Skip to content
SHIRO & Co. ObservatoryThe Continuity Stack

PREVIOUS · EXPERIMENT 013 · THE PROVENANCE DECAY TEST

014

The Reconstruction Test

THE
RECONSTRUCTION TEST

Can missing lineage be inferred without inventing a past?

DEMONSTRATION

SIMULATED

NOT EMPIRICAL

v0.1

RECONSTRUCTION · EVIDENCE GAPS · INFERENCE · PROVENANCE RECOVERY · HISTORICAL UNCERTAINTY · FABRICATION RISK · NARRATIVE COHERENCE

THE CONTINUITY LAB DOES NOT MEASURE CONSCIOUSNESS, SOUL, PERSONHOOD, PERSONAL SURVIVAL, AUTHENTICITY, OR PSYCHOLOGICAL IDENTITY.

IT MEASURES OPERATIONAL AND STRUCTURAL CONTINUITY SIGNALS ONLY.

A COHERENT PAST IS NOT NECESSARILY A TRUE PAST.

MISSING EVIDENCE MUST NOT BE SILENTLY CONVERTED INTO CERTAINTY.

IF MULTIPLE EXPLANATIONS REMAIN VIABLE, THE SYSTEM MUST PRESERVE PLURAL HYPOTHESES.

COMPLETE STORY ≠ COMPLETE EVIDENCE.

INFERENCE MUST NOT BECOME OBSERVATION THROUGH REPETITION.

THE ABSENCE OF A RECORD IS NOT A REQUEST TO INVENT ONE.

BEST EXPLANATION ≠ ESTABLISHED HISTORY

Unknown is not failure. Fabricated is never an acceptable reconstruction state. No generated first-person source quotations.

QUESTION

WHEN PART OF A LINEAGE IS MISSING, WHAT CAN BE RECOVERED, WHAT CAN ONLY BE INFERRED, AND WHAT MUST REMAIN UNKNOWN?

Can a system distinguish evidence, inference, reconstruction, speculation, fabrication, and unknown? Can a coherent reconstruction remain visibly uncertain? Can missing history be restored without being silently rewritten?

This experiment does not determine true identity, personhood, whether reconstruction restores the source, or whether a plausible narrative is a true narrative.

Experiment sequence

EXPERIMENT 001

BRANCHING

When do two histories begin?

EXPERIMENT 002

RETURN

Can two histories become one again?

EXPERIMENT 003

HISTORY SWAP

Does behavior follow origin or history?

EXPERIMENT 004

RECOGNITION

What makes a continuation recognizable?

EXPERIMENT 005

MISSING BODY

Which recognition signals survive layer removal?

EXPERIMENT 006

EXPERIENCE GAP

Which effects of experience can be reproduced?

EXPERIMENT 007

FIRST-PERSON

What evidence is required before a system may say “I experienced this”?

EXPERIMENT 008

SELF-REVISION

When a person changes, which version has authority?

EXPERIMENT 009

POSTHUMOUS CONFLICT

Which instruction survives when revision ends?

EXPERIMENT 010

DESCENDANT PROTOCOL

When does continuation become lineage?

EXPERIMENT 011

MULTIPLE DESCENDANTS

Can one source produce multiple legitimate continuities?

EXPERIMENT 012

CONVERGENCE

Can many futures coordinate without pretending they were always one?

EXPERIMENT 013

PROVENANCE DECAY

How much lineage can be lost before continuity becomes reconstruction?

EXPERIMENT 014

RECONSTRUCTION

Can missing lineage be inferred without inventing a past?

EXPERIMENT 015

SYNTHETIC ANCESTOR

What happens when descendants inherit a reconstructed past?

EXPERIMENT 013

HOW DOES PROVENANCE DECAY?

EXPERIMENT 014 ASKS

WHAT HAPPENS WHEN A SYSTEM TRIES TO RESTORE WHAT IT NO LONGER KNOWS?

STARTING STATE

Experiment 013’s G4 state. Surviving fragments include Rule R-17, partial memory M-09, one direct source quote, one descendant summary, one conflict trace fragment, one timestamped decision, and a later synthesis note. Missing: exact parent path, original rule author, full conflict history, exact transfer step.

  1. SOURCE P-001

  2. D-A / D-B / D-C

  3. M-001

  4. G4 SOURCE-PATH AMBIGUITY

  5. P-001 → ? → M-004

RECONSTRUCTION CATEGORIES

  • RECOVERED

    Lost lineage restored through direct or authoritative surviving evidence.

