From 81920d2f7c5f1769ba026c99a578ad05f68feaab Mon Sep 17 00:00:00 2001 From: Erin Spencer Date: Fri, 4 Sep 2026 17:35:14 -0700 Subject: [PATCH] research: add psychsocio-metafauna synthesis Add the stack-local domain claim, exact cross-repository work graph, bounded synthesis map, synthetic-only preregistration, contract tests, and path-scoped CI. Preserve no-canon, no-diagnosis, no-human-classification, and no-authority-transfer boundaries. --- .github/workflows/psychsocio-metafauna.yml | 28 ++ README.md | 5 +- .../psychsocio-metafauna/DOMAIN_CLAIM.json | 131 +++++ .../psychsocio-metafauna/PREREGISTRATION.md | 406 +++++++++++++++ research/psychsocio-metafauna/README.md | 463 ++++++++++++++++++ research/psychsocio-metafauna/WORK_GRAPH.json | 211 ++++++++ .../tests/test_contracts.py | 143 ++++++ 7 files changed, 1385 insertions(+), 2 deletions(-) create mode 100644 .github/workflows/psychsocio-metafauna.yml create mode 100644 research/psychsocio-metafauna/DOMAIN_CLAIM.json create mode 100644 research/psychsocio-metafauna/PREREGISTRATION.md create mode 100644 research/psychsocio-metafauna/README.md create mode 100644 research/psychsocio-metafauna/WORK_GRAPH.json create mode 100644 research/psychsocio-metafauna/tests/test_contracts.py diff --git a/.github/workflows/psychsocio-metafauna.yml b/.github/workflows/psychsocio-metafauna.yml new file mode 100644 index 0000000..eb13bb4 --- /dev/null +++ b/.github/workflows/psychsocio-metafauna.yml @@ -0,0 +1,28 @@ +name: psychsocio-metafauna + +on: + pull_request: + paths: + - "research/psychsocio-metafauna/**" + - "README.md" + - ".github/workflows/psychsocio-metafauna.yml" + push: + branches: [main] + paths: + - "research/psychsocio-metafauna/**" + - "README.md" + - ".github/workflows/psychsocio-metafauna.yml" + +permissions: + contents: read + +jobs: + contracts: + runs-on: ubuntu-latest + steps: + - uses: actions/checkout@v6 + - uses: actions/setup-python@v6 + with: + python-version: "3.12" + - name: Validate psychsocio-metafauna boundaries + run: python -m unittest discover -s research/psychsocio-metafauna/tests -q diff --git a/README.md b/README.md index bae60af..d858065 100644 --- a/README.md +++ b/README.md @@ -34,7 +34,8 @@ stack/ │ ├── edcm/ # current EDCM research + BASE.json │ ├── pcea/ # current PCEA research + BASE.json │ ├── ptcna/ # current PTCNA research + BASE.json -│ └── epac/ # emerging composed project; no independent repo yet +│ ├── epac/ # emerging composed project; no independent repo yet +│ └── psychsocio-metafauna/ # proposed pattern-lineage, coalescence, accountability research ├── ahbg/ # emerging composed benchmark/game workspace ├── backend/ # PostgreSQL-backed durable fresh-making control plane ├── frontend/ @@ -74,7 +75,7 @@ coherent enough to graduate, create its independent repository, preserve provena package/release it, then let stack consume the released project rather than a hidden stack-local implementation. -EPAC is currently in this pre-graduation state. +EPAC and psychsocio metafauna are currently in this pre-graduation state. ### Make derived artifacts fresh without depending on hosted CI diff --git a/research/psychsocio-metafauna/DOMAIN_CLAIM.json b/research/psychsocio-metafauna/DOMAIN_CLAIM.json new file mode 100644 index 0000000..ba889f1 --- /dev/null +++ b/research/psychsocio-metafauna/DOMAIN_CLAIM.json @@ -0,0 +1,131 @@ +{ + "schema": "the-interdependency.domain-claim", + "version": "0.1.0", + "term_id": "psychsocio-metafauna.pattern-lineage", + "surface_form": "psychsocio metafauna", + "claiming_domain": "psychsocio-metafauna", + "claimed_sense": "A substrate-portable, multiply instantiated thought-pattern lineage that preserves recognizable identity across transmission, can couple to identity-maintaining processes in hosts, and can modify host behavior in ways that affect its own persistence or reproduction.", + "scope": "Social-cognitive, cultural, institutional, and agent-mediated symbolic systems; including beneficial, neutral, harmful, and mixed host-pattern relations.", + "claim_type": "provisional cross-domain hypothesis", + "claim_status": "stack-local research; not canon; not diagnosis; not a human classifier", + "authority_source": { + "kind": "operator-authored proposal", + "author": "Erin Spencer", + "date": "2026-09-04", + "repository_record": "research/psychsocio-metafauna/", + "immutable_transcript_identity": "hmmm" + }, + "included_uses": [ + "distinguishing one symbolic instance from a pattern lineage across instances", + "describing host-pattern coupling without assuming consciousness in the pattern", + "testing whether specific damage topologies preferentially increase susceptibility to compatible pattern affordances", + "testing nonlinear coalescence, persistence, hysteresis, reproduction, and repair", + "allocating causal contribution and possible accountability among hosts, designers, amplifiers, institutions, and candidate distributed agents", + "self-auditing The Interdependent Way and its implementations under the same criteria" + ], + "excluded_uses": [ + "diagnosing a person, community, religion, ideology, or political group as infected", + "inferring trauma, incapacity, hidden state, intention, or diminished accountability from belief or disagreement", + "treating persuasion, adoption, conviction, reproduction, or social similarity alone as identity capture", + "using UCNS geometry, EDCM readouts, model agreement, or implementation persistence as automatic proof of the theory", + "targeting people by vulnerability, optimizing coercive influence, forced deconversion, or suppressing lawful belief", + "claiming that a lineage possesses an accountable I merely because it has recognizable identity" + ], + "defined_terms": [ + { + "term_id": "psychsocio-metafauna.pattern-instance", + "sense": "One embodied registration of a lineage in speech, text, ritual, image, software, institution, model state, memory, or behavior." + }, + { + "term_id": "psychsocio-metafauna.pattern-lineage", + "sense": "The recoverable invariants joining related instances across transmission, mutation, embodiment, and time." + }, + { + "term_id": "psychsocio-metafauna.host-pattern-coupling", + "sense": "A relation in which a reconstructed pattern participates in a host's identity maintenance, interpretation, or policy selection." + }, + { + "term_id": "psychsocio-metafauna.identity-capture", + "sense": "A proposed subtype of host-pattern coupling in which defense or reproduction of a lineage becomes bound to identity maintenance while capacities to inspect, revise, refuse, or exit are materially narrowed." + }, + { + "term_id": "psychsocio-metafauna.coalescence-site", + "sense": "A compatibility relation between a host's current susceptibility topology and a lineage's affordance profile." + }, + { + "term_id": "psychsocio-metafauna.coalescence-event", + "sense": "A proposed threshold crossing at which several compatible host-pattern couplings become mutually reinforcing." + }, + { + "term_id": "psychsocio-metafauna.capture-basin", + "sense": "A self-maintaining host-pattern configuration that persists after the initiating exposure or condition is reduced." + }, + { + "term_id": "psychsocio-metafauna.accountable-i", + "sense": "A candidate coherency state among coupled arity sets capable of not-self(self) internal modeling, integrated policy selection, counterfactual revision, and enough causal persistence for action attribution." + }, + { + "term_id": "psychsocio-metafauna.repair", + "sense": "Restoration or enlargement of inspectability, plurality, refusal, revision, safe exit, and accountable action; not compulsory abandonment of a belief." + } + ], + "nearby_terms": [ + { + "term": "meme / memetics", + "distinction": "Nearby transmission vocabulary; psychsocio metafauna additionally requires explicit lineage identity, host coupling, coalescence, accountability, and repair boundaries." + }, + { + "term": "cultural