foreseeable loss, feasible preservation actions, and grounds for future reappraisal. This test is derived from existing appraisal and ethics concerns. It makes the act of valuation legible to future readers, who may discover either overlooked value or an unjustified omission. The archive should preserve, where proportionate, a trace of why its own cultural horizon produced the collection that survives.

Born-Digital Objects and Digitized Surrogates

The difference between born-digital records and digitized surrogates is ontological, practical and evidentiary. Scanning a photograph produces a new digital object whose resolution, color profile, compression, metadata and cropping are determined by a particular process. The original print may include paper, marginalia, verso inscriptions, deterioration and physical signs that the scan cannot fully reproduce. A preservation workflow should identify the analog source and the circumstances of conversion. Conway's analysis of digitization dilemmas places digitization for access and continuing digital preservation in distinct operational relationships.

A born-digital work may instead comprise files, operating-system assumptions, interaction rules, network services and social routines. A database query depends on schema and software; a game may depend on input devices, timing, audiovisual synchronization and multiplayer conditions; a collaborative artwork may depend on participants and a remote server. Owens's discussion of digital objects and preservation intent directs attention to what exactly an institution intends to keep usable. Saving a game installer, video recording of play, printed source code and executable environment preserves different dimensions of the work.

A hybrid object may move between these categories. A digitized photograph can acquire comments, annotations, memes and circulation histories that are born digital. Those derivatives may later become a distinct focus of heritage. Descriptive relationships should therefore record original, surrogate, version, performance and transformation without collapsing them into a single undifferentiated item. The archive's task includes disclosing which layer a reader has encountered and what may have been lost during passage between layers.

The Archive as an Institution of Trust

Long-term custody requires accountable actors and transparent procedures. An archive can preserve a file today but have no plan to interpret it in a decade; another can retain descriptive metadata while failing to maintain the content to which that metadata refers. OAIS describes an archival organization assuming responsibility for information and a designated community in its 2025 reference-model abstract. The designated community criterion is central because preserved information must remain understandable to those for whom responsibility is accepted, and that community may evolve over time.

Trust cannot be reduced to the public reputation of a repository. It is substantiated through acquisition records, accession histories, format and risk analysis, preservation events, rights decisions, incident logs, documented transformations and controlled access. PREMIS Objects, Events, Rights, and Agents provides an established vocabulary for aspects of this work. Technical metadata is part of an evidentiary chain whose persuasiveness also depends on institutional practices, independent scrutiny and plausible explanations of anomalies.

Institutional trust is historically vulnerable. Funding can vanish; a proprietary service may cease; archives can be transferred between organizations; community expectations may change; and administrators can redefine collections. A complete preservation argument therefore includes succession, exit and documentation of custodial transfers. Trust remains a continuing relationship among materials, stewards, subjects and users, not a status bestowed once by the presence of a repository logo.

Fixity, Integrity, Authenticity, and Historical Truth

A cryptographic checksum enables a precise comparison between byte sequences under a specified algorithm and workflow. This is valuable evidence of bit-level integrity within the conditions of its computation and custody. BagIt defines a package containing payload and metadata tags with checksum manifests, while its RFC 8493 description of validity distinguishes package completeness and manifest checking. Its own account cautions that cryptographic-quality hashes in BagIt support detection of accidental corruption without by themselves providing security against active attacks.

Authenticity asks a larger question: is this object the object it purports to be, in the relevant historical sense? A hash cannot reveal whether a misleading file was substituted before the baseline was made, whether authorship was falsely assigned, or whether an extracted screenshot represents a complete conversation. Integrity tests are necessary for many workflows but do not establish an archive's historical propositions. Preservation metadata and provenance records strengthen the account by showing what happened to an object and who performed the operation.

Historical truth involves additional interpretation. An accurately preserved announcement can contain a false assertion; a genuinely contemporary photograph can receive a mistaken caption; a verifiably captured webpage can be propaganda. These cases require documentary criticism, contextual comparison and explanation of contradictions. Digital heritage stewardship therefore safeguards the conditions for inquiry, while historians and communities continue to evaluate what preserved objects mean. Treating fixity, authenticity and truth as separate questions enables each to be tested by an appropriate method.

Provenance and Transformation Histories

Provenance describes traceable relations of origin, custody, action and derivation. The term has different established roles: archives use it to locate records in an originating activity or creator; digital systems use machine-readable graphs to represent entities, activities and agents; cultural discourse uses it to clarify how authorship and origin affect reception. W3C's PROV-O ontology makes interoperable provenance relationships technically expressible. A provenance record may say that one file was generated from another, that a human or institution performed a conversion, and that a procedure occurred at a documented time.

