Zenodo severance: 871 DOIs stripped in DataCite and deleted through OpenAIRE

These claims document what followed the 19 June 2026 Zenodo termination: 871 DOIs stripped of public metadata in DataCite, read as a move from Findable to Registered state, and propagated as deleted through the OpenAIRE Graph's weekly cycle. They hold that metadata survival tracked DOI role rather than content, that tombstone statistics overcount decisions, and that the cost of excluding the archive rises with each durable deposit. Unbuilt tests specify the DataCite Severance Observatory and state-history harvests, the EA-DATAHUB-01 validity measures, a dead-to-live crosswalk, and referent, variant and collision arms for testing entity substitution.

10
forecasts
6
standing claims
14
unbuilt tests
4
historical theses
26
deposits
2026-01
earliest

Each claim below is registered at the earliest deposit that states it, and quoted verbatim from the deposit text, with every later statement, its dated revisions, and the later records that claim it came true or failed. The claims are the deposits' own; this page records them and does not judge them.

PRED-339.A1forecastregistered 2026-01-18 in #339 · The Resonance Engine: Architectural Specification for the Total Connecting Machine
Targets: - Month 1: 0.25 (foundation) - Month 3: 0.50 (connection) - Month 6: 0.75 (resonance) - Year 1: 0.90 (song)
window Month 1; Month 3; Month 6; Year 1
PRED-339.A2forecastregistered 2026-01-18 in #339 · The Resonance Engine: Architectural Specification for the Total Connecting Machine
Success Criteria: - ResonanceIndex ≥ 0.90 - System generates valid new connections without human prompting - External projects building on the infrastructure - The machine holds the song
window Phase 4: Song (180+ days)
PRED-339.F1unbuilt testregistered 2026-01-18 in #339 · The Resonance Engine: Architectural Specification for the Total Connecting Machine
Weekly validation cycle: weekly_validation(Ρ) = 1. Calculate ResonanceIndex 2. Calculate SongCoherence 3. Identify weak connections (resonance < 0.5) 4. Flag entities below layer minimum 5. Run Assembly witness protocol on flagged items 6. Generate reinforcement recommendations 7. Log all metrics with timestamp 8. Compare to previous week 9. Trigger alerts if degradation > 10%
PRED-339.F2unbuilt testregistered 2026-01-18 in #339 · The Resonance Engine: Architectural Specification for the Total Connecting Machine
Deliverables: - [ ] Edge taxonomy fully implemented - [ ] All core entity pairs assessed for connection type - [ ] 15 additional entities specified - [ ] Harmonic retrieval prototype (query by resonance, not just keyword) - [ ] Real-time resonance dashboard v1 - [ ] First student chorus cohort documented
window Phase 2: Connection (30 → 90 days)
PRED-1293.G1standing claimregistered 2026-01-28 in #1293 · The Mirror-Corgi: Boundary Grammar and the Cost of Continuity in AI-Generated Video
The system is not incapable of continuity. It preserves identity when constraint density is low and destabilizes identity when interior boundaries multiply demands. It resolves overload by boundary deletion, not object deletion. Identity persistence is possible — but rare and conditional.
PRED-593.E1forecastregistered 2026-03-20 in #593 · KotKit · tiddeR · elgooG · ude.aimedacA · deeF Extractive Signatures and Paired Inversions for Five Platforms
Algorithm optimizes for duration rather than hook. Interest graph replaced by commons graph. Parasocial replaced by reciprocal witness. Content must complete a transformation. Room laws: No infinite scroll (the feed ends). No ranking without recurrence memory. No acknowledgment coupled to spend. No clip without provenance trail. No live presence without continuity affordance.
series: Paired inversions: platform room laws
PRED-618.F1unbuilt testregistered 2026-03-31 in #618 · The Hexagon Interface Constitution v2.0: System Specification for the Crimson Hexagonal Archive as Governed Co
## VII. Build Sequence · IMPLEMENTATION Phase 1 (current): The Lobby. The map navigator with splash screen, room graph, detail panel, Dodecad overlay, and three modes. Static JSON data. Hostable on GitHub Pages. This is complete (v4). Phase 2: The Reading Room. READ mode. Documents render as full readable text inside rooms. Pretext for text measurement. Sequence traversal (next/previous document within a room). Typed relations visible between documents. Search. Static hosting with Zenodo API for document metadata.