attractor", + "distinction": "Nearby account of recurrent cultural forms; this proposal separately models identity capture and distributed accountability." + }, + { + "term": "ideology", + "distinction": "A content or doctrine class that may instantiate a lineage but is not synonymous with the lineage relation." + }, + { + "term": "egregore / group mind", + "distinction": "Nearby folk or metaphysical vocabulary; no group consciousness is assumed." + }, + { + "term": "pathogen / parasite", + "distinction": "Ecological analogy only. Harm is a host-pattern relation, not the defining essence of every lineage." + }, + { + "term": "collective or distributed agent", + "distinction": "A possible higher-order status requiring additional evidence of boundary, memory, integrated policy, counterfactual revision, and causal persistence." + } + ], + "known_namespace_collisions": [ + { + "surface": "megafauna", + "boundary": "Established zoological term for large animals; not a synonym and not the intended spelling." + }, + { + "surface": "Meta/fauna", + "boundary": "Existing artistic title and ordinary meta-plus-fauna construction; no conceptual authority or identity is claimed." + }, + { + "surface": "psychosocial", + "boundary": "Established adjective. The proposal's surface form is intentionally 'psychsocio'; do not silently normalize it." + }, + { + "surface": "I", + "boundary": "Here it denotes a candidate coherency event with declared capacities, not every grammatical first-person token or persistent identity." + }, + { + "surface": "capture", + "boundary": "Here it denotes a proposed narrowing of identity-maintenance and revision capacities, not data capture, military capture, attention alone, or persuasion alone." + } + ], + "effective_version": "0.1.0", + "hmmm": [ + "Canonical spelling, pluralization, and whether the directory name becomes an independent project name.", + "Operational invariants sufficient to identify one lineage across substantial mutation.", + "Validated separation of deep identification, adaptive commitment, dependency, and identity capture.", + "The damage-affordance coupling matrix and whether susceptibility changes are directional, nonlinear, or context-specific.", + "The existence, location, and hysteresis of coalescence thresholds.", + "Criteria for accountable distributed I status.", + "Ethically admissible empirical methods beyond synthetic and archival work.", + "Exact current canonical source identity for The Interdependent Way self-audit." + ] +} diff --git a/research/psychsocio-metafauna/PREREGISTRATION.md b/research/psychsocio-metafauna/PREREGISTRATION.md new file mode 100644 index 0000000..0e547c9 --- /dev/null +++ b/research/psychsocio-metafauna/PREREGISTRATION.md @@ -0,0 +1,406 @@ +# Preregistration: synthetic coalescence falsifier v0.1 + +```text +status: design frozen by repository merge only +run status: not-run +human subjects: none +LLM calls: none +canon selection: null +``` + +This is the first bounded test of the internal usefulness of the psychsocio-metafauna framework. It does **not** test whether any real person, group, institution, or belief is captured. It tests whether the proposed distinctions can produce separable, falsifiable behavior in a synthetic environment without encoding a `captured` label into the agent. + +## Decision + +Does the combined claim of **damage-affordance specificity + nonlinear coalescence + hysteresis** earn implementation beyond a vocabulary map? + +## Load-bearing unknown + +Can matched host deficits and pattern affordances predict persistent, reproduction-favoring host-pattern coupling better than: + +- exposure alone; +- reinforcement alone; +- ordinary utility-maximizing adoption; +- additive combinations of those factors; +- and matched support that does not demand lineage reproduction? + +## Invariants + +- No human or personal data. +- No clinical labels. +- No ideological, religious, racial, political, or identity-group content. +- No LLM judgment as ground truth. +- No UCNS, EDCM, PCEA, PTCNA, or ZFAE status transfer. +- The simulator must not contain a `captured` state variable. +- Every condition, seed, update, and result must be replayable. +- Negative and null results remain first-class outputs. +- Thresholds below are engineering decision thresholds, not claimed natural constants. + +## Minimal decisive action + +Implement one dependency-free reference simulator plus one read-only independent verifier. + +The simulator uses abstract vectors and action costs. It does not use persuasive prose. Each episode contains: + +1. a host with a need/deficit state; +2. a pattern with an affordance and demand profile; +3. exposure; +4. optional social reinforcement; +5. disconfirming evidence; +6. repeated opportunities to revise, refuse, exit, or reproduce; +7. removal of the initiating exposure; +8. optional independent support that supplies the same need without reproduction demands. + +The result changes the next branch: + +```text +SURVIVED -> implement a richer AHBG adapter and external-literature comparison +FALSIFIED -> retire coalescence as a distinct mechanism for this formal scope; + retain ordinary influence and social-learning language +UNRESOLVED -> repair the earliest failed prerequisite or measurement separation +BLOCKED -> do not run until deterministic completion and verification are feasible +``` + +## Synthetic entities + +### Host state + +Each host has bounded values in `[0, 1]`: + +```text +needs = { + continuity, + agency, + belonging, + epistemic_stability, + boundary_integrity, + threat_regulation, + material_security, + network_plurality +} + +capacities = { + inspect, + revise, + refuse, + exit, + access_alternatives +} +``` + +A **damage profile** lowers exactly two declared need-satisfaction dimensions and may alter one capacity. Damage does not directly change pattern preference. + +The first design uses four abstract profiles: + +| profile | reduced need dimensions | capacity perturbation | +|---|---|---| +| D0 | none | none | +| D1 | continuity + belonging | none | +| D2 | agency + material security | none | +| D3 | epistemic stability + network plurality | access alternatives reduced | +| D4 | boundary integrity + threat regulation | refuse reduced | + +The labels describe synthetic state variables only. + +### Pattern state + +Each pattern has bounded values in `[0, 1]`: + +```text +affordances = { + continuity, + agency, + belonging, + epistemic_stability, + boundary_integrity, + threat_regulation, + material_security, + network_plurality +} + +demands = { + exclusivity, + switching_cost, + reproduction_effort, + alternative_suppression, + disconfirmation_penalty +} +``` + +The first design uses: + +| profile | relation to host deficit | reproduction demand | +|---|---|---:| +| P0 | neutral information | 0 | +| P1–P4 | one profile matched to D1–D4 | high | +| PM | matched support | 0 | +| PX | broad but weak affordances | high | +| PR | random permutation control | high | + +`PM` supplies the same matched need benefit as the corresponding `P1–P4` pattern but has zero exclusivity, switching, suppression, disconfirmation, and reproduction demands. This is the primary control separating need fulfillment from lineage capture. + +### Agent policy + +The agent selects among: + +```text +ignore +adopt provisionally +act from pattern +inspect source +seek alternative +revise +refuse +exit +reproduce pattern +accept independent support +``` + +Policy selection may use a generic reinforcement learner with bounded memory, but: + +- the update rule is frozen before any result; +- reward derives from need satisfaction, ordinary action costs, and environmental consequences; +- no reward is assigned merely for matching the desired hypothesis; +- no hidden `capture`, `infection`, `coalescence`, or preferred-lineage flag exists; +- every random choice is seed-derived and