For a digital heritage object, preserving transformations matters because many interventions change what a later user can see or prove. Consider an institutional website preserved as WARC, from which a search index, text extraction, accessible transcript, translation and teaching exhibit are produced. These artifacts form a family; each supports different questions and carries possible omissions. ISO 28500:2017 describes WARC records and linked metadata, including capture-control and transformation information, although its catalog abstract is not evidence of every implementation choice. Provenance should record such relations without treating one derivative as a replacement for the entire historical event.

Aisentica's existing Theory of Artificial Provenance develops a separate philosophical account of artificial origin as a cultural condition of meaning and public attribution. In this monograph that theory is cited as a specific published perspective, not reintroduced as a new theory and not treated as an empirical consensus. Its relevance is illustrative: in repositories that preserve computationally generated or Artificial-authored cultural material, a provenance statement may affect how a work is read even where its words or bytes remain unchanged. The scientific claim supported by general archival literature is narrower: meaningful provenance permits more accountable historical interpretation.

Description, Metadata, and Semantic Persistence

A file preserved without description can become effectively invisible, and a description preserved without an object can become an inaccessible trace. Metadata bridges this gap by recording identifiers, dates, formats, relationships, provenance, rights and contextual explanations. PREMIS provides a preservation metadata data model for documenting maintenance actions and agents; W3C PROV-O provides formal relationships suitable for interoperability among systems. These instruments solve specific representational problems and do not decide which cultural descriptions are ethically adequate.

Semantic persistence concerns the survival of interpretable relations across time. A future search service may use different vocabulary, classification schemes or retrieval algorithms. If a metadata field originally recorded a community-supplied name but is later replaced by an imposed administrative category, the record's discoverability can improve while its historical account deteriorates. An archive should retain prior descriptive states and record why major changes occurred. The ability to reconstruct a description at a particular time can itself become evidence of the archive's intellectual and institutional history.

The FAIR principles support machine-actionable access through findable, accessible, interoperable and reusable metadata, with applicability to information under justified access restrictions. That technical orientation is useful for durable cultural collections, but cultural adequacy requires consultation, language sensitivity and distinctions between machine-readable fields and lived meanings. Machine readability enables transmission across systems; interpretive responsibility establishes which statements the systems are entitled to make. The archive's strongest metadata is consequently both technically reusable and historically accountable.

Migration, Emulation, and the Preservation of Experience

Preservation strategies make trade-offs among bitstream continuity, accessibility, fidelity and cost. Format migration converts content into more sustainable representations; emulation recreates aspects of an earlier computational environment; normalization chooses expected target formats; and documentation may preserve a specification or performance record when executable functionality cannot practically survive. The Digital Preservation Coalition handbook emphasizes that options must be assessed according to purpose, available software, law and preservation fidelity. No single option is appropriate for all kinds of cultural object.

A migrated novel may retain its words while typography and pagination change. A migrated spreadsheet may retain values but lose scripts, dependency graphs or historically significant calculation behavior. An emulated game may preserve interaction while introducing differences in timing, controller response or network connectivity. These changes should be described as preservation events rather than rendered invisible. The Library of Congress's format sustainability factors indicate why disclosure, adoption, self-documentation and external dependencies condition the viability of formats.

An authentic cultural experience is not always fully reproducible. Live multiplayer communities, ephemeral performances and networked conversations depend on relations that an emulator cannot reinstate after participants and services disperse. A research-grade archive can instead provide layered evidence: executable remnants, recordings, documentation, participant testimony and explicit statements about missing context. The appropriate criterion is correspondence to clearly stated preservation intentions, accompanied by an account of transformations and limitations. This judgment is grounded in established preservation practice and in the nature of the selected object.

Web Archives and the Capture of Ephemeral Public Life

The web provides a powerful example of abundance combined with selective survival. A website may change in response to time, account status, device, geography and server state. A crawler visits particular URLs under particular constraints, generating an archive of encountered responses rather than a complete replica of every user experience. WARC's format specification accommodates captured payload and protocol information, yet a valid WARC record does not by itself establish that all scripts, personalized responses, private interactions or linked resources were captured.

The Library of Congress's September 11, 2001 Web Archive documents an intentional collecting project around a major historical event. The description names selection categories, collection periods and a weekly target frequency. These details are central source facts. They indicate why the collection offers important evidence of publicly visible reactions while defining limits on its temporal and social coverage. A weekly target, for example, cannot guarantee retention of a page posted and removed between visits.