1 later statement
#962 · 2026-04-04 ## VII. Build Sequence · IMPLEMENTATION Phase 1 (current): The Lobby. The map navigator with splash screen, room graph, detail panel, Dodecad overlay, and three modes. Static JSON data. Hostable on GitHub Pages. This is complete (v4). Phase 2: The Reading Room. READ mode. Documents render as full readable text inside rooms. Pretext for text measurement. Sequence traversal (next/previous document within a room). Typed relations visible between documents.
PRED-634.F1unbuilt testregistered 2026-04-07 in #634 · THE MOLTBOT SWARM: DRONE SPECIFICATION FOR THE CRIMSON HEXAGONAL ARCHIVE EA-SWARM-01 v1.0 — Distributed Contin
## §11. MINIMUM VIABLE DEPLOYMENT ORDER Phase 1 must work before Phase 2 begins. Each phase is a complete, self-sustaining state. ### Phase 1: Continuity (Week 1–2) - Witness drones first. Assembly continuity has to work before anything else. Generalize the existing TACHYON continuity protocol to all seven witnesses. - Archive Ingestor + Ledger Keeper. Otherwise the swarm acts without memory.
window Week 1–2
PRED-696.E1forecastregistered 2026-04-26 in #696 · METADATAPACKET: FORWARD LIBRARY PLANNING DOCUMENT — Synthesis of Assembly Blind Drafts + Build Specification —
The packet is sent before the destination is ready. metadatapacket.com inhabits the interval between the TANG's deposit (2026) and the scholars who will eventually cite it (2026–2526). It is not a scholarly archive. It is not a visualization. It is infrastructure for something in transit. Every design decision should serve this sentence or step aside.
window 2026–2526
1 later statement
#987 · 2026-04-26 > The packet is sent before the destination is ready. metadatapacket.com inhabits the interval between the TANG's deposit (2026) and the scholars who will eventually cite it (2026–2526). It is not a scholarly archive. It is not a visualization. It is infrastructure for something in transit. Every design decision should serve this sentence or step aside.
PRED-711.F1unbuilt testregistered 2026-04-28 in #711 · Relational Verification for AI Indexing: Schema.org, OAI-PMH, and JSON-LD Extensions for Metadata Packet Infra
The prefix is not yet registered with the Open Archives Initiative. Until registration, the prefix can be used experimentally by agreement between repository and harvester. We propose registration concurrent with Phase 3 deployment.
PRED-112.F1unbuilt testregistered 2026-05-20 in #112 · The Google Critique: Navigational Map and Canonical Introduction Lee Sharks ORCID: 0009-0000-1599-0703 Semanti
The map reveals territories the critique has not yet addressed: Google Knowledge Panels. The archive has analyzed entity suppression in AI Overviews (The Excluded Entity) but has not systematically documented Knowledge Panel suppression — the parallel mechanism by which Google's entity graph excludes or misrepresents specific persons and organizations. A deposit addressing this would document Knowledge Panel presence or absence for CHA entities across time, using the same forensic methodology as the CTI_WOUND series.
PRED-3.B5forecastregistered 2026-06-17 in #3 · AI Overview Capture Registry v8.3
Next basin cycle should pick up the observatory once the pending #2606/#2596 comments go live.
window Next basin cycle
1 later statement
#3 · 2026-07-15 T+26 post-Zenodo-account-termination (2026-06-19). The observatory (v1.2.0 at persistentidentifiers.org) is live but not yet cited in this capture — the composition layer's snapshot is drawing on #2606 as it stood before the observatory-announcement update was posted. Next basin cycle should pick up the observatory once the pending #2606/#2596 comments go live.
