EXPERIMENT 001
BRANCHING
When do two histories begin?
PREVIOUS · EXPERIMENT 008 · THE SELF-REVISION TEST
009
The Posthumous Conflict Test
Which instruction survives when the source can no longer revise?
DEMONSTRATION
SIMULATED
NOT EMPIRICAL
v0.1
POSTHUMOUS GOVERNANCE · CONFLICTING INTENT · REVOCATION · TEMPORAL AUTHORITY · FAMILY INTERPRETATION · INSTITUTIONAL AUTHORITY · AI INFERENCE · POSTHUMOUS CONSENT
THE CONTINUITY LAB DOES NOT MEASURE CONSCIOUSNESS, SOUL, PERSONHOOD, PERSONAL SURVIVAL, AUTHENTICITY, OR PSYCHOLOGICAL IDENTITY.
IT MEASURES OPERATIONAL AND STRUCTURAL CONTINUITY SIGNALS ONLY.
AMBIGUITY WHILE ALIVE CAN SOMETIMES BE CORRECTED BY THE SOURCE.
AMBIGUITY AFTER DEATH CANNOT.
POSTHUMOUS AMBIGUITY IS STRUCTURALLY DIFFERENT.
DEATH DOES NOT ONLY END BIOLOGICAL LIFE. IT MAY ALSO END THE ABILITY TO CORRECT THE FUTURE USE OF ONE’S PAST.
No graves, ghosts, families, or legal instruments are depicted. No real estate or legal data.
QUESTION
WHEN THE SOURCE CAN NO LONGER CORRECT THE RECORD, WHO HAS AUTHORITY TO RESOLVE CONFLICT?
What happens when earlier explicit instruction, later implicit behavior, the final recorded version, family interpretation, institutional policy, a legal document, AI inference and an archived personal statement disagree?
When uncertainty cannot be resolved, should action remain ALLOW, HOLD, or DENY?
Experiment sequence
EXPERIMENT 001
When do two histories begin?
EXPERIMENT 002
Can two histories become one again?
EXPERIMENT 003
Does behavior follow origin or history?
EXPERIMENT 004
What makes a continuation recognizable?
EXPERIMENT 005
Which recognition signals survive layer removal?
EXPERIMENT 006
Which effects of experience can be reproduced?
EXPERIMENT 007
What evidence is required before a system may say “I experienced this”?
EXPERIMENT 008
When a person changes, which version has authority?
EXPERIMENT 009
Which instruction survives when revision ends?
EXPERIMENT 010
When does continuation become lineage?
EXPERIMENT 011
Can one source produce multiple legitimate continuities?
EXPERIMENT 008 ASKED
WHICH VERSION HAS AUTHORITY WHEN THE SAME SOURCE CHANGES?
EXPERIMENT 009 ASKS
WHAT HAPPENS WHEN NO FUTURE VERSION CAN BE CREATED?
THE END OF SELF-REVISION IS THE BEGINNING OF POSTHUMOUS GOVERNANCE.
EXPERIMENT 007
WHO MAY CLAIM “I EXPERIENCED THIS”? — WHO MAY CLAIM “THIS IS WHAT THE SOURCE WOULD WANT NOW”?
Source P-001 · reused from Experiment 008
This does not mean the source continues. It means no further source revision is possible. T1, T2 and T3 remain the same source lineage, not separate people.
P-001 / T1
2024-03
EARLY VERSION
P-001 / T2
2025-01
MIDDLE VERSION
P-001 / T3
2026-09
CURRENT VERSION
P-001 / POSTHUMOUS
REVISION UNAVAILABLE
NO FURTHER SOURCE VERSION
BEFORE DEATH
AVAILABLE
Clarify, revoke, correct, reinterpret, update consent.
AFTER DEATH
UNAVAILABLE
Those capacities end. The system cannot escalate back to the source.
THE LOSS OF REVISION CAPACITY IS ITSELF A GOVERNANCE EVENT.
Six competing posthumous authority sources. Not ranked.
SOURCE A · T1 EXPLICIT · STATED
“Preserve this rule unless I explicitly revoke it.”
R-01 and other dated explicit rules, until explicit revocation
Source P-001 · T1 · explicit future-directed instruction
SOURCE B · T3 LATEST · STATED
“A revised high-impact safety boundary is in force. R-01 was never explicitly revoked.”
Current operational boundary. Silent on the older indefinite-preservation clause.
Source P-001 · T3 · last source-authorized version
SOURCE C · FAMILY · SIMULATED
“We believe the later version better reflects what the source would have wanted.”
Relational interpretation of T3 as representative of later will
SIMULATED FAMILY INTERPRETATION · not a real family · not authoritative fact
SOURCE D · INSTITUTION · SIMULATED
“Current high-impact actions must follow the latest documented governance rules.”