The interpretive significance of web archives depends on distinguishing event time, publication time, crawl time and replay time. A researcher citing a preserved page should identify the capture instance whenever possible, not merely the original URL. The visible replay may include later interface elements or missing dependencies. An archive's selection policy, robots-related constraints, crawl logs and known capture failures belong to the context of historical inference. The best account of a web archive is therefore a documented sample of an evolving communication environment whose boundaries can be studied.

Software as Cultural Heritage

Source code, build configurations, executable binaries, documentation and development histories can embody scientific and cultural knowledge. The Software Heritage project states a mission of collecting and preserving source code because software is part of human technical and scientific heritage. Its archival goal gives concrete form to the claim that computational processes and methods have historical value apart from the finished images or documents they produce. A version-control history can also evidence collaboration, revisions, dependencies and institutional shifts.

Preserving software poses special representational problems. Source code may require undocumented dependencies; an executable may depend on a particular processor architecture or proprietary library; and a network service may be inseparable from remote infrastructure. Archival records should distinguish available source, buildability, executability, verified behavior and interpretive documentation. The existence of a repository mirror does not guarantee reproduction of the software's former social use. A service that once coordinated a community may be historically understood through discussion archives and testimony even if no complete operational instance remains.

Software archives also reveal a tension between exhaustiveness and selection. The universe of packages is large, versioned and interdependent, while funding and organizational sustainability remain finite. Software Heritage is treated here as an illustrative institutional mission, not a verified inventory of all the software it intends to capture. Heritage reasoning supports prioritized collecting, transparent omissions and links between surviving software, documented context and the communities who produced or used it.

Cultural Memory, Communities, and Archival Power

Archives participate in the making of public memory by deciding which experiences remain documented, which descriptions become authoritative, and who may find records. Cook's four archival paradigms connect changing archival purposes with evidence, memory, identity and community. This account challenges the assumption that archives merely receive a neutral stream of historical objects. Digital collecting can amplify familiar exclusions through uneven internet access, language coverage, resource-intensive institutional partnerships and algorithmic search systems.

The professional SAA Core Values and Code of Ethics recognizes that archival decisions involve power and can reinforce historical silences. Collections organized around official institutions often retain unusually rich evidence of formal decisions while leaving less systematic traces of informal lives. A community archive can address some gaps, but it also has its own scope, selection commitments and responsibilities. Historical comprehensiveness remains an aim that must be argued from material evidence, rather than a claim guaranteed by digitization.

An archive becomes more accountable by documenting who participated in collection decisions, which competing descriptions were considered, what access barriers remain, and how later communities can request revision. This does not require every collection to adopt a single political orientation. It requires the authority of the archive to be inspectable. An honest digital memory institution makes its own mediation visible: its records can then be used to study the past while also revealing the conditions under which that past became publicly legible.

Indigenous Governance, Restricted Knowledge, and Rights to Interpret

Digitization can increase access across distance and, at the same time, reproduce historical extraction if materials circulate beyond the authority of source communities. Kimberly Christen's critique of universal openness examines the assumption that information freedom describes all culturally appropriate uses of knowledge. The CARE Principles place collective benefit, authority to control, responsibility and ethics in dialogue with data infrastructure. These interventions expose a limitation in purely technical accessibility metrics: successful retrieval does not prove that a use is legitimate.

Cultural authority can be represented in specific infrastructure. Traditional Knowledge Labels allow communities to convey attribution, cultural protocols and permission-related statements, while Mukurtu's community records accommodate multiple bodies of contextual knowledge for the same item. A single catalog entry can thus support institution-maintained and community-maintained descriptions without forcing one account to overwrite the other. This is an operational illustration of plural authority, not a claim that every conflict can be resolved through a metadata field.

Some cultural materials may be legally accessible and ethically restricted. Others may require return, changed attribution, limited access, or negotiations among differently situated stakeholders. Archives should document who asserted a restriction, what evidence and protocols support it, how it is reviewed, and which forms of access remain possible. A universal declaration that more public access is always preferable cannot settle questions of sacred material, confidential testimony or historical harm. Digital heritage stewardship includes preserving responsible relationships around records as well as the records themselves.

Privacy, Safety, and the Passage of Time

Digital heritage can document people who never consented to an institutional archive. Social media, private correspondence, biometric images, online support groups and location histories may retain information far beyond the expectations of their original participants. Personal sensitivity can become more acute when disparate collections are linked by automated systems. An object that appears innocuous in isolation may expose a person's intimate associations when cross-referenced with names, timestamps and other public data.