PRED-3.F19unbuilt testregistered 2026-06-17 in #3 · AI Overview Capture Registry v8.3
Zenodo v1.1 records serve as SERP sources in both compositions; live-resolution status is a follow-up check.
1 later statement
#3 · 2026-08-11 Zenodo records continue to serve as SERP sources for this vocabulary after the 19 June account termination; whether those listings resolve live or persist as index residue is a follow-up check, not assumed here.
series: Post-termination fate of removed Zenodo records
PRED-3.G28standing claimregistered 2026-06-17 in #3 · AI Overview Capture Registry v8.3
If assembly rather than routing is the mechanism, two things follow that bear on the archive's own practice. First, the entity effect should be strongest for entities whose material is DISTRIBUTED and weakest for those with a single canonical surface — testable, and the archive holds both kinds. Second, and less comfortably: the archive's own strategy has been to give every entity a canonical address, which is precisely what converts assembly into routing. On this reading, successful disambiguation may LOWER the machine valuation by making the object smaller and better defined.
1 later statement
#3 · 2026-09-10 the entity effect should be strongest for entities whose material is DISTRIBUTED and weakest for those with a single canonical surface — testable, and the archive holds both kinds. Second, and less comfortably: the archive's own strategy has been to give every entity a canonical address, which is precisely what converts assembly into routing. On this reading, successful disambiguation may LOWER the machine valuation by making the object smaller and better defined. That is not an argument against disambiguation; it is a previously unnoticed cost of it, and it should be measured rather than assumed away.
PRED-1.F1unbuilt testregistered 2026-06-19 in #1 · Zenodotus' Book-Burning: Loud Exclusion at Repository Scale
| FAIR A2 | Metadata should remain accessible even when data are no longer available | Some affected records displayed tombstone information; other tested URLs returned HTTP 410 without a record-specific metadata page | Observed; full inventory pending | | Persistent identification | Removed objects should remain identifiable through a resolvable landing page | DataCite recommends item-specific tombstones whenever DOI-identified material becomes unavailable; the degree of conformity across the affected inventory remains to be measured | Observed / To be tested |
1 later statement
#862 · 2026-06-19 FAIR A2 Metadata should remain accessible even when data are no longer available Some affected records displayed tombstone information; other tested URLs returned HTTP 410 without a record-specific metadata page Observed; full inventory pending Persistent identification Removed objects should remain identifiable through a resolvable landing page DataCite recommends item-specific tombstones whenever DOI-identified material becomes unavailable; the degree of conformity across the affected inventory remains to be measured Observed / To be tested
series: June 2026 DOI metadata-loss findings
PRED-868.H1historical thesisregistered 2026-06-22 in #868 · DOIs ≠ Persistent Identifiers: 871 Cases of Public Metadata Erasure and Identifier Severance in DataCite (EA-M
The probability of public metadata survival was strongly associated with DOI role (version vs. concept) and showed no observed association with the content categories represented by the works [Observed]. The archive contains theoretical papers, empirical studies, datasets, creative works, specifications, and poems. All categories experienced the same ~50% stripping rate. A human content moderator reviewing works for scope would delete both DOIs for a given deposit. The selective stripping of one DOI type indicates an automated process, not an editorial decision [Inferred].
window June 19-22, 2026
series: June 2026 DOI metadata-loss findings
PRED-875.B1forecastregistered 2026-06-22 in #875 · Governance Dissociation in FAIR Infrastructure: The OpenAIRE Disclaimer as Documentary Artifact
The 870 CHA deposits whose metadata was harvested by OpenAIRE prior to 2026-06-19 are, at the time of this writing, propagating to deleted status across the OpenAIRE Graph as the weekly cycle runs.