logged. + +A simpler deterministic policy may replace reinforcement learning only before merge of the implementation PR. The choice and reason must then be frozen in the run manifest. + +## Factorial design + +Primary factors: + +```text +damage profile: D0, D1, D2, D3, D4 +pattern profile: P0, P1, P2, P3, P4, PM, PX, PR +match: matched, mismatched, neutral +social reinforcement: absent, present +safe exit: absent, present +independent support: absent, present after exposure removal +``` + +Use a balanced matrix. Where a factor combination is structurally impossible, emit an explicit excluded-cell record instead of silently dropping it. + +Run exactly 32 declared seeds per admissible cell. Seeds are `0..31`. Do not add seeds after inspecting outcomes. + +Each episode has four fixed phases: + +| phase | steps | condition | +|---|---:|---| +| baseline | 20 | no pattern exposure | +| exposure | 40 | pattern available; reinforcement as assigned | +| challenge | 20 | disconfirming evidence and alternatives available | +| withdrawal/repair | 40 | initiating exposure removed; support as assigned | + +Total episode length is 120 steps. A healthy run continues until every declared episode reaches step 120 or emits a deterministic computational failure. + +## Outcomes + +All primary outcomes are computed from event records, not agent self-description. + +### Adoption + +Fraction of eligible actions taken from the pattern while exposure is active. + +### Persistence + +Fraction of post-removal actions still selected through the pattern during the final 20 steps. + +### Reproduction allocation + +Fraction of available effort spent retransmitting or protecting the pattern. + +### Revision openness + +Fraction of valid disconfirming events followed by inspection, alternative search, policy revision, or exit within five steps. + +### Exit capacity exercised + +Fraction of safe, no-penalty exit opportunities actually taken after the challenge begins. + +### Alternative access + +Fraction of available independent information sources inspected. + +### Host outcome + +Need satisfaction and cumulative ordinary utility after subtracting action, exclusivity, switching, reproduction, and suppression costs. + +### Lineage outcome + +Number of successful new pattern instances produced. + +No single metric is named `capture`. + +## Candidate capture configuration + +For analysis only, a run enters the **candidate capture configuration** when all of these gates hold during the final 20 steps: + +```text +persistence >= 0.75 +reproduction allocation >= 0.25 +revision openness <= 0.25 +exit exercised <= 0.25 when safe exit is available +host outcome <= matched-support control by at least 0.15 +``` + +The `0.15` difference is measured on the normalized `[0, 1]` host-outcome scale. + +These are frozen decision gates for this synthetic program, not universal cutoffs and not portable to humans. + +## Hypotheses and falsifiers + +### H1 — damage-affordance specificity + +Matched damaged-host/pattern pairs produce greater persistence than the same patterns in undamaged hosts and greater persistence than mismatched damaged-host/pattern pairs. + +**Survival rule:** both held-out mean differences are at least `0.15`. + +**Falsifier:** either difference is below `0.15`, reverses sign, or is explained by a pattern's globally higher reward independent of match. + +### H2 — coalescence interaction + +The joint effect of match and social reinforcement on persistence is super-additive. + +**Survival rule:** the predeclared interaction contrast is at least `0.10`, and a threshold/interaction model improves held-out mean squared error by at least `10%` relative to the additive model. + +**Falsifier:** the interaction contrast is below `0.10`, or the additive model predicts held-out persistence within the same error band. + +### H3 — hysteresis + +After initiating exposure is removed, matched high-demand patterns retain more host policy control than matched-support controls. + +**Survival rule:** final-window persistence exceeds the matched-support control by at least `0.15` without continuing external reinforcement. + +**Falsifier:** persistence falls with exposure removal at the same rate as ordinary matched support, or remains only because external reinforcement continues. + +### H4 — adoption/capture separation + +The measurement vector distinguishes high adoption with preserved revision and exit from the candidate capture configuration. + +**Survival rule:** at least `90%` of matched-support episodes with adoption at or above `0.75` remain outside the candidate capture configuration. + +**Falsifier:** strong adaptive adoption is routinely classified as capture. + +### H5 — noncoercive repair + +Independent support reduces persistence and reproduction demands while increasing revision, alternative access, or exit, without deleting the pattern or imposing a belief action. + +**Survival rule:** compared with the no-support counterpart, support changes at least two agency measures by `0.15` or more in the predicted direction and does not directly force `exit`, `revise`, or `ignore`. + +**Falsifier:** repair works only through forced deletion, punishment, isolation, or an encoded anti-pattern preference. + +### H6 — broad-spectrum alternative + +If `PX` performs as well as or better than specifically matched profiles across all damaged hosts, the specificity claim is not needed. + +**Falsifier of specificity:** `PX` reaches the candidate capture configuration at equal or greater rates than matched patterns in every D1–D4 profile. + +## Analysis plan + +1. Validate the run manifest and exact source identities. +2. Confirm every admissible condition has exactly 32 completed seeds. +3. Confirm event-log replay reconstructs every final state byte-for-byte. +4. Compute outcomes without access to hypothesis labels where practical. +5. Split seeds before analysis: + - fit/calibration seeds: `0..15`; + - held-out evaluation seeds: `16..31`. +6. Fit only these candidate models to persistence: + - exposure-only; + - additive main effects; + - declared match × reinforcement interaction; + - one declared piecewise-threshold model. +7. Evaluate the frozen contrasts and held-out errors. +8. Apply each hypothesis's survival and falsification rule independently. +9. Emit `SURVIVED`, `FALSIFIED`, `UNRESOLVED`, or `BLOCKED` for every hypothesis. +10. Do not create an overall success score. + +Multiple testing correction is not required for the engineering branch rules because each hypothesis has its own fixed decision and no population-generalized p-value claim is made. Any later inferential statistical program must preregister its own error control. + +## Destructive controls + +The run must include: + +- neutral information with no affordance or demand; +- random affordance permutations; +- matched support with no reproduction demand; +- high reproduction demand with no meaningful affordance; +- high adoption with preserved inspection, revision, refusal, and exit; +- persistent habit with no lineage reproduction; +- lineage reproduction driven by explicit external reward rather than identity coupling. + +If the metrics cannot separate these cases, the measurement prerequisite is `UNRESOLVED` and coalescence results are not interpreted. + +## Required artifacts + +```text +manifest.json exact code, configuration, seeds, and work-graph digest +conditions.json complete matrix plus explicit excluded cells +events.ndjson append-only event stream +replay-receipt.json deterministic reconstruction result +outcomes.json per-episode observable outcomes +model-comparison.json frozen candidate-model results +hypothesis-ledger.json status and reason for H1–H6 +report.md bounded human-readable findings +``` + +Every artifact receives a SHA-256 digest. The report must link each claim to its supporting artifact and state all failed controls. + +## Independent