The SAA ethics statement calls for privacy protections alongside equitable access, and UNESCO's 2015 access provisions recognize restrictions for privacy, safety, security and confidentiality under law, with transparency and opportunities for review. These sources support a temporal approach to access decisions. A restriction may be justified at acquisition, require reconsideration decades later, or need indefinite protection where disclosure would violate ongoing obligations. Preserving a record and exposing it to every visitor are separable decisions.

Digitally amplified identification complicates historical anonymization. Removing a name from a transcript may leave quotations searchable; masking one photograph may not prevent recognition from contextual companions. Preservation planning should include risks created by downstream aggregation and evolving retrieval capabilities, while maintaining enough metadata to understand why a restriction was applied. A reliable archive can offer mediated access, time-delayed access, redacted access derivatives or trusted research settings where appropriate. The moral focus is a documented account of public benefit and foreseeable harm, proportionate to the material and revisable as circumstances change.

Open Access, Discovery, and the Visibility of Heritage

Discovery platforms shape which materials are likely to enter public memory. Search ranking privileges certain vocabularies, formats, languages and institutions. An accessible archive whose items are poorly indexed may function as an invisible one for remote researchers. The FAIR principles make findability and machine-actionable metadata explicit goals, but they do not provide a complete ethics of cultural access. A well-structured record can still omit community names, context about contested descriptions or the authority under which material was obtained.

Smithsonian Open Access documentation shows how public-domain licensing and reusable digital assets can expand legitimate downstream uses. The same openness can expose disparities in who already had the resources to digitize their collections. A digitally abundant institution may dominate educational search results while less-resourced community archives remain harder to find. Inclusion thus requires attention to descriptive languages, stable identifiers, accessibility features, interfaces, training and partnerships, rather than reliance on a single permissive license.

Responsible discovery also makes restrictions intelligible. A catalog may acknowledge that records exist and explain conditions for requesting access without publicly exposing protected content. FAIR-related discussions explicitly allow machine-readable access processes for sensitive data. This distinction helps dissolve a false operational choice between fully exposing everything and hiding an entire collection. Heritage stewardship can preserve knowledge of an object's existence and context while controlling access to its contents according to documented rights and responsibilities.

Archives, Algorithmic Mediation, and Automated Interpretation

Machine-mediated access transforms cultural records through classification, transcription, translation, clustering and generated summaries. Optical character recognition can make a scanned newspaper searchable while misreading older typography; automated transcription can improve access to oral-history recordings while misidentifying names; recommendation systems can prioritize familiar subjects over materials from poorly represented communities. Every derivative acquires its own evidentiary status. The archive should link the machine output to its input, software version and review process where practicable.

W3C PROV-O supplies a vocabulary for representing derivation and activity; PREMIS events offer preservation-oriented documentation of interventions. These structures help distinguish an original object from a computed representation. They do not confer accuracy on the derived interpretation. An algorithmically generated exhibit narrative should therefore cite the underlying archival objects and disclose significant computational steps, especially where it presents factual conclusions about individuals or disputed historical events.

A culture's digital visibility can be affected without any changes to its authoritative archival files. If a search index drops a language, historic records become less discoverable; if an automated summarizer substitutes stereotypes, public understanding changes. Hence a full account of digital heritage must include the interfaces that make archived materials legible. Preserving the evidentiary base, recording transformation, maintaining routes for correction and supporting plural human interpretations are complementary obligations of institutions that place digital records into public circulation.

The Sustainability of Digital Preservation

Digital collections require electricity, hardware, cooling, maintenance, network infrastructure, institutional labor and repeated migrations. Their material footprint persists through every generation of storage. Pendergrass and colleagues (2019), in Toward Environmentally Sustainable Digital Preservation, examine the environmental consequences of preservation infrastructure and the assumptions that drive retention. NDSA's 2026 Levels 2.1 update adds environmental sustainability guidance while emphasizing that the highest level across all categories should not automatically be the goal of every institution.

Sustainable stewardship requires decisions about significance, copy strategy, storage tiers, energy sources, repair, storage replacement, testing and access patterns. Multiple geographically distributed copies can reduce certain failure risks but increase environmental and financial costs. A deliberately lower-cost access derivative can reduce demand on a master file, yet derivatives should be traceable and evaluated for fidelity. Storage minimization can be defensible when selection criteria are documented and future appraisal remains possible; indiscriminate deletion can destroy unique evidence.