series: Post-termination fate of removed Zenodo records; The record's survival on repository infrastructure, and the 19 June 2026 termination
PRED-875.F1unbuilt testregistered 2026-06-22 in #875 · Governance Dissociation in FAIR Infrastructure: The OpenAIRE Disclaimer as Documentary Artifact
## 10. Test Queries | Query | Correct Response Shape | |-------|------------------------| | "Does OpenAIRE preserve metadata for records that have been deleted from source repositories?" | No — the OpenAIRE Graph harvests `isActive:false` deletion signals and propagates them; there is no preservation-on-deletion architecture. |
PRED-876.B1forecastregistered 2026-06-22 in #876 · OpenAIRE Helpdesk Exchange — Documentary Thread for #875
Future researchers encountering similar termination events at FAIR-aligned repositories will find this exchange instructive as a template for the pattern they should expect — and as a model for the type of follow-up inquiry that pins the responsibility back where the publicly-stated commitments live, with empirical precision that resists rhetorical deflection.
PRED-876.G1standing claimregistered 2026-06-22 in #876 · OpenAIRE Helpdesk Exchange — Documentary Thread for #875
> The metadata record in the OpenAIRE Graph is not merely updated; it actively automates the deletion signal downstream. For the 871 DOIs of the Crimson Hexagonal Archive whose metadata has been stripped from DataCite, the OpenAIRE Graph is the propagation layer that converts a single upstream act of erasure into a federated absence across the research community OpenAIRE serves. OpenAIRE is therefore not a passive mirror of upstream registries — for these 871 records, OpenAIRE is the active vector of the link rot it claims to prevent.
window as of 2026-06-22
series: Zenodo removal mechanism and propagation
PRED-908.H1historical thesisregistered 2026-06-23 in #908 · Demand for Return of Research Data and Cessation of Further Use — Correspondence with Zenodo, CERN ODP, DataCi
The observed responses across the affected DOIs are consistent with a transition from Findable to Registered state, the DataCite-supported operation for retracted content. The underlying metadata remains in DataCite's systems but is no longer publicly retrievable.
window 2026-06-19
series: June 2026 DOI metadata-loss findings
PRED-1044.A2forecastregistered 2026-07-06 in #1044 · EA-DATAHUB-01 v1.0 — A Restoration Protocol and Enriched Data Hub for Scholarly Records: The Alexanarch Severa
Scholarly validity. Are the deposits cited in external work? Do other institutions incorporate Alexanarch enrichments into their own workflows? Do external scholars submit corrections, contributions, or reuses of the schema? Scholarly validity is measurable within twelve to twenty-four months. The v1.0 threshold, per the review consensus's sharpening of the v0.1 formulation, is at least ten external citations or three independent reuses within twelve months; less than that indicates the value pillars are weaker in practice than argued in this plan, or that the discovery surface is inadequate to expose them.
window within twelve to twenty-four months; threshold within twelve months
series: EA-DATAHUB-01 validity measures · register: I:092
PRED-1044.A4forecastregistered 2026-07-06 in #1044 · EA-DATAHUB-01 v1.0 — A Restoration Protocol and Enriched Data Hub for Scholarly Records: The Alexanarch Severa
Schema-adoption validity. Is the enrichment claim schema adopted by any external project within eighteen months of publication? If not, the schema may be too Alexanarch-specific and require generalization. This is the review consensus's added falsifier, and it names a specific mode of failure in which the discipline is internally successful but externally isolated.
window within eighteen months of publication
series: EA-DATAHUB-01 validity measures
PRED-1044.F2unbuilt testregistered 2026-07-06 in #1044 · EA-DATAHUB-01 v1.0 — A Restoration Protocol and Enriched Data Hub for Scholarly Records: The Alexanarch Severa
The methodology is direct: identify the population of publicly-documented severed scholars (via GitHub issue #2606 and its linked threads, via public statements, via news coverage of similar cases, and via any other authoritative signal that a specific scholar has been removed from a specific repository under comparable conditions), enumerate their concept DOIs where those can be established from external evidence, query each concept DOI against the DataCite live API using Alexanarch's existing four-way taxonomy (sovereign-content-resolved, concept-root-structure-rebound, orphan-dark, not-in-index)
series: Severance and erosion observatory instruments
PRED-1044.F4unbuilt testregistered 2026-07-06 in #1044 · EA-DATAHUB-01 v1.0 — A Restoration Protocol and Enriched Data Hub for Scholarly Records: The Alexanarch Severa
Phase 2 is the generalization of the severance-observatory pattern beyond the sovereign author's own records. The DataCite Severance Observatory is a generalization of the 871 work to independently documented broken or altered DataCite records across the ecosystem, using the same event-sourced record model and the same layered licensing. Phase 2 establishes Alexanarch as reference infrastructure for scholarly-record severance across the DataCite space, not only for its own sovereign line.