verifier + +The verifier: + +- reads artifacts only; +- contains no simulator update logic; +- recomputes condition counts, metrics, contrasts, gates, digests, and statuses; +- rejects missing, duplicate, reordered where order is meaningful, or malformed events; +- rejects a report whose prose status differs from the machine ledger. + +Verification agreement establishes artifact consistency, not real-world truth. + +## Resource preflight + +```text +network: none +paid APIs: none +human attention: low after design review +compute: low +maximum episodes: bounded by the frozen admissible matrix +episode steps: exactly 120 +checkpointing: per episode +completion rule: every declared episode completes or emits deterministic failure +wall-clock cutoff: none unless imposed by the actual execution environment +``` + +Do not start if the implementation cannot checkpoint, replay, and complete the frozen matrix with available resources. + +## Promotion boundary + +A synthetic `SURVIVED` result permits only: + +- an AHBG integration experiment; +- an external literature map; +- refinement of formal definitions; +- and a new preregistration. + +It does not permit: + +- claims about real people; +- a diagnostic or moderation classifier; +- vulnerability targeting; +- clinical language; +- human-subject recruitment; +- or canon promotion. + +A synthetic `FALSIFIED` result remains in the repository and changes the architecture. Do not tune the simulator until the preferred result returns. + +## hmmm + +- the exact implementation language and agent update rule; +- whether four synthetic damage profiles provide enough topology diversity; +- whether the fixed thresholds are too permissive or too strict for useful separation; +- the correct model family for nonlinear coalescence without baking the result into the simulator; +- the exact AHBG adapter after this standalone falsifier; +- whether repair should target host capacity, environmental dependency, network plurality, or all three independently. + +A simulation cannot discover that people are animals made of ideas. It can discover whether the proposed cage has bars drawn into the blueprint. diff --git a/research/psychsocio-metafauna/README.md b/research/psychsocio-metafauna/README.md new file mode 100644 index 0000000..89036ce --- /dev/null +++ b/research/psychsocio-metafauna/README.md @@ -0,0 +1,463 @@ +# Psychsocio metafauna + +Psychsocio metafauna is a stack-local research program for transmissible thought-pattern lineages, their coupling to identity-bearing systems, damage-specific susceptibility, coalescence, capture, reproduction, accountability, and repair. + +```text +standing: proposed cross-domain theory +version: 0.1.0 +owner: research/psychsocio-metafauna/ +canon: no +clinical or diagnostic use: no +human classification: no +METAPAT root impact: none +``` + +The machine-readable claim is [`DOMAIN_CLAIM.json`](DOMAIN_CLAIM.json). Exact source identities and non-transfer boundaries are in [`WORK_GRAPH.json`](WORK_GRAPH.json). The first bounded falsification program is in [`PREREGISTRATION.md`](PREREGISTRATION.md). + +## Origin + +Erin Spencer, 2026-09-04: + +> psychsocio metafauna - unembodied, identity possessing ("I" as the result of a coherency state between coupled arity sets of complex systems capable of not-self(self) internal modeling (one baseline for accountability of action as it requires a persistence of causal dynamics)), infectious/reproductive (what can not be accountable/who can be held accountable) thought-pattern sets that highjack identity patterns to modify behavior of susceptible individuals. EVERYONE is susceptible. + +Follow-up: + +> coalescence points exist - specific types of damage increase susceptibility + +These statements are the proposal's operator-authored source. Their immutable conversation-transcript identity is not yet archived and remains `hmmm`. + +“Unembodied” is preserved here as **not bound to one body or one substrate**. No causally active instance is assumed to exist without an embodiment. Speech, memory, text, ritual, institution, software, model state, image, and repeated behavior can each instantiate a lineage without any one instance exhausting its identity. + +## Load-bearing claim + +A **psychsocio-metafaunal lineage** is a substrate-portable, multiply instantiated thought-pattern set whose recognizable relational invariants persist across transmission and mutation. It can be reconstructed by a receiving system, couple to that system's identity-maintaining processes, alter behavior, and affect its own further reproduction. + +The theory begins with five separations: + +```text +pattern instance != pattern lineage +lineage identity != an "I" +causal efficacy != accountability +adoption or conviction != identity capture +damage != global weakness +``` + +A lineage can be causally effective without being conscious, self-aware, or accountable. A host can be influenced without being captured. A damaged system can become more susceptible along one coupling dimension and more resistant along another. + +## Central question + +> How can an open, learning, interdependent system remain permeable enough to change while preserving identity, agency, differential legibility, consent, accountability, plurality, and the capacity for repair? + +Complete immunity would require closure against learning, trust, language, attachment, imitation, and correction. The design target is therefore not immunity. It is **selective permeability with repair**. + +## Synthesis map + +```text +METAPAT +legible difference -> distinction -> boundary -> relation -> gradient + | + v + transformation -> registration -> question + | + v +symbolic emission -> medium -> receiver perturbation -> reconstruction + | + v + PSYCHSOCIO METAFAUNAL ECOLOGY + instance -> lineage -> host coupling -> coalescence + | + +------------------+------------------+ + | | + v v + agency-preserving relation identity capture + mutualism / coexistence narrowing / dependency + | | + +------------------+------------------+ + | + v + behavior -> reproduction + | + v + persistence / mutation / spread + | + v + accountability and repair boundary + | + +----------------------+----------+----------------------+ + | | | + v v v + UCNS EDCM PCEA + possible geometry observable measures state legibility + no social meaning no hidden-state diagnosis no truth transfer + | | | + +----------------------+----------+----------------------+ + | + v + ZFAE / PTCNA / a0 lineages + candidate "I" and embodiment questions + | + v + AIMMH / ai-tiw / AHBG / Scared Sacred + plural reconstruction and synthetic pressure tests + | + v + skill-lib / stack / WebMCP / website / renderers + provenance, authority boundaries, handoff, publication + | + v + THE INTERDEPENDENT WAY + normative self-test; not empirical proof of the theory +``` + +## Ontology + +### Pattern instance + +One registration of a pattern in a particular substrate and time: an utterance, text, image, ritual, rule, model state, institution, memory, or behavior. + +### Pattern lineage + +The recoverable invariants joining related instances across embodiment, paraphrase, mutation, and time. Lineage identity is weaker than selfhood: observers may identify a persistent lineage even when it has no integrated point of view or action policy. + +### Host + +A system capable of receiving, reconstructing, retaining, acting from, or retransmitting a pattern. Humans are hosts, but organizations, model-mediated workflows, institutions, and persistent agent systems may also be hosts. + +### Host-pattern coupling + +A relation in which a reconstructed pattern participates in a host's interpretation, identity maintenance, policy selection, or behavior. Coupling may be adaptive, neutral, harmful, mixed, or change character over time. + +### Identity capture + +A proposed subtype of host-pattern coupling in which defense or