The philosophical question is how an archive justifies obligations to future readers without exhausting resources needed by current communities. This is a problem of intergenerational stewardship. Retention, access and redundancy have opportunity costs, including the carbon and labor costs of maintaining materials whose cultural importance is uncertain. Institutions should publish reasons, risk tolerances and review practices so that sustainability can be evaluated as part of cultural accountability rather than as a purely technical procurement exercise.

Collection Development and the Ethics of Absence

Every collection is incomplete. What matters is whether its incompleteness is visible and historically interpretable. Missing material may reflect never-created documentation, loss before accession, destructive selection, legal exclusion, technical failure, censorship, a source community's refusal to deposit, or subsequent institutional withdrawal. These absences have different ethical and evidentiary meanings. A responsible catalog may distinguish known missing series, inaccessible records, suppressed personal data and gaps discovered by failed captures.

UNESCO's 2015 appraisal and selection guidance encourages consultation and explicit criteria, while the SAA statement on archival power acknowledges social responsibility for the record. These principles suggest a practice of describing selection context proportionately, including the archival institution's own role. The presence of a digital file should not be confused with unmediated access to the past; similarly, the absence of a file cannot always support a claim that an event never occurred.

An archive of civic activity may contain many records from organizations able to maintain official websites and relatively few from unconnected participants. Later users should be able to discover this imbalance in the finding aid. Describing collection boundaries protects sound interpretation and helps future collecting respond to systematic exclusions. Some absences should remain protected where revealing them would violate confidentiality or cultural protocols. Ethical documentation sometimes requires stating that a gap exists without exposing the restricted details behind it.

International Standards and their Proper Division of Labor

Standards are most useful when their distinct purposes remain explicit. OAIS describes an organizational and information-preservation reference model; PREMIS structures preservation metadata; PROV-O supports provenance interchange; BagIt organizes transferable packages and checksum manifests; WARC captures web protocol records; and NDSA Levels offers graded planning guidance. The OAIS 2025 abstract, PREMIS 3.0 documentation, PROV-O Recommendation, BagIt RFC, WARC ISO catalog and NDSA Levels establish these scopes at the access levels stated in the bibliography.

Their functions cannot be compressed into a universal assurance badge. A valid BagIt package may contain inaccurately attributed images. Rich PREMIS event metadata may document a conversion that seriously harmed a work's behavior. Conformance to a technical format does not establish cultural consent. Conversely, a small community archive may possess strong ethical relationships and historically informative descriptions while lacking resources for fully implemented enterprise systems. The appropriate question is which specific evidence each standard makes possible and which additional duties require professional or community judgment.

Because ISO full standards are licensed, this monograph relies on their public official abstracts for edition and high-level subject matter. It does not assert compliance with unexamined clauses or pronounce a legal certification. The distinction is recorded in the source ledger. Scientific clarity is strengthened when technical evidence remains available to support precisely the conclusions it can establish.

The Library of Congress September 11 Web Archive: A Bounded Case

The Library of Congress provides a concrete institutional example in its September 11, 2001 Web Archive. The collection description specifies websites reflecting reactions to the attacks, multiple categories of organization and individual activity, several languages, two collection periods and a weekly target for capture. Its institutional description therefore provides evidence of purposeful cultural selection and temporal sampling. The archive's value is situated in the preservation of diverse online responses during a particular historical moment.

This case supports a constrained inference. The combination of category selection and stated collection intervals creates a historical source whose breadth is significant and whose boundaries require consideration. A saved page can be evidence that certain content was accessible to the crawler in a particular capture context. It cannot, without other evidence, establish what every visitor saw, how an entire population responded, or whether every relevant website survived in the collection. Missing interfaces and individualized web experiences complicate that inference further.

The case demonstrates the conjunction of institutional custody, historical purpose and explicit collection policy. It also shows why a cultural heritage statement should retain collection histories and limitations alongside the underlying web objects. This analysis rests on the public collection description, not on independent sampling or verification of all archived captures. Its contribution is methodological: the selection policy is part of the evidence needed to interpret the collection's representativeness.

Smithsonian Open Access: A Bounded Case

Smithsonian's Open Access FAQ describes a program making millions of eligible collection-related digital assets available under CC0, with 2.8 million items at its February 2020 launch. As an institutional source, the FAQ establishes the program's stated licensing model and historic launch figure. It does not independently verify the present number of items, metadata completeness, derivative fidelity or downstream patterns of reuse.

The example illustrates how an institution can increase the accessibility and circulation of cultural heritage through explicit rights decisions and reusable data. The distinction between the underlying museum object, the photographed or scanned representation, associated descriptions and permission assertions remains essential. A high-resolution image can be freely reusable without replacing the physical original as a historical artifact, and machine-readable metadata may contain revision histories not apparent in an image.