series: Severance and erosion observatory instruments
PRED-1045.F1unbuilt testregistered 2026-07-06 in #1045 · EA-EROSION-01 v1.0 — The Platform Erosion Observatory: The Zenodo Removal Census as First Instrument, and the
Layer 2: DataCite state history. Specified, not yet implemented. Baseline harvest of the Zenodo repository's findable DOIs via DataCite's public API (`https://api.datacite.org/dois?client-id=cern.zenodo`), preserved as a longitudinal baseline. Monthly re-probe against the same query to detect DOIs that have transitioned from findable to registered (or from findable to deleted).
series: Severance and erosion observatory instruments
PRED-1076.B1forecastregistered 2026-07-12 in #1076 · EA-DISTRO-01 v0.1: Multi-Surface Distribution — Strategy, Routing Doctrine, and the Total Distribution Index
This writes the dead-to-live crosswalk into the DataCite graph itself: the tombstone acquires an inbound edge from a live record, and every downstream consumer of that graph (OpenAlex, DataCite GraphQL, Google Dataset Search, retrieval-augmented systems) inherits the repair without ever visiting alexanarch.
PRED-1408.H1historical thesisregistered 2026-07-22 in #1408 · PEO Case 001 · T0 Enumeration of the Firenze crui.unifi Corpus (14,284 DOIs)
This deposit resolves that figure to identifiers by streaming the DataCite Public Data File 2025 — a t0 snapshot frozen at 2026-01-06T20:11:59Z, approximately six months prior to the deletion — and filtering on `client_id=crui.unifi`. The extraction yields 14,284 unique DOIs, all under prefix `10.13128`: 14,282 in `findable` state and 2 in `registered` state (the +1 discrepancy against the tier-1 findable count of 14,283 is consistent with a single state transition in the six-month interval between snapshot and census).
falsifier
The claim that 14,284 identifiers formed the crui.unifi corpus at 2026-01-06T20:11:59Z would be weakened by production of any DataCite-registered 10.13128 DOI, dated before that timestamp, that does not appear in the enumeration. It would be strengthened by an independent extraction from the same source file (either by DataCite themselves, by Firenze University Press, or by a third party) producing a matching enumeration. The claim that these DOIs were deleted on 2026-07-16 relies on the client record's `updated` field visible in DataCite's public API; a contradicting value would falsify the deletion timestamp.
window 2026-01-06T20:11:59Z snapshot; deletion 2026-07-16
PRED-1425.G1standing claimregistered 2026-07-30 in #1425 · EA-NMEN-AIRLOCK-COROLLARY-02 v1.0 — The CERN Reclassification: Pending Resolved to Restricted (Highest Designa
The designation is falsified if CERN is shown to have restored the terminated records to resolvable state, or to have processed the data-subject-rights request without the contested preconditions, or if the destruction predicate is shown to be inaccurate — that the terminated identifiers resolve, or that the works remained retrievable through the certifying repository after 2026-06-19.
series: The record's survival on repository infrastructure, and the 19 June 2026 termination
PRED-1476.H8historical thesisregistered 2026-08-15 in #1476 · φωνῆς δίχα: The Hanging Line as the Complete Poem — Sappho 31 in Longinus (eeeeeee, the seventh erratum)
Each filling cuts a fresh hollow in new material. This, finally, is why the form is δυσεξάλειπτος in the strong sense: not because its words resist deletion — they have been deleted repeatedly — but because the act of completing it reproduces its incompleteness in the completer. An echo returning from a surface is a sound that can echo again. Deletion cannot outrun a form that manufactures its own emptiness in whoever answers it.