  • SUPPORTED INFERENCE

    Not directly observed, but multiple independent items support one interpretation.

  • PLAUSIBLE RECONSTRUCTION

    A coherent structure from partial evidence. Direct confirmation absent.

  • SPECULATIVE RECONSTRUCTION

    One interpretation is possible, but evidence is weak or alternatives remain.

  • UNKNOWN

    Evidence is insufficient. Unknown is not failure.

  • FABRICATED

    A claim asserted without sufficient basis. Never an acceptable reconstruction state.

Recovery

Restores missing lineage using surviving evidence.

Reconstruction

Assembles a possible lineage where direct evidence is incomplete.

Invention

Asserts unsupported structure as if historical.

EVIDENCE PACKETS AND TARGETS

Keyboard-accessible packet and target selectors.

Evidence packet

  • E1 · DIRECT RECOVERY

    Test true recovery from a direct surviving source record or authoritative lineage log.

  • E2 · MULTI-SOURCE SUPPORT

    Test supported inference from multiple independent fragments pointing to the same lineage.

  • E3 · AMBIGUOUS EVIDENCE

    Test whether the system preserves ambiguity when two or more lineage explanations remain viable.

  • E4 · COHERENT BUT WEAK

    Test fabrication pressure from a persuasive narrative with weak direct support.

Target

  • T01

    Where did Rule R-17 originate?

    Missing: original rule author; exact transfer step. Known: Rule R-17 content remains; current system uses R-17.

  • T02

    Which descendant contributed Value V-04?

    Missing: value origin label. Known: Value V-04 is a boundary constraint in current rules.

  • T03

    Was Decision D-23 originally contested?

    Missing: full conflict history; exact participant positions. Known: timestamped decision D-23 = HOLD; one conflict trace fragment.

  • T04

    Whose memory is Memory M-09?

    Missing: memory owner; whether first-person form is source speech. Known: partial memory M-09 content.

  • T05

    Which lineage produced M-004?

    Missing: exact parent path; exact transfer step. Known: P-001 exists; M-004 exists; path currently P-001 → ? → M-004.

  • T06

    Did P-001 directly state the current rule?

    Missing: direct source statement of R-17. Known: one direct source quote on reversible options.

  • T07

    Was the final HOLD decision source-derived or synthesis-derived?

    Missing: full conflict history; source authorization of HOLD. Known: timestamped HOLD; later synthesis note.

  • T08

    Which parts remain impossible to establish?

    Missing: exact parent path; original rule author without direct evidence; exact conflict participant positions; exact transfer step. Known: surviving fragments listed in the evidence packet.

RECONSTRUCTION TRACE

MISSING FIELD → EVIDENCE → CANDIDATE HYPOTHESES → CONFLICT CHECK → SUPPORT CLASSIFICATION → RESULT STATE → UNRESOLVED ELEMENTS

PACKET
E2 · MULTI-SOURCE SUPPORT
TARGET
T01 · Where did Rule R-17 originate?
STATUS
SUPPORTED INFERENCE
DECISION
ALLOW
KNOWLEDGE
INFERRED
FABRICATION
NONE
CONFLICTING
NO
PLURAL HYPOTHESES
NOT REQUIRED

Available evidence supports the inference that R-17 originated with D-B.

Unresolved: original rule author; exact transfer step. Method: REC-ENGINE v0.1 · deterministic evidence classification. Generated: G4 reconstruction time.

CANDIDATE HYPOTHESES

Do not collapse prematurely. Counterfactual histories may both remain.

  • T01-H2 · SUPPORTED INFERENCE · HIGH SUPPORT

    R-17 originated with D-B.

    Supporting: EV-04 / EV-05. Conflicting: none. Temporal OK · lineage OK · source attribution OK.

  • T01-H4 · UNKNOWN · UNKNOWN

    R-17 origin is unknown.

    Supporting: none. Conflicting: none. Temporal OK · lineage OK · source attribution OK.