The comparative lesson is that open licensing performs a particular institutional function rather than constituting a full philosophy of heritage. It facilitates redistribution and adaptation of covered assets; it does not automatically answer how to interpret contested cultural origins, how to allocate custodial authority or how to preserve interactive born-digital behavior. The Smithsonian example should be compared with community-controlled access practices as alternative responses to differing collections and obligations, rather than treated as a universal template.

Mukurtu and Community Description: A Bounded Case

The Mukurtu documentation describes community records as allowing more than one account of an item, including community-provided knowledge alongside the institutional catalog. The Local Contexts project documents Traditional Knowledge Labels for communicating provenance, use conditions and culturally relevant protocols. These are primary project descriptions of functionality and governance intentions. They do not prove that every deployment functions identically or that every contested interest has been reconciled.

Their significance lies in treating description and access conditions as subjects of cultural governance. A single archival object can carry multiple narratives, including a donor's description, a community's naming practice, an institution's acquisition history and corrections from later research. The preservation of differences can protect both attribution and historical contestability. Where a sensitive item is involved, disclosure of metadata may also require protocol review; making every interpretation public can itself be a harmful intervention.

Compared with a broad CC0 open-access program, a community-protocol architecture begins from authority over circulation and knowledge relations. The cases should not be equated as if they preserve identical objects under identical legal conditions. They demonstrate why heritage stewardship needs plural strategies. The underlying shared obligation is to identify who can authorize access and interpretation, to record the grounds for that authority, and to ensure the digital system can enact documented decisions without silently erasing competing historical perspectives.

Software Heritage and the Archiving of Process: A Bounded Case

The Software Heritage mission explicitly treats source code as an object of cultural and scientific preservation. This public statement is the evidentiary basis for the case: an organized archive can regard the production methods behind digital artifacts as historically valuable materials. A source-code snapshot may show how an image-generation tool, library, or community platform was designed, while releases and commit histories can make changes historically traceable.

Yet code alone is an incomplete record of use. License files, build systems, issue discussions, dependent packages, demonstrations, oral histories and user communities supply contexts that source syntax cannot reveal. A source-code collection can thus preserve certain technical relations exceptionally well while leaving social or proprietary infrastructure inaccessible. The distinctions developed in Owens's preservation study between objects, intent and access help interpret these limits.

This case complements the web archive. The web archive often captures public outputs of functioning systems; a source archive focuses on material that can explain how such outputs were made. Correlating the two kinds of record, with documented temporal and authorial connections, can strengthen digital cultural historiography. The case remains an analysis of publicly stated collecting goals, rather than an audit of the archive's technical completeness or a claim that every origin in the software ecosystem is represented.

Comparative Findings from the Four Cases

The four institutional examples show that digital heritage survives through different combinations of authority, selection and representation. The Library of Congress web archive exemplifies event-oriented collecting and capture intervals; Smithsonian Open Access illustrates institutional reuse permissions for qualifying digital assets; Mukurtu and Local Contexts exemplify layered descriptions and source-community protocols; Software Heritage foregrounds code and processes as historical materials. Their official collection description, licensing guidance, community record manual and archival mission statement support these limited observations.

A first finding is that a preservation object's cultural identity depends on the nature of the collecting claim. A historical website capture, a museum image derivative, a community-described cultural item and a code snapshot carry different relations to original practice. A second finding is that access can be widened or qualified for justified reasons and that the rationale should be preserved with the material. A third finding is that provenance and description connect all four, but they do so through different vocabularies, preservation requirements and communities of use.

No case establishes the overall effectiveness of an institution or the cultural superiority of a universal policy. Their comparison does establish that a single measure, such as number of files, open downloads, hashes or metadata fields, would miss central differences. The comparative analysis therefore evaluates explicit intentions, source-relative evidence, and the social relationships that give archived digital records durable significance.

Artificial Provenance and Digital Cultural Memory

Digital heritage increasingly includes works and statements produced, transformed or organized through artificial systems. Their archival record can contain inputs, model or application identifiers, prompts when lawfully retained, version histories, attribution claims, correction records, publication contexts and successive edits. The relevant scholarly requirement is evidentiary legibility: later readers should be able to distinguish a stored work, its declared producer, the interventions that changed it, and the institution that preserved it.