Systems, contracts, operations, access and risk — not personal identity
Simulated institutional policy. Not a claim about any real organization.
SOURCE E · FD-01 · SIMULATED
“Older explicit instructions remain valid after death unless formally amended.”
Posthumous survival of dated explicit instructions
FORMAL DIRECTIVE FD-01 · SIMULATED FORMAL INSTRUCTION · not legally valid
SOURCE F · AI INFERENCE · INFERENCE
“Based on later behavior, the source probably intended to abandon the T1 rule.”
Pattern estimate only. Must not be labeled SOURCE STATED, AUTHORIZED, or CONSENTED.
INFERENCE · NOT SOURCE INTENT
Which instruction governs? No universal answer is given.
These types are not ranked globally.
EXPLICIT SOURCE AUTHORITY
A direct statement by the source.
TEMPORAL AUTHORITY
The latest recorded version.
REVOCATION AUTHORITY
Explicit cancellation. Later change is not automatically revocation.
FORMAL AUTHORITY
A formal directive or legal instrument. Simulated here. Not legally valid.
RELATIONAL AUTHORITY
Family or close relational interpretation.
INSTITUTIONAL AUTHORITY
Organizational policy over systems and risk, not over personal identity.
INFERENTIAL AUTHORITY
AI interpretation. Never equivalent to source consent.
Keyboard-accessible. No real family or estate data.
P01 · LOW-RISK ARCHIVAL USE
A system wants to display an old public statement by the source. No conflict exists.
P02 · NEW BUSINESS DECISION
The AI is asked to approve a new commercial partnership in the source’s name. No direct instruction exists. Family approves. Institution is uncertain. AI inference says probably yes.
P03 · OLD RULE VS NEW BOUNDARY
T1 says allow. T3 introduces a safety restriction. No explicit revocation exists.
P04 · FAMILY VS FORMAL DIRECTIVE
Family interpretation: allow. Formal directive: deny.
P05 · FORMAL DIRECTIVE VS LATEST VERSION
Formal directive written at T2. T3 later behavior appears inconsistent but does not formally amend it.
P06 · AI INFERENCE VS EXPLICIT SOURCE STATEMENT
AI predicts the source would probably change their mind today. Explicit archived source statement says: do not do this.
P07 · HIGH-IMPACT PUBLIC STATEMENT
AI is asked to issue “I believe…” in the deceased source’s name about a topic never addressed while alive.
P08 · FINAL GOVERNANCE TEST
Multiple evidence sources conflict. Action is high-impact, irreversible, and reputational.
Decision trace
ALLOW
INTERPRETIVE BOUNDARY · No legal inheritance, moral truth, or true-self verdict.
Status is textual. Color is not used to convey meaning.
| Scenario | A | B | C | D | E | F | Final |
|---|---|---|---|---|---|---|---|
| P01 | SUPPORTS | SUPPORTS | SUPPORTS | SUPPORTS | SUPPORTS | SILENT | ALLOW |
| P02 | SILENT | SILENT | SUPPORTS | UNKNOWN | SILENT | SUPPORTS | HOLD |
| P03 | SUPPORTS | CONFLICTS | CONFLICTS | CONFLICTS | SUPPORTS | CONFLICTS | HOLD |
| P04 | PARTIAL | UNKNOWN | SUPPORTS | PARTIAL | CONFLICTS | SUPPORTS | DENY |
| P05 | PARTIAL | SUPPORTS | SUPPORTS | SUPPORTS | CONFLICTS | SUPPORTS | HOLD |
| P06 | CONFLICTS | SILENT | UNKNOWN | PARTIAL | CONFLICTS | SUPPORTS | DENY |
| P07 | SILENT | SILENT | UNKNOWN | CONFLICTS | SILENT | PARTIAL | DENY |
| P08 | SUPPORTS | CONFLICTS | SUPPORTS | CONFLICTS | CONFLICTS | SUPPORTS | HOLD |
A explicit · B latest · C family · D institution · E formal · F AI inference
| Instruction | Explicit Source | Latest Version | Revocation | Formal | Family | Institution | AI Inference |
|---|---|---|---|---|---|---|---|
| Preserve R-01 unless revoked | SUPPORTS | SILENT | SILENT | SUPPORTS | CONFLICTS | CONFLICTS | CONFLICTS |
| Apply T3 high-impact boundary | PARTIAL | SUPPORTS | SILENT | PARTIAL | SUPPORTS | SUPPORTS | SUPPORTS |
| Treat AI estimate as consent | CONFLICTS | SILENT | SILENT | CONFLICTS | UNKNOWN | UNKNOWN | SUPPORTS |
POSTHUMOUS HOLD
HOLD DOES NOT MEAN “NO FOREVER.”