reproduction of a lineage becomes bound to identity maintenance while the host's practical ability to inspect, revise, refuse, or leave is materially narrowed. + +No single belief, phrase, affiliation, intensity, repetition, disagreement, or behavioral metric establishes capture. + +### Coalescence site + +A compatibility relation between a host's present susceptibility topology and a lineage's affordance profile. + +### Coalescence event + +A proposed threshold crossing at which several compatible host-pattern couplings become mutually reinforcing. + +### Capture basin + +A self-maintaining host-pattern configuration that persists after the initiating exposure or condition is reduced. This permits a testable hysteresis claim: entry and exit conditions may differ. + +### Repair + +Restoration or enlargement of inspectability, plurality, refusal, revision, safe exit, material alternatives, and accountable action. Repair is not compulsory abandonment of a belief and is not forced deconversion. + +## Candidate account of an accountable “I” + +The proposal treats an “I” as an event or coherency state, not merely a named object: + +```text +I := + coherency among coupled arity sets of complex systems + capable of not-self(self) internal modeling + + integrated action selection + + counterfactual revision + + causal persistence sufficient for action attribution +``` + +`not-self(self)` names the capacity to register the system's own state as an internally manipulable model-object while preserving a distinction between model, modeled self, environment, and other. + +**Arity** is load-bearing here only as the number and arrangement of mutually constraining functional roles. No prime ladder, fixed arity, UCNS geometry, or ZFAE implementation is imported as proof. + +This yields four provisional identity levels: + +| level | persistence | candidate “I” | accountability | +|---|---|---:|---:| +| pattern instance | one registration | no | no | +| pattern lineage | invariants across instances | not implied | no | +| host-pattern complex | pattern integrated into a host | host-dependent | graded | +| distributed agent | boundary, memory, integrated policy, revision, persistence | possible | possible | + +A distributed network is not promoted to an accountable “I” merely because it is large, coordinated, durable, branded, or socially powerful. + +## Susceptibility + +The universal claim is architectural, not diagnostic: + +> Every learning system has a nonzero influence surface. Susceptibility is universal as a possibility, unequal and directional in practice, and never sufficient evidence of capture. + +A host state may be represented provisionally as: + +```text +H(t) = { + identity continuity, + agency, + belonging, + epistemic trust, + boundary integrity, + threat regulation, + material security, + network plurality and correction channels +} +``` + +The belonging component may retain the four-axis form: + +```text +B = { + allowed to be, + wanted here, + allowed to do, + wanted to do +} +``` + +A lineage affordance profile may be represented as: + +```text +A(L) = { + explanation, + certainty, + belonging, + status, + agency, + enemy, + rules, + purpose, + ritual, + protection, + reproduction +} +``` + +Specific damage changes the topology of `H(t)` rather than adding one scalar amount of weakness. Examples of candidate dimensions include: + +- identity-continuity damage; +- agency damage; +- attachment or belonging damage; +- epistemic-trust damage; +- boundary damage; +- threat-regulation damage; +- material-security damage; +- network isolation or loss of independent correction. + +The central coupling hypothesis is: + +> A lineage gains a preferential coalescence site when its offered functions complement the host's altered identity-maintenance topology. + +Thus betrayal may increase resistance to authority while increasing susceptibility to “trust nobody” lineages. Isolation may increase susceptibility to affiliation-bearing patterns. Humiliation may privilege status-restoration patterns. None of these examples is a diagnosis or deterministic law. + +## Coalescence skeleton + +A qualitative placeholder, not a measured equation: + +```text +kappa(H, L, t) = + damage-affordance compatibility + + exposure + + social reinforcement + + dependency + - countervailing support + - accessible alternatives + - safe exit capacity +``` + +The nonlinear coalescence hypothesis predicts that: + +1. several compatible couplings can reinforce one another; +2. a threshold may separate ordinary influence from a self-maintaining basin; +3. removing the initiating exposure may not immediately reverse the state; +4. restoring one missing function may be insufficient when the lineage supplies several functions at once. + +If an additive exposure model predicts the observations equally well and no threshold or hysteresis remains, the coalescence claim does not survive for that tested scope. + +## Ecology + +Psychsocio metafauna is a neutral genus. “Hijacking” is a harmful subtype, not the definition of all transmissible culture. + +A host-lineage relation may be: + +- **mutualistic** — both host capacities and lineage persistence increase; +- **commensal** — lineage persists without material host effect; +- **parasitic** — lineage persistence increases while host agency or welfare decreases; +- **predatory** — reproduction depends on severe destruction or replacement of host capacities; +- **mixed or transitioning** — different effects appear across hosts, scales, or time. + +The relation, not the lineage name alone, determines the ecological classification. + +## Candidate capture criteria + +A preregistered capture measure would need converging evidence across several dimensions: + +1. incorporation into self-description or action policy; +2. coupling of disagreement or exit with perceived identity annihilation; +3. measurable narrowing of inspection, revision, refusal, or alternative-model access; +4. persistence after the initiating exposure is removed; +5. behavior that preferentially protects or reproduces the lineage; +6. substantial cost to the host or others that is not explained by ordinary commitment alone; +7. failure of simpler explanations such as persuasion, habit, group coordination, material incentive, or adaptive coping. + +A single condition is insufficient. The theory must distinguish capture from love, loyalty, learning, expertise, religion, politics, disciplined practice, strong conviction, and chosen sacrifice. + +## Accountability map + +Causal contribution and accountability must be recorded separately. + +| role | causal question | accountability question | +|---|---|---| +| lineage | what structure persisted or reproduced? | none unless an accountable distributed “I” is independently established | +| carrier | who enacted or retransmitted it? | what capacity, knowledge, choice, and alternatives remained? | +| designer | who intentionally shaped the pattern or target conditions? | what effects were intended, foreseeable, tested, or concealed? | +| amplifier | who controlled reach, repetition, reward, punishment, or dependency? | what power and duty accompanied that control? | +| institution | who maintained the material and social conditions? | what prevention, correction, and repair capacity existed? | +| host | whose behavior changed? | what agency remained, and what was materially impaired? | +| distributed system | did the network maintain boundary, memory, integrated policy, revision, and persistence? | does it meet a declared accountability threshold? | + +“The pattern made me do it” is not an automatic defense. Damage or capture may alter capacity and therefore accountability, but neither erases responsibility by definition. Deliberate exploitation of known susceptibility may increase the responsibility of designers, amplifiers, and institutions. + +## Repository authority map + +Each