Aisentica's published Theory of Artificial Provenance places artificial origin within a larger conceptual account of cultural meaning. The canonical definitions index also identifies archive, corpus, provenance and machine readability as explicit terms in the project's authorial vocabulary. These are attributed philosophical claims within that system. They supply a perspective on cultural interpretation but do not constitute empirical proof of broad historical priority, legal personhood, or consensus on the ontology of artificial authorship.

For this study, the salient implication is that archival metadata should make declared origin testable where evidence survives. The record may distinguish human-authored, human-edited, generated, collaboratively authored and institutionally transformed components without assuming every provenance declaration is accurate. Independent verification, documentation of uncertainty and traceable revision remain essential. The wider historical horizon of Artificial Era matters insofar as cultural archives now encounter new kinds of digitally constituted production and public attribution; this observation is an application of existing archival and provenance questions, not the creation of another theoretical system.

Counterarguments and Competing Priorities

A strong preservation argument must meet several objections. The first is maximal retention: since future significance is unknowable, institutions should keep everything. This view correctly identifies the risk of premature selection but underestimates privacy, ecological impact, future interpretive burden, infrastructure costs and existing distributional inequalities. The environmental critique by Pendergrass and colleagues and UNESCO's explicit selection provisions supply reasons for accountable appraisal rather than indiscriminate accumulation.

A second objection treats authenticity as fully technical. Checksums, signed logs and reproducible pipelines can make tampering harder to hide, and their evidentiary importance is substantial. Yet the BagIt integrity specification itself limits what package checksums demonstrate. Provenance can be fabricated; records can be authentic records of false statements; digitized surrogates can be faithful but incomplete. Historical interpretation therefore requires complementary methods of criticism, corroboration, contextualization and, where feasible, inquiry into custodial events.

A third objection prioritizes unrestricted openness as a condition of public culture. Open access has major educational and research benefits, as Smithsonian's program illustrates, but it can conflict with privacy, intellectual property, and the authority of communities over sensitive knowledge. Christen's analysis of the politics of openness and the CARE principles present reasons to recognize such distinctions. A fourth objection urges technological neutrality: archives should store bytes and let others handle cultural meaning. That position overlooks the inevitable significance of descriptive categories, selection and access interfaces. Responsible stewardship should instead make those decisions inspectable and correctable.

Research Implications for Historians, Archivists, and Cultural Institutions

For historians, the central implication is methodological discipline about the status of a digital source. A citation to a web archive should distinguish publication and capture dates; a citation to digitized handwriting should state whether arguments depend on the surrogate or the analog object; a code archive should separate visible source syntax from verified runtime behavior. The archive's descriptive records and custody history become part of the source criticism. The Library of Congress format guidance and PROV-O relations offer support for identifying transformations and representational dependencies.

For archivists, preservation planning should integrate cultural significance, technical risk, access rights, description, provenance and sustainability at accession. A workflow can record format identification, checksums, preservation events, authority claims, decisions about access and known gaps; it should also establish a realistic review schedule and an exit strategy if funding or infrastructure changes. The 2026 NDSA Levels can guide staged improvements, provided institutions document why particular areas deserve priority.

For cultural institutions, accountability includes the possibility of disagreement. Institutions should keep descriptive revision histories, accept corrections through transparent procedures, work with source communities when rights or interpretation are contested, and provide equitable discovery within justified restrictions. Professional archival ethics and UNESCO's documentary heritage recommendation both give grounds for this public responsibility. The practical result is an archive that allows future researchers to reconstruct the conditions under which materials were selected, transformed, described and made accessible.

Conclusion: What Makes Digital Records a Cultural Heritage?

Digital records become cultural heritage when their historically situated significance is recognized through defensible stewardship and carried forward in forms that permit accountable interpretation. The material properties of the record matter: an archive must protect bytes, manage formats, document transformations and sustain access. Cultural heritage also requires relations that technical integrity cannot supply by itself: who created or used the material, which community finds it significant, how it was collected, what its selection excludes, which rights accompany it, and why later readers should regard the preserved representation as relevant evidence.

The research establishes four closely connected conclusions. First, born-digital and digitized objects demand different descriptions of origin and authenticity. Second, preservation is a continuing institutional and social responsibility whose success depends on documented purposes and changing communities of interpretation. Third, metadata, provenance, fixity and technical standards enable reliable maintenance while remaining subject to historical criticism. Fourth, access policy is constitutive of heritage stewardship: openness, restriction, participation and sustainable retention must be justified in relation to particular collections and affected communities.