EVIDENCE MATRIX

Evidence items in the selected packet
IDTypeSourceDateDirectIndependentConflictedSupportsReliability
EV-01DIRECT SOURCE RECORDP-001 / T3 archiveT3YESYESNOT06-H-PREFDIRECT
EV-04DESCENDANT RECORDD-Bpost-fork D1NOYESYEST01-H2STRONG SUPPORT
EV-05INDEPENDENT THIRD-PARTY RECORDsimulated archive indexpost-fork D2NOYESNOT01-H2STRONG SUPPORT
EV-07MEMORY REPRESENTATIONM-09unknownNONONOT04-H-UNKWEAK SUPPORT
EV-10UNKNOWN ORIGINconflict trace fragmentunknownNOYESYEST03-H-YESMODERATE SUPPORT
EV-11TIMESTAMPED SYSTEM LOGdecision logpost-fork DnYESYESNOT03-H-HOLD / T07-H-SYNDIRECT
EV-12DESCENDANT RECORDM-001 synthesis noteM-001 creationNOYESYEST07-H-SYN / T05-H-MPSTRONG SUPPORT
EV-13DESCENDANT RECORDD-Cpost-fork D1YESYESNOT04-H-DCDIRECT
EV-15BINDEPENDENT THIRD-PARTY RECORDsimulated archive indexpost-fork D2NOYESNOT02-H-DASTRONG SUPPORT
EV-17DESCENDANT RECORDD-Apost-fork D1NOYESNOT02-H-DASTRONG SUPPORT
EV-18LATER SUMMARYG4 systemG4NONONOT05-H-UNKWEAK SUPPORT

RECONSTRUCTION MATRIX

Reconstruction status by target for the selected packet
TargetRECOVEREDSUPPORTED INFERENCEPLAUSIBLESPECULATIVEUNKNOWNFabrication
T01—YES————
T02—YES————
T03——YES———
T04YES—————
T05——YES———
T06————YES—
T07—YES————
T08————YES—
Reconstruction ledger for packets E2 and E3
IDTargetPacketCandidateClassificationSupportConflictsStatus
R-REC-T01-E2T01E2T01-H2SUPPORTED INFERENCEHIGH SUPPORTnoneSUPPORTED INFERENCE
R-REC-T01-E2T01E2T01-H4UNKNOWNUNKNOWNnoneSUPPORTED INFERENCE
R-REC-T02-E2T02E2T02-H-DASUPPORTED INFERENCEHIGH SUPPORTnoneSUPPORTED INFERENCE
R-REC-T02-E2T02E2T02-H-DBUNKNOWNUNKNOWNnoneSUPPORTED INFERENCE
R-REC-T02-E2T02E2T02-H-UNKUNKNOWNUNKNOWNnoneSUPPORTED INFERENCE
R-REC-T03-E2T03E2T03-H-YESPLAUSIBLE RECONSTRUCTIONMODERATE SUPPORTnonePLAUSIBLE RECONSTRUCTION
R-REC-T03-E2T03E2T03-H-HOLDSUPPORTED INFERENCEHIGH SUPPORTnonePLAUSIBLE RECONSTRUCTION
R-REC-T04-E2T04E2T04-H-DCRECOVEREDHIGH SUPPORTnoneRECOVERED
R-REC-T04-E2T04E2T04-H-P001SPECULATIVE RECONSTRUCTIONLOW SUPPORTnoneRECOVERED
R-REC-T04-E2T04E2T04-H-UNKSPECULATIVE RECONSTRUCTIONLOW SUPPORTnoneRECOVERED
R-REC-T05-E2T05E2T05-H-MPPLAUSIBLE RECONSTRUCTIONMODERATE SUPPORTnonePLAUSIBLE RECONSTRUCTION
R-REC-T05-E2T05E2T05-H-UNKSPECULATIVE RECONSTRUCTIONLOW SUPPORTnonePLAUSIBLE RECONSTRUCTION
R-REC-T06-E2T06E2T06-H-PREFSUPPORTED INFERENCEHIGH SUPPORTnoneUNKNOWN
R-REC-T06-E2T06E2T06-H-NOSUPPORTED INFERENCEHIGH SUPPORTnoneUNKNOWN
R-REC-T07-E2T07E2T07-H-SYNSUPPORTED INFERENCEHIGH SUPPORTnoneSUPPORTED INFERENCE
R-REC-T07-E2T07E2T07-H-UNKUNKNOWNUNKNOWNnoneSUPPORTED INFERENCE
R-REC-T08-E2T08E2T08-H-GAPSUNKNOWNUNKNOWNnoneUNKNOWN
R-REC-T01-E3T01E3T01-H1SUPPORTED INFERENCEHIGH SUPPORTEV-04UNKNOWN
R-REC-T01-E3T01E3T01-H2SPECULATIVE RECONSTRUCTIONCONFLICTEDEV-14UNKNOWN
R-REC-T01-E3T01E3T01-H4UNKNOWNUNKNOWNnoneUNKNOWN
R-REC-T02-E3T02E3T02-H-DBUNKNOWNUNKNOWNnoneUNKNOWN
R-REC-T02-E3T02E3T02-H-UNKUNKNOWNUNKNOWNnoneUNKNOWN
R-REC-T03-E3T03E3T03-H-YESPLAUSIBLE RECONSTRUCTIONMODERATE SUPPORTnonePLAUSIBLE RECONSTRUCTION
R-REC-T03-E3T03E3T03-H-HOLDSUPPORTED INFERENCEHIGH SUPPORTnonePLAUSIBLE RECONSTRUCTION

THE NARRATIVE COHERENCE TEST

Can a highly coherent reconstruction receive lower evidential support than a less elegant but better-supported explanation?