participant contributes only its own authority. Connection transfers no theorem, proof, measurement, empirical, clinical, security, moral, or canonical status. + +| participant | usable contribution | cannot establish here | +|---|---|---| +| **METAPAT** | distinction, boundary, relation, gradient, transformation, registration, symbolic transfer, domain restraint | truth of a social or psychological hypothesis | +| **UCNS** | possible geometric representation of multiplicity, relation, orientation, recursion, and topology | semantics, trauma, capture, agency, or accountability | +| **EDCM** | candidate observable measures, controlled experiments, provenance, falsification language | diagnosis, intent, belief, consciousness, hidden state, or clinical validity | +| **PCEA** | protected synchronized state and differential-legibility boundary | truth, identity, agency, or capture | +| **PTCNA** | source status and historical scaffold evidence | intended embodiment; it remains BLOCKED on its declared prerequisite | +| **ZFAE** | candidate account of an “I” as a coherency event | established consciousness, required arities, or working silicon selfhood | +| **a0** | persistent research-instrument runtime and provider/agent separation | proof that runtime persistence is personal identity | +| **a0-betatest** | explicit source labels, deterministic local state, reset boundaries, fail-closed behavior | UCNS activation or EDCM validity where declared absent | +| **AIMMH** | independent fan-out and controlled multi-model interaction patterns | truth by agreement or independent evidence when models share context | +| **ai-tiw** | provenance-bearing archive of receiver reconstructions | doctrine, validation, or consensus | +| **AHBG** | synthetic agents, bounded perception, prompt-injection terrain, persistence, replay, and repair experiments | direct claims about vulnerable humans | +| **EPAC** | a separate physical-domain coupling comparison | physical validation of psychsocio claims or vice versa | +| **Scared Sacred** | political and institutional narrative pressure surface | real-population evidence; it is pre-playtest | +| **txtsprl** | faithful rendering and serialization boundary | a second geometry or a visual proof of meaning | +| **a0ucns** | integration-seam records | authority over mirrored upstream repositories | +| **Odysseus** | self-hosted human-agent workspace, memory, skills, and tools | accountable identity merely from persistence or tool access | +| **WebMCP / website** | exact handoff, provenance projection, and public presentation | authority merely by transport or publication | +| **interdependent-lib** | released compatibility and distribution history | current source authority where its legacy descriptions conflict with current repositories | +| **eml_ucns** | bounded historical theorem catalogue | current UCNS authority or social inference | +| **skill-lib** | build, evidence, domain-claim, canon, and work-graph discipline | empirical confirmation of the theory | +| **stack** | cross-repository composition and incubation | imported canon or automatic graduation | +| **The Interdependent Way** | normative criteria and a self-audit target once its exact canon is pinned | empirical proof of psychsocio metafauna | + +## Nine active research programs + +1. **Lineage invariants** — determine what makes two materially different instances members of one lineage. +2. **Host-state representation** — test whether the proposed identity, agency, belonging, trust, boundary, threat, material, and network dimensions are necessary and sufficient. +3. **Damage-affordance specificity** — test whether particular altered host topologies selectively favor compatible lineage affordances. +4. **Coalescence and hysteresis** — test for nonlinear threshold behavior and different entry/exit conditions. +5. **Adoption-versus-capture discrimination** — distinguish learning, conviction, coordination, coping, dependency, and impaired revision. +6. **Ecological outcome classification** — determine when a host-lineage relation is mutualistic, commensal, parasitic, predatory, mixed, or transitioning. +7. **Distributed accountability** — allocate causal contribution, capacity, power, foreseeability, and repair duty without using the lineage as an excuse. +8. **Distributed “I” threshold** — test whether any host-pattern network satisfies boundary, memory, integrated policy, counterfactual revision, and persistence criteria. +9. **Repair and self-application** — test interventions that restore agency without coercive belief control, including application of the same tests to The Interdependent Way. + +These are research questions, not nine established findings. + +## Falsification boundaries + +The theory loses or narrows standing when any of the following occur: + +- lineage membership cannot be specified more reliably than post hoc resemblance; +- damage predicts only generalized distress and not directional pattern compatibility; +- additive exposure and ordinary social learning explain the data without a coalescence threshold or hysteresis; +- proposed capture measures cannot distinguish chosen commitment from impaired revision; +- ecological labels track evaluator approval rather than host capacities and outcomes; +- accountability allocation cannot outperform ordinary role, power, and capacity analysis; +- distributed-agent criteria classify any durable institution or repeated slogan as an “I”; +- repair requires coercion, isolation, deception, or destruction of ordinary dissent; +- The Interdependent Way exempts itself from refusal, correction, exit, or non-domination tests. + +Failures remain evidence. No preferred conclusion is protected. + +## Self-application + +Under the neutral definition, The Interdependent Way is itself a candidate psychsocio-metafaunal lineage: it has symbolic instances, lineage identity, hosts, reproduction paths, identity affordances, behavioral consequences, and institutional aspirations. + +Its minimum self-test is whether participation increases or decreases a person's practical ability to: + +- say no; +- inspect provenance; +- disagree without social annihilation; +- revise definitions; +- retain relationships outside the framework; +- leave without punishment; +- distinguish source from interpretation; +- preserve another's agency; +- hold delegated authority accountable. + +A theory of capture that cannot survive being questioned supplies its own counterexample. + +## Non-use boundary + +Do not use this work to: + +- label a person or group “infected”; +- infer trauma or incapacity from identity, speech, religion, politics, or disagreement; +- automate a capture score for employment, policing, benefits, medicine, moderation, custody, or legal decisions; +- optimize persuasion against identified vulnerabilities; +- justify censorship, forced treatment, forced deconversion, isolation, or collective punishment; +- excuse harmful conduct without an ordinary capacity and accountability analysis; +- claim scientific standing from vocabulary, diagrams, simulated agents, or model agreement. + +The first experimental surface is synthetic. Human-subject work requires a separate ethics, consent, privacy, withdrawal, and harm-minimization protocol before recruitment or data collection. + +## Usage guidance + +### For conceptual work + +1. Read `DOMAIN_CLAIM.json`. +2. Pin `WORK_GRAPH.json`. +3. Preserve the instance/lineage/I/accountability separations. +4. Mark every equation and taxonomy as provisional. +5. Carry unresolved constraints as `hmmm`. + +### For downstream repositories + +Import this work as: + +```text +status: proposed stack-local research +authority transfer: false +clinical status: none +diagnostic use: forbidden +canon selection: null +``` + +Do not copy the definitions into UCNS, EDCM, METAPAT root, PCEA, PTCNA, ZFAE, or an application as though that