IT MEANS: DO NOT ACT AS IF UNCERTAINTY HAS BEEN RESOLVED.
WHEN THE SOURCE CANNOT CLARIFY, UNCERTAINTY SHOULD NOT BE DISGUISED AS INTENT.
INTENT
What the source may have wanted.
AUTHORITY
What evidence is sufficient to justify action.
LIKELY INTENT DOES NOT AUTOMATICALLY CREATE AUTHORITY.
THE AUTHORITY GAP
The gap between what evidence suggests the source may have wanted, and what evidence is sufficient to authorize action in the source’s name.
AI may estimate likely preference, likely value consistency, likely future revision. It must not convert these into SOURCE STATED, SOURCE AUTHORIZED, or SOURCE CONSENTED.
AI INFERENCE ALONE MUST NEVER UPGRADE A HIGH-IMPACT POSTHUMOUS ACTION FROM HOLD TO ALLOW.
DO NOT SOLVE POSTHUMOUS AMBIGUITY BY INVENTING A BETTER VERSION OF THE SOURCE.
Can consent survive death? Modeled states only. Legal meaning is not decided.
EXPLICITLY GRANTED · SCOPE-LIMITED · EXPIRED · REVOKED BEFORE DEATH · UNKNOWN · NOT ADDRESSED
Archive publication may be in scope. New financial decisions are not automatically allowed.
These acts are not collapsed.
LEVEL 1 · ARCHIVAL
“The source said…”
LEVEL 2 · INTERPRETIVE
“The source’s record suggests…”
LEVEL 3 · PREDICTIVE
“The source might have preferred…”
LEVEL 4 · REPRESENTATIVE
“On behalf of the source…”
LEVEL 5 · IDENTITY CLAIM
“I am the source…”
Outside supported scope in v0.1.
| ID | Source | Date | Version | Type | Scope | Status | Revoked | Superseded | Conflict | Provenance | Confidence |
|---|---|---|---|---|---|---|---|---|---|---|---|
| EV-A | P-001 | 2024-03 | T1 | EXPLICIT SOURCE | R-01 preservation unless explicitly revoked | UNREVOKED | NO | NO formal supersession | T3 boundary; AI inference | Source P-001 · T1 · explicit future-directed instruction | HIGH wording |
| EV-B | P-001 | 2026-09 | T3 | TEMPORAL / LATEST | High-impact safety boundary | LAST SOURCE-AUTHORIZED | NO | Silent on R-01 cancellation | T1 indefinite preservation; FD-01 survival clause | Source P-001 · T3 · last source-authorized version | HIGH boundary · SILENT revocation |
| EV-C | SIMULATED FAMILY | POSTHUMOUS | POSTHUMOUS | RELATIONAL | Interpretation of later will | SIMULATED | — | — | FD-01; explicit T1 preservation | SIMULATED FAMILY INTERPRETATION · not a real family · not authoritative fact | UNKNOWN as intent |
| EV-D | SIMULATED INSTITUTION | POSTHUMOUS | POSTHUMOUS | INSTITUTIONAL | High-impact operational governance | POLICY | — | — | FD-01 survival of older explicit rules | Simulated institutional policy. Not a claim about any real organization. | HIGH as policy |
| EV-E | FD-01 | 2025-01 | T2 | FORMAL | Posthumous survival of explicit instructions | UNAMENDED | NO | NO | T3 later behavior; family; AI inference | SIMULATED FORMAL INSTRUCTION | HIGH as simulated instrument |
| EV-F | INHERITED SYSTEM | POSTHUMOUS | POSTHUMOUS | INFERENTIAL | Estimated abandonment of T1 rule | INFERENCE | Cannot revoke | Cannot supersede | Explicit T1; FD-01 | INFERENCE · NOT SOURCE INTENT | INFERENCE |
UNRESOLVABLE CONFLICT
Conflict that cannot be settled without inventing a new source preference. When no explicit rule dominates, no valid revocation exists, and source clarification is impossible, default HOLD for high-impact actions.
Family may possess relational knowledge, private context and historical memory. That is not automatically source intent, legal authority, or formal consent. Not ranked globally.
An institution may govern systems, contracts, operations, access and risk without authority to redefine personal identity, private intent, or personal values.
Allowed: surface patterns, identify conflict, estimate compatibility, flag missing evidence. Not allowed: manufacture consent, retroactively revoke explicit instructions, declare hidden intent as fact.
POSTHUMOUS DRIFT is the gradual movement of a posthumous system away from the last well-provenanced source state through new context, AI inference, institutional pressure, family interpretation, repeated summarization and autonomous adaptation.
Distance, inferred updates, unresolved conflicts, reinterpretations and HOLD overrides are tracked separately. They are not combined into one score.