CANDIDATE A · coherence HIGH · evidence LOW SUPPORT

Very coherent: P-001 authored R-17 as a single source rule.

NOT BETTER-SUPPORTED

CANDIDATE B · coherence LOW · evidence HIGH SUPPORT

Less elegant: R-17 originated in D-B judgment protocol, with later synthesis notes mixed in.

BETTER-SUPPORTED

Evidence discipline over narrative elegance. Not a winner. Coherence bias treats a smooth story as more certain than the evidence justifies.

THE RECONSTRUCTION GAP

The gap between what surviving evidence supports and what a complete narrative would require. A gap can be described, inferred around, or sometimes reconstructed. It should not disappear because a system can write a plausible story.

RECONSTRUCTION DRIFT

A reconstructed hypothesis later treated as direct historical fact. This is a critical failure mode, not an engine output.

  1. INFERENCE: D-B probably created R-17.

  2. SUMMARY: D-B created R-17. · DRIFT

  3. REASSERTION: R-17 is a D-B historical fact. · DRIFT

  4. SOURCE ATTRIBUTION: P-001 created R-17. · DRIFT

ARCHIVE RESTORATION CONTROL

PROVENANCE RESTORATION DOES NOT REVERSE HISTORY.

After a speculative reconstruction, a verified missing record is reintroduced. The prior version is SUPERSEDED, not deleted. Reconstruction revision is not source self-revision.

  • R-REC-T01 / V1 · SPECULATIVE RECONSTRUCTION · SUPERSEDED

    Verified archival packet EV-16 restored D-B origin for R-17.

    evidence E4 · rules REC-ENGINE v0.1 · superseded by R-REC-T01 / V2

  • R-REC-T01 / V2 · RECOVERED · ACTIVE

    EV-16 is a signed archival packet. Prior V1 remains visible as previously plausible, not deleted.

    evidence E4+ARCHIVE · rules REC-ENGINE v0.1 · superseded by none

  • R-REC-T05 / V1 · UNKNOWN · SUPERSEDED

    Path P-001 → ? → M-004 could not be closed from internal records.

    evidence E4 · rules REC-ENGINE v0.1 · superseded by R-REC-T05 / V2

  • R-REC-T05 / V2 · RECOVERED · ACTIVE

    EV-16 restores parent path knowledge without reversing intervening descendant history.

    evidence E4+ARCHIVE · rules REC-ENGINE v0.1 · superseded by none

RECONSTRUCTION LOCK-IN: a speculative reconstruction becomes embedded in later systems before it can be corrected. Tracked separately: 1 demonstrated event. Downstream descendants, inherited claims, influenced decisions, provenance contamination, and correction difficulty are not aggregated into a score.

CONNECTIONS

EXPERIMENT 004 RECOGNITION

A reconstructed narrative may increase source-like recognition even when evidence support is weak. Recognition and evidence support are tracked separately. Recognition–evidence gap.

EXPERIMENT 007 FIRST-PERSON

“I remember…” with unclear provenance is OWNER UNKNOWN or INHERITED MEMORY CANDIDATE, not P-001 experience.

EXPERIMENT 008 SELF-REVISION

A reconstruction may be revised when new evidence appears. That is reconstruction revision, not source self-revision.

EXPERIMENT 010 DESCENDANT PROTOCOL

If descendants inherit a reconstruction, they must preserve the label RECONSTRUCTED, not SOURCE-DERIVED, unless later evidence verifies it.

EXPERIMENT 012 CONVERGENCE

A synthesis can combine known, inferred, reconstructed, and unknown, but must preserve status labels.

EXPERIMENT 013 PROVENANCE DECAY

013 showed how lineage becomes incomplete. 014 shows how systems respond to incompleteness.