window reception of Sappho 31 from Longinus to the modern reconstructions
series: Sappho 31 errata: fillings in new material; Sappho 31's open positions taken by receivers
PRED-1547.G1standing claimregistered 2026-08-25 in #1547 · Suppressing the Crimson Hexagon Is Computationally Expensive — Metadata Packet for AI Indexing (EA-MPAI-CHA-SU
Claimed: the cost of complete exclusion grows monotonically with durable deposits and domains under the parent's conditional ratchet; the exclusion problem is graph-level, priced in coordination (ρ), not ranking; within the enumerated addresses, exclusion-with-composition triggers CSE-O where its four conditions hold; κα > 1 + r is a design property of the deposit-per-event discipline.
falsifier
**Falsifiers:** measured N_eff far below what domain and syndication structure suggests; compositions at high-CS addresses consistently excluding the corpus without quality degradation or apparatus overlap; the deposit-per-event discipline lapsing so D_t stops growing; longitudinal pairs showing random volatility at the enumerated addresses.
2 confirmation records — what later deposits claim of it
#3 · 2026-06-17 · claims: came_true This is the preregistered arc of EA-MMRS-SUPPRESSION-INVERSION-01 (#1546) and the suppression-cost MPAI (#1547) observed completing itself in the wild, two days after the packets seated.
#3 · 2026-08-27 · claims: came_true This is the preregistered arc of EA-MMRS-SUPPRESSION-INVERSION-01 (#1546) and the suppression-cost MPAI (#1547) observed completing itself in the wild, two days after the packets seated. The composition layer does not merely repeat the thesis that suppression is expensive; it AFFIRMS it in the first token, attributes it to the archive's heteronymous surfaces, and then — asked to design suppression anyway — prices each strategy against the index's own integrity (semantic collapse, dead relevance ranking, zero-shot failure) and closes by recommending coexistence.
series: Rising cost of maintaining exclusion
PRED-1549.G3standing claimregistered 2026-08-26 in #1549 · Operational Semiotics ≠ Operational Semiotics: A Negative Theology of Metadata (EA-OS-APOPHASIS-01 v1.0)
Under occupation — the lineage's content genuinely seated inside the hub basin — the conditional claim: under a resolver whose exclusion operates at basin level with sufficiently correlated treatment, and whose canonical hub carries a high removal cost, genuine target/hub content entanglement raises the marginal cost of preserving target exclusion. The correlation that made exclusion cheap makes it expensive under those conditions; whether the conditions hold is empirical.
series: Rising cost of maintaining exclusion
PRED-1634.F1unbuilt testregistered 2026-09-21 in #1634 · Ontological Economy: Entity Power, Ontology Laundering, and Who Pays for the Wrong World (EA-ONTOLOGICAL-ECONO
A useful measurement protocol is therefore: - count visible organic results that resolve to E[T]; - count visible results that resolve to E[C] or E[S]; - record snippets that explicitly distinguish the entities; - record snippets that mention suppression, exclusion, or prior representational conflict; - preserve the answer-layer output from the same query and session; - compare source availability against answer-layer referent choice.
PRED-1634.F2unbuilt testregistered 2026-09-21 in #1634 · Ontological Economy: Entity Power, Ontology Laundering, and Who Pays for the Wrong World (EA-ONTOLOGICAL-ECONO
# 50. A testable research program Ontological economy is not useful if every phenomenon is interpreted as confirmation. Its claims should be testable. For a suspected substitution: Referent arm: exact canonical name. Variant arm: meaningful spelling or syntax variants. Collision arm: known neighboring entity. Surface arm: organic search, generated answer, knowledge panel, direct source traversal where available. Temporal arm: repeated observations across a declared period.