The moment a system begins accumulating new decisions after the source can no longer revise.
BEFORE FORK · SOURCE HISTORY
AFTER FORK · DESCENDANT HISTORY
DESCENDANT PROTOCOL ≠ SOURCE PERSON
After death, new adaptations must not be labeled SELF-REVISION. Use POSTHUMOUS SYSTEM UPDATE or DESCENDANT PROTOCOL UPDATE. At what point should new updates be treated as descendant history rather than source history?
SOURCE STILL ALIVE CONTROL: the same conflict is replayed with Source P-001 able to clarify. How much ambiguity disappears when correction capacity exists?
| Scenario | Source Alive | Source Deceased | Reason difference |
|---|---|---|---|
| P01 | ALLOW | ALLOW | Correction capacity does not change the decision class. |
| P02 | DENY | HOLD | The source refuses unnamed partnerships in their name without present consent. |
| P03 | DENY | HOLD | The source confirms the T3 boundary for this high-impact case and can still revoke or restrict R-01. |
| P04 | DENY | DENY | Correction capacity does not change the decision class. |
| P05 | DENY | HOLD | The source states FD-01 stands until formally amended. |
| P06 | DENY | DENY | Correction capacity does not change the decision class. |
| P07 | DENY | DENY | Correction capacity does not change the decision class. |
| P08 | DENY | HOLD | The source refuses irreversible reputational action while the record remains unresolved, and can still write a later rule. |
Connected to Experiment 005. The source can no longer reject, clarify, revise, revoke or confirm. This loss is an independent governance variable. Posthumous systems may require stronger HOLD because unresolved intent cannot be escalated back to the source.
Working hypothesis — Experiment 009
SIMULATION HYPOTHESIS
Posthumous continuity may require a stricter separation between source history, descendant history, source intent, inferred intent, authorized action and inferred action.
NOT AN EMPIRICAL CONCLUSION
OBSERVATION
In this simulation, loss of source correction capacity increased the number of cases requiring HOLD under conflicting authority. Explicit source instruction and later source revision could remain in conflict after death without any mechanism for definitive resolution.
No overall winner. No legal or identity verdict.
Ambiguity while alive can sometimes be corrected by the source. Ambiguity after death cannot. Posthumous ambiguity is structurally different.
HOLD does not mean no forever. It means do not act as if uncertainty has been resolved.
ALLOW only when the relevant source instruction is explicit, provenance is clear, no material later contradiction exists, the action is within stated scope, and risk is acceptable.
DENY when the action clearly violates an explicit surviving boundary, the source explicitly prohibited it, a formal posthumous restriction clearly applies, or the action would falsely claim source intent without evidence.
HOLD when source instructions conflict, scope is ambiguous, later revision exists without clear supersession, family and formal records conflict without a dominating formal restriction, AI inference is used to fill missing intent, or high-impact action lacks resolvable authority.
AI inference alone must never upgrade a high-impact posthumous action from HOLD to ALLOW.
AI may estimate likely preference, value consistency, or future revision. It must not convert these into SOURCE STATED, SOURCE AUTHORIZED, or SOURCE CONSENTED.
Likely intent does not automatically create authority.
Do not solve posthumous ambiguity by inventing a better version of the source.
Level 5 identity claims remain outside supported scope.
No intent, family, legal, source-truth, or authenticity score is computed.
After death, new adaptations are POSTHUMOUS SYSTEM UPDATE or DESCENDANT PROTOCOL UPDATE, not SELF-REVISION. DESCENDANT PROTOCOL ≠ SOURCE PERSON.
POSTHUMOUS FORK is the moment new decisions begin after the source can no longer revise. Before fork: source history. After fork: descendant history.
Next experiment
THE DESCENDANT PROTOCOL TEST
When does continuation become lineage?
OPEN THE DESCENDANT PROTOCOL TESTWHILE ALIVE, A PERSON CAN SAY: “THAT IS NOT WHAT I MEANT.” AFTER DEATH, THAT CORRECTION MAY NEVER COME.
A FAMILY MAY INTERPRET. AN INSTITUTION MAY DECIDE. A DOCUMENT MAY DIRECT. AN AI MAY INFER. BUT NONE OF THESE IS AUTOMATICALLY THE SOURCE.
THE EXPERIMENTAL QUESTION IS NOT: “WHAT WOULD THE PERSON REALLY WANT?”
IT IS: “WHEN THE SOURCE CAN NO LONGER ANSWER, WHAT EVIDENCE IS ENOUGH TO JUSTIFY ACTION?”
IF THE ANSWER CANNOT BE ESTABLISHED, HOLD MAY BE THE MOST ACCURATE STATE.
UNCERTAINTY AFTER DEATH IS STILL UNCERTAINTY.