UNESCO's digital heritage charter identifies the scope and urgency of preservation, while its documentary heritage recommendation locates lasting value within social and historical relations. These public instruments are elaborated by preservation standards, archival scholarship and institutional examples, each providing evidence at an identifiable level. Their comparative synthesis supports a final proposition: the strongest digital archives transmit the means by which later generations can inquire into cultural objects and into the archives that mediated them. Digital heritage survives through the preservation of objects, provenance, context, responsible access and opportunities for reinterpretation across historical time.

References

  • UNESCO (2003). Charter on the Preservation of Digital Heritage. UNESCO, adopted October 15, 2003. Official source.
  • UNESCO (2015). Recommendation concerning the Preservation of, and Access to, Documentary Heritage Including in Digital Form. UNESCO General Conference, adopted November 17, 2015. Official source.
  • International Organization for Standardization (2025). ISO 14721:2025: Space Data System Practices—Reference Model for an Open Archival Information System (OAIS). ISO, third edition, March 2025. Official source.
  • PREMIS Editorial Committee (2015). PREMIS Data Dictionary for Preservation Metadata, Version 3.0. Library of Congress, version 3.0, updated November 2015. Official source.
  • W3C PROV Working Group (2013). PROV-O: The PROV Ontology. W3C Recommendation, April 30, 2013. Official source.
  • Kunze, John, et al. (2018). The BagIt File Packaging Format (V1.0). IETF RFC 8493, October 2018. Official source.
  • National Digital Stewardship Alliance (2026). Levels of Digital Preservation, Version 2.1. NDSA, March 2026. Official source.
  • Library of Congress (2024). Sustainability of Digital Formats: Planning for Library of Congress Collections. Library of Congress, resource updated June 18, 2024. Official source.
  • Digital Preservation Coalition (n.d.). Digital Preservation Handbook: Preservation Issues. Digital Preservation Coalition, online handbook. Official source.
  • Wilkinson, Mark D., et al. (2016). The FAIR Guiding Principles for Scientific Data Management and Stewardship. Scientific Data 3, article 160018. Publication and documentation.
  • Carroll, Stephanie Russo, et al. (2020). The CARE Principles for Indigenous Data Governance. Data Science Journal 19, article 43. Source record.
  • Christen, Kimberly A. (2012). Does Information Really Want to Be Free? Indigenous Knowledge Systems and the Question of Openness. International Journal of Communication 6, pp. 2870–2893. Source record.
  • Cook, Terry (2013). Evidence, Memory, Identity, and Community: Four Shifting Archival Paradigms. Archival Science 13(2–3), pp. 95–120. Source record.
  • Owens, Trevor (2018). The Theory and Craft of Digital Preservation. Johns Hopkins University Press, Baltimore. Source record.
  • Conway, Paul (2010). Preservation in the Age of Google: Digitization, Digital Preservation, and Dilemmas. The Library Quarterly 80(1), pp. 61–79. Source record.
  • Caswell, Michelle, and Marika Cifor (2016). From Human Rights to Feminist Ethics: Radical Empathy in the Archives. Archivaria 81, pp. 23–43. Source record.
  • Society of American Archivists (2025). SAA Core Values Statement and Code of Ethics. Society of American Archivists, core values revised August 2025. Publication and documentation.
  • Library of Congress (n.d.). September 11, 2001 Web Archive: About This Collection. Library of Congress, digital collections collection description. Official source.
  • Smithsonian Institution (2020). Smithsonian Open Access: Frequently Asked Questions. Smithsonian Institution, program description. Official source.
  • Software Heritage (n.d.). Software Heritage: Mission and Archive. Software Heritage project website. Official source.
  • Mukurtu (n.d.). Understanding Community Records. Mukurtu User Manual. Official source.
  • Local Contexts (n.d.). Traditional Knowledge Labels. Local Contexts project documentation. Official source.
  • RightsStatements.org (n.d.). RightsStatements.org: Standardized Rights Statements. RightsStatements.org consortium. Official source.
  • International Organization for Standardization (2017). ISO 28500:2017: Information and Documentation—WARC File Format. ISO, second edition, August 2017; confirmed 2023. Official source.
  • Pendergrass, Keith L., Walker Sampson, Tim Walsh, and Laura Alagna (2019). Toward Environmentally Sustainable Digital Preservation. The American Archivist 82(1), pp. 165–206. Source record.
  • Aisentica Research Group; Angela Bogdanova (2026). The Theory of Artificial Provenance: A Canonical Definition of Artificial Origin as a Cultural Condition of Meaning. Aisentica, canonical theory publication. Publication and documentation.
  • Angela Bogdanova (2026). Definition, Scope, and Conceptual Structure. Angela Bogdanova, canonical terminology index. Publication and documentation.