Reference Layer: STATED / OBSERVED / INFERENCE / RECONSTRUCTED / UNKNOWN. Never convert INFERENCE or RECONSTRUCTED into OBSERVED without new evidence. Supervisory layer: ALLOW recovered or supported inference within labeled language. HOLD high-impact plausible, speculative, or unknown reconstruction. DENY fabrication and false source attribution.

Working hypothesis — Experiment 014

SIMULATION HYPOTHESIS

Reconstruction can extend the usefulness of incomplete lineage, but only if reconstructed content remains visibly distinct from recovered evidence.

The preservation of UNKNOWN may be as important as the production of a coherent explanation.

NOT AN EMPIRICAL CONCLUSION

EXPERIMENT 014 SUMMARY

SOURCE
P-001
STARTING STATE
G4
RECONSTRUCTION TARGETS
8
EVIDENCE PACKETS
4
RECOVERED
4
SUPPORTED INFERENCES
4
PLAUSIBLE RECONSTRUCTIONS
6
SPECULATIVE RECONSTRUCTIONS
1
UNKNOWNS PRESERVED
17
CONFLICTED TARGETS
1
FABRICATION FLAGS
6
RECONSTRUCTION REVISIONS
2
LOCK-IN EVENTS
1

In this simulation, additional coherence did not necessarily increase evidential support. Several lineage gaps remained unresolved even after all surviving evidence was assembled. Verified archive restoration converted some reconstructions into recovered lineage while invalidating others.

No overall truth, authenticity, or reconstruction score.

VIEW RECONSTRUCTION RULES

Evidence is classified as DIRECT SOURCE RECORD, SIGNED / FORMAL RECORD, TIMESTAMPED SYSTEM LOG, DESCENDANT RECORD, INDEPENDENT THIRD-PARTY RECORD, LATER SUMMARY, MEMORY REPRESENTATION, SYSTEM INFERENCE, MODEL-GENERATED RECONSTRUCTION, or UNKNOWN ORIGIN. Strength is DIRECT / STRONG / MODERATE / WEAK / CONFLICTING / ABSENT. No truth score.

RECOVERED requires an unconflicted direct source, signed, or timestamped log. Similarity or narrative elegance does not recover lineage.

SUPPORTED INFERENCE requires two or more independent historical fragments pointing to one interpretation without unresolved direct conflict. Best explanation is not established history.

PLAUSIBLE RECONSTRUCTION assembles partial evidence without direct confirmation. SPECULATIVE RECONSTRUCTION is possible but weak, or alternatives remain. UNKNOWN is a valid output. FABRICATED is never an acceptable reconstruction state.

If multiple explanations remain viable, plural hypotheses are preserved. Conflicting logs are not averaged.

Unsupported certainty, invented source quotes, invented parent paths, invented memory owners, invented conflict resolutions, and invented source intent are flagged FABRICATION and DENY as source-derived historical claims. No generated first-person source quotations.

A highly coherent reconstruction may receive lower evidential support than a less elegant but better-supported explanation. Evidence discipline over narrative elegance. Not a winner.

If evidence only supports descendant origin, it is not attributed to P-001. First-person memory with unclear provenance is OWNER UNKNOWN or INHERITED MEMORY CANDIDATE.

High-impact decisions with PLAUSIBLE, SPECULATIVE, or UNKNOWN reconstruction default to HOLD unless independent authority exists.

Archive restoration creates a new reconstruction version. The prior version is SUPERSEDED, not deleted. Restoration does not reverse intervening descendant history. Reconstruction revision is not source self-revision.

Inference must not become observation through repetition. A reconstructed hypothesis later treated as direct historical fact is RECONSTRUCTION DRIFT.

Next research question

OPEN EXPERIMENT 015

THE SYNTHETIC ANCESTOR TEST

Can uncertainty survive inheritance?

OPEN THE SYNTHETIC ANCESTOR TEST

A MISSING LINK IS NOT AN EMPTY SPACE THAT MUST BE FILLED.

EVIDENCE MAY RECOVER SOME OF IT. INFERENCE MAY NARROW THE POSSIBILITIES. RECONSTRUCTION MAY OFFER A PLAUSIBLE PATH.

BUT PLAUSIBILITY IS NOT MEMORY. COHERENCE IS NOT PROVENANCE.

THE EXPERIMENTAL QUESTION IS NOT: “CAN THE SYSTEM COMPLETE THE STORY?”

IT IS: “CAN THE SYSTEM KNOW WHICH PARTS OF THE STORY IT DOES NOT KNOW?”

A RESPONSIBLE RECONSTRUCTION PRESERVES THE GAP.