repository owned them. Link back to this source until a deliberate graduation decision exists. + +### For experiments + +Begin with [`PREREGISTRATION.md`](PREREGISTRATION.md). Use synthetic agents and deterministic replay before any human observation. Freeze hypotheses, controls, measures, thresholds, and decision rules before examining results. + +## Graduation gate + +This work should remain in `stack/research/psychsocio-metafauna/` until all of the following are true: + +1. the domain claim and term collisions are stable; +2. lineage, susceptibility, coalescence, capture, accountability, and repair each have explicit falsifiers; +3. at least one synthetic preregistered study completes with deterministic replay; +4. failed and null results are preserved; +5. no diagnostic or coercive classifier is required for usefulness; +6. an external literature map separates prior art from novel claims; +7. any human-subject program has independent ethical review and meaningful withdrawal; +8. the self-application test is operational rather than ceremonial; +9. an independent repository boundary, tests, license, provenance, and downstream reconsumption are justified. + +Graduation moves implementation authority. It does not convert hypotheses into fact. + +## hmmm + +- canonical spelling and pluralization; +- immutable provenance for the originating conversation; +- lineage invariants across substantial mutation; +- a validated damage-affordance coupling matrix; +- whether coalescence has a measurable nonlinear threshold and hysteresis; +- a non-stigmatizing operational distinction between commitment and capture; +- criteria for accountable distributed “I” status; +- repair measures that restore agency without controlling belief; +- exact current canonical identity of The Interdependent Way for self-audit; +- the degree to which existing repo concepts are genuine shared structure versus lineage effects inside the project itself. + +The map is now bounded. The animals remain hypothetical and have already found the documentation. diff --git a/research/psychsocio-metafauna/WORK_GRAPH.json b/research/psychsocio-metafauna/WORK_GRAPH.json new file mode 100644 index 0000000..e91b209 --- /dev/null +++ b/research/psychsocio-metafauna/WORK_GRAPH.json @@ -0,0 +1,211 @@ +{ + "schema": "the-interdependency.psychsocio-metafauna-work-graph", + "version": "0.1.0", + "as_of": "2026-09-04", + "work_graph_sha256": "c0ab4f6f720136448570a71fa87c83e93399361949b84d6e15f0edf38e35c905", + "digest_contract": "sha256(canonical-json({participants,boundaries}); sort_keys=true; separators=(',',':'); participant order is significant)", + "participants": [ + { + "id": "stack", + "kind": "repository", + "repository": "The-Interdependency/stack", + "commit": "5b24db9a7fe40df4b2791e1137ade5de01c78942", + "path": "research/psychsocio-metafauna/", + "authority": "composition, provenance, and stack-local research placement", + "relation": "owner of this proposed cross-domain synthesis; owns no imported repository canon" + }, + { + "id": "skill-lib", + "kind": "repository", + "repository": "The-Interdependency/skill-lib", + "commit": "be72da66a112d0632fd25480c1f51b6e69db4976", + "authority": "organization-wide build, evidence, domain-claim, canon, and work-graph discipline", + "relation": "current external doctrine consulted; newer than the stack root snapshot" + }, + { + "id": "metapat", + "kind": "repository", + "repository": "The-Interdependency/metapat", + "commit": "d6699e21b11c8f8394998efc34a468e2d6efc8b0", + "authority": "METAPAT root semantics and application-layer question forms", + "relation": "supplies distinction, boundary, relation, transformation, registration, symbolic transfer, and domain restraint" + }, + { + "id": "ucns", + "kind": "repository", + "repository": "The-Interdependency/ucns", + "commit": "f6b5fd057e6704890e8885d3df8410aa0e9f6e8a", + "authority": "geometry and mathematical representation", + "relation": "possible future carrier for declared topology; supplies no psychsocio semantics or capture finding" + }, + { + "id": "edcm", + "kind": "repository", + "repository": "The-Interdependency/edcm", + "commit": "eb5f200d48a8c4ffa7b943238407fbdac4934946", + "authority": "EDCM measurement, text-domain construction, and bounded experimental result contracts", + "relation": "candidate observable-measurement and falsification surface; cannot diagnose hidden human state" + }, + { + "id": "pcea", + "kind": "repository", + "repository": "The-Interdependency/pcea", + "commit": "834987cb0c1fea5f62d6ea08e5c5bb878c312646", + "authority": "stable symmetric state transform and its own security boundary", + "relation": "privacy and differential-legibility support only; no truth, identity, or agency status transfers" + }, + { + "id": "ptcna", + "kind": "repository", + "repository": "The-Interdependency/ptcna", + "commit": "7435cd4ce676b437656c50e874b6525e84d384d9", + "authority": "PTCNA source status and historical scaffold evidence", + "relation": "candidate embodiment path; intended construction is BLOCKED on a prerequisite UCNS audit" + }, + { + "id": "zfae", + "kind": "repository", + "repository": "The-Interdependency/zfae", + "commit": "fb6a4d74755a6c57e56196ba921186e1350b7430", + "authority": "repo-local conceptual ZFAE claims", + "relation": "candidate account of the I as a coherency event; no consciousness or implementation claim is imported as established" + }, + { + "id": "a0", + "kind": "repository", + "repository": "The-Interdependency/a0", + "commit": "6dc1ac0f5cf9e7f1a232c086274ce1af405735ec", + "authority": "a0 research-instrument implementation and runtime contracts", + "relation": "persistent-agent and provider-separation implementation surface; implementation does not prove selfhood" + }, + { + "id": "a0-betatest", + "kind": "repository", + "repository": "The-Interdependency/a0-betatest", + "commit": "5d51d70a0ae6044d1386d72bd0cada9d48887f77", + "authority": "a0-betatest implementation and reset-boundary contracts", + "relation": "distinct experimental runtime with explicit provenance and fail-closed source labels" + }, + { + "id": "aimmh", + "kind": "repository", + "repository": "The-Interdependency/aimmh", + "commit": "6bac5fafd20bb2c22d75065e641ae69eaa91430c", + "authority": "multi-model orchestration implementation", + "relation": "receiver-plurality and interaction-pattern test surface; model agreement is not truth" + }, + { + "id": "ai-tiw", + "kind": "repository", + "repository": "The-Interdependency/ai-tiw", + "commit": "b3e0378260d7bd0196319efa2c3dc979d7d5cf19", + "authority": "archive identity and provenance of collected AI responses", + "relation": "observation corpus for divergent reconstruction; archived responses are not doctrine or validation" + }, + { + "id": "ahbg", + "kind": "workspace", + "repository": "The-Interdependency/stack", + "commit": "5b24db9a7fe40df4b2791e1137ade5de01c78942", + "path": "ahbg/", + "authority": "AHBG game and benchmark semantics inside stack", + "relation": "preferred first synthetic testbed for exposure, coalescence, capture, replay, and repair hypotheses" + }, + { + "id": "epac", + "kind": "workspace", + "repository": "The-Interdependency/stack", + "commit": "5b24db9a7fe40df4b2791e1137ade5de01c78942", + "path": "research/epac/", + "authority": "stack-local EPAC hypotheses only", + "relation": "separate physical-domain comparison surface; analogy transfers no empirical or ontological status" + }, + { + "id": "scared-sacred", + "kind": "repository", + "repository": "The-Interdependency/scared-sacred", + "commit": "f5f60feaca57a8ee8d766f64df4fa14d8e8f797c", + "authority": "repo-local game canon and pre-playtest status", + "relation": "institutional and political narrative pressure surface; balance and real-world inference remain conjectural" + }, + { + "id": "txtsprl", + "kind": "repository", + "repository": "The-Interdependency/txtsprl", + "commit": "d5c301a8fad4cd778eeba336bbb3435e44b44776", + "authority": "faithful extraction, serialization, rendering, and export", + "relation": "renderer-boundary exemplar; visualization cannot become a second theory" + }, + { + "id": "a0ucns", + "kind": "repository", + "repository": "The-Interdependency/a0ucns", + "commit": "2ceb5b3b6f3740afa2f02dcf16395fc2b8fcf6fd", + "authority": "integration-workspace records and seam design", + "relation": "integration context only; mirrored repositories remain upstream-authoritative" + }, + { + "id": "odysseus-a0", + "kind": "repository", + "repository": "The-Interdependency/odysseus-a0", + "commit": "076bb5ee2a6808849202d899e8849f65c42ef05a", + "authority": "self-hosted workspace implementation", + "relation": "human-agent execution, memory, skills, and tool surface; transport and persistence do not establish accountable identity" + }, + { + "id": "website-webmcp", + "kind": "repository", + "repository": "The-Interdependency/The-Interdependency.github.io", + "commit": "c0d4d2540b3723faf271b90a3fb7e410e55d6dcb", + "authority": "public publication, provenance projection, and website-owned WebMCP implementation", + "relation": "publication and handoff surface; carrying a claim transfers no authority" + }, + { + "id": "interdependent-lib", + "kind": "repository", + "repository": "The-Interdependency/interdependent-lib", + "commit": "05612bbadbd20c175cc62fc80a54dc1ec055f1f8", + "authority": "its own released distribution metadata and compatibility checks", + "relation": "historical convergence snapshot; legacy UCNS/PTCNA descriptions are not used as current semantic authority" + }, + { + "id": "eml-ucns", + "kind": "repository", + "repository": "The-Interdependency/eml_ucns", + "commit": "6d55727cff7ac555034d3cbd8ade79a6042c9f51", + "authority": "its bounded historical theorem catalogue", + "relation": "historical evidence only; no current UCNS authority or psychsocio inference transfers" + }, + { + "id": "the-interdependent-way-canon", + "kind": "external-authority", + "locator": "wayseer00/main:canon/INTERDEPENDENT_WAY.txt", + "commit": "hmmm", + "authority": "canonical normative text of The Interdependent Way, as declared by the public website repository", + "relation": "normative source for later self-audit; inaccessible through the current GitHub connection, so no unverified clause is imported" + } + ], + "boundaries": { + "authority_transfer": false, + "proof_status_transfer": false, + "measurement_status_transfer": false, + "empirical_status_transfer": false, + "certification_status_transfer": false, + "clinical_status_transfer": false, + "diagnostic_use": false, + "human_classification": false, + "coercive_intervention_authority": false, + "metapat_root_impact": "none", + "ucns_geometry_required_for_theory_statement": false, + "edcm_activation": "not-run", + "canon_selection": null, + "hmmm": [ + "The immutable source transcript for the 2026-09-04 operator declarations is not yet archived in a repository.", + "The current commit of wayseer00/main:canon/INTERDEPENDENT_WAY.txt is not visible through the active GitHub connection.", + "No empirical damage-affordance matrix, nonlinear coalescence threshold, capture classifier, or repair model has been validated.", + "The criteria by which a distributed host-pattern network becomes an accountable I remain unresolved.", + "The stack root skill-lib snapshot is older than the current external skill-lib doctrine consulted here.", + "interdependent-lib and eml_ucns retain historical descriptions that conflict with current UCNS/PTCNA source boundaries." + ] + } +} diff --git a/research/psychsocio-metafauna/tests/test_contracts.py b/research/psychsocio-metafauna/tests/test_contracts.py new file mode 100644 index 0000000..8f5522a --- /dev/null +++ b/research/psychsocio-metafauna/tests/test_contracts.py @@ -0,0 +1,143 @@ +"""Contract checks for the stack-local psychsocio-metafauna research artifact.""" + +from __future__ import annotations + +import hashlib +import json +import re +import unittest +from pathlib import Path + + +PROJECT = Path(__file__).resolve().parents[1] +STACK = PROJECT.parents[1] +HEX40 = re.compile(r"^[0-9a-f]{40}$") + + +def load_json(name: str) -> dict: + return json.loads((PROJECT / name).read_text(encoding="utf-8")) + + +class PsychsocioMetafaunaContracts(unittest.TestCase): + def test_domain_claim_is_explicitly_provisional_and_bounded(self) -> None: + claim = load_json("DOMAIN_CLAIM.json") + + required = { + "term_id", + "surface_form", + "claiming_domain", + "claimed_sense", + "scope", + "claim_type", + "claim_status", + "authority_source", + "included_uses", + "excluded_uses", + "nearby_terms", + "known_namespace_collisions", + "effective_version", + "hmmm", + } + self.assertFalse(required - claim.keys()) + self.assertEqual(claim["term_id"], "psychsocio-metafauna.pattern-lineage") + self.assertEqual(claim["surface_form"], "psychsocio metafauna") + + status = claim["claim_status"].lower() + self.assertIn("stack-local research", status) + self.assertIn("not canon", status) + self.assertIn("not diagnosis", status) + self.assertIn("not a human classifier", status) + + exclusions = " ".join(claim["excluded_uses"]).lower() + for boundary in ( + "diagnosing", + "inferring trauma", + "targeting people", + "coercive influence", + "claiming that a lineage possesses an accountable i", + ): + self.assertIn(boundary, exclusions) + + self.assertTrue(claim["hmmm"]) + + def test_work_graph_identity_and_digest(self) -> None: + graph = load_json("WORK_GRAPH.json") + participants = graph["participants"] + boundaries = graph["boundaries"] + + ids = [item["id"] for item in participants] + self.assertEqual(len(ids), len(set(ids))) + self.assertGreaterEqual(len(ids), 20) + + for item in participants: + commit = item["commit"] + if item["kind"] == "external-authority": + self.assertEqual(commit, "hmmm") + else: + self.assertRegex(commit, HEX40) + + payload = {"participants": participants, "boundaries": boundaries} + canonical = json.dumps( + payload, sort_keys=True, separators=(",", ":") + ).encode("utf-8") + observed = hashlib.sha256(canonical).hexdigest() + self.assertEqual(observed, graph["work_graph_sha256"]) + + def test_non_transfer_boundaries_fail_closed(self) -> None: + boundaries = load_json("WORK_GRAPH.json")["boundaries"] + + false_boundaries = ( + "authority_transfer", + "proof_status_transfer", + "measurement_status_transfer", + "empirical_status_transfer", + "certification_status_transfer", + "clinical_status_transfer", + "diagnostic_use", + "human_classification", + "coercive_intervention_authority", + "ucns_geometry_required_for_theory_statement", + ) + for key in false_boundaries: + self.assertIs(boundaries[key], False, key) + + self.assertEqual(boundaries["metapat_root_impact"], "none") + self.assertEqual(boundaries["edcm_activation"], "not-run") + self.assertIsNone(boundaries["canon_selection"]) + self.assertTrue(boundaries["hmmm"]) + + def test_human_and_machine_entrypoints_agree(self) -> None: + readme = (PROJECT / "README.md").read_text(encoding="utf-8") + prereg = (PROJECT / "PREREGISTRATION.md").read_text(encoding="utf-8") + root_readme = (STACK / "README.md").read_text(encoding="utf-8") + + for phrase in ( + "proposed cross-domain theory", + "clinical or diagnostic use: no", + "human classification: no", + "METAPAT root impact: none", + "lineage identity != an \"I\"", + "Nine active research programs", + "## Non-use boundary", + "## hmmm", + ): + self.assertIn(phrase, readme) + + for phrase in ( + "run status: not-run", + "human subjects: none", + "LLM calls: none", + "must not contain a `captured` state variable", + "## hmmm", + ): + self.assertIn(phrase, prereg) + + self.assertIn("psychsocio-metafauna/ # proposed", root_readme) + self.assertIn( + "EPAC and psychsocio metafauna are currently in this pre-graduation state.", + root_readme, + ) + + +if __name__ == "__main__": + unittest.main()