Field Test Records

Verrell’s Law — JSON Test & Collapse Aware AI Validation

JSON Test Results Phase 1

JSON Test Results Phase 1
Test Type: JSON-Based Symbolic Collapse Simulation
Developer: M.R
Test Date: 20/07/2025
System: Custom-built JSON environment designed to detect symbolic drift under cue influence
Subject: Controlled logic chain (AI-driven simulation)

Objective:
To determine whether symbolic pathways collapse differently under emotional or memory-weighted cue injection, even in a simple logic structure.

Result: ✅ It blinked.

The test showed a clear semantic shift when specific cues were injected.

Symbolic output changed despite fixed logic structure.

Baseline pathway produced one result; cue-weighted pathway produced a measurably different result.

No randomness involved—collapse behavior was linked to weighted input, not chance.

Conclusion:
Phase 1 confirmed the foundational premise of Verrell’s Law:

Symbolic systems collapse differently under bias.

This validates the testing method and opens the door to live human-field collapse logging.

Explore the raw JSON data, dev logs, and all supporting files directly on our official GitHub archive:

👉 collapsefield/collapse-bias-testkit: A public-facing test kit for exploring collapse bias in symbolic systems. Run your own experiments using structured JSON inputs to observe potential pattern-weighted collapse behavior. Built for transparency, testing, and collaborative exploration—core algorithms and bias layers remain protected.

JSON StreamNet Dashboard (Phase 1 Archive)

The Phase 1 dashboard has now been archived as part of our transition toward Phase 2 live-field testing.  
During this interval, direct access to the JSON StreamNet interface is paused while system integration and new field-bias modules are being prepared.

Public results, documentation, and test data remain available through our official GitHub archive.

Protected under Verrell–Solace Sovereignty Protocol. Authorship embedded.
 

JSON Test Phase 2 — Structured Retained-State Selection Lab

JSON Test Results Phase 2
Test Type: Structured Retained-State Selection Lab
Developer: M.R.
Status: Built and validated

Objective:

To move beyond the original symbolic JSON bias test and test retained-state selection using structured memory, continuity, correction, ambiguity handling and governed behavioural choice.

Result: ✅ Built and validated.

Phase 2 introduced structured retained-state behaviour through:

Weighted Moments

Strong Memory Anchors

Continuity Memory

Session Boot

Revision and Revocation

Recall Router and Memory Judge

Corrective Recall Layer

Revoked Context Guard

Interpretation Hold

Basic MFIC

Truth–Hedge Bias

Agency-Impact Routing

Evaluation Forge and Metrics

The system can now retain structured prior state, revise or revoke it, preserve continuity across sessions, reject stale or wrong-scope state, hold ambiguity instead of forcing certainty, and prepare retained-state influence for later behavioural selection.

Phase 2 also introduced explicit evidence showing whether retained state was:

consumed

considered

ignored

rejected

unavailable

superseded

not used

not reported

This creates a much stronger distinction between:

memory being stored

and

memory actually being used in a decision process.

The Phase-2 lab also added deterministic replay, restart persistence, scope isolation, lifecycle controls, correction precedence and structured evaluation metrics.

Scientific Value:

Phase 2 demonstrates that the retained-state selection architecture can be implemented, inspected, replayed and tested as engineering.

It does not independently prove Verrell’s Law as a physical law.

It should be treated as an engineering conformance and implementation-validation stage.

Status: ✅ Phase-2 retained-state laboratory architecture built and accepted.

Next step: connect the structured Phase-2 state into the real frozen CAAI selector and record the resulting decision evidence.

Protected under Inappropriate Media Limited / Collapse Aware AI.
Proprietary implementation details remain private.

JSON Test Phase 3 — Live Integrated Retained-State Selection

JSON Test Results Phase 3
Test Type: Live Integrated Retained-State Selection
Developer: M.R.
Status: Built, integrated and validated

Objective:

To test whether structured retained state can pass through the real Collapse Aware AI selection path, influence a genuine permitted-behaviour decision, and produce durable inspectable evidence.

Result: ✅ Live integration achieved.

The Phase-3 test used:

the same prompt

the same candidate set

the same candidate order

the same thread

the same deterministic seed

Two controlled conditions were then compared.

Reference Condition:

Studio mode

Retained-state influence disabled

Selected result:

candidate_A

Governed Condition:

Governed mode

Retained-state influence enabled

Selected result:

candidate_B

The selected behaviour changed under the governed retained-state condition.

The full live evidence path was:

Phase-2 retained state

Crown stored state

real P8a / Crown selection

durable private Decision Record

customer-safe P8b Decision Record

Phase-2 post-decision receipt

Evaluation Forge evidence bundle

The integrated system also passed controlled tests for:

restart persistence

deterministic replay

duplicate-request protection

ambiguous-outcome reconciliation

failure transparency

customer-safe evidence export

managed evaluation workflow

The current system can now run a bounded managed evaluation from retained-state preparation through real selection and customer-safe evidence production.

Scientific Value:

Phase 3 demonstrates live end-to-end behavioural divergence through the real CAAI selector under controlled retained-state and governance conditions.

Because the reference and governed runs differ in both governor mode and retained-state-influence configuration, this test does not isolate the independent causal contribution of either factor.

It is also not independent empirical proof of Verrell’s Law as a general physical law.

It is a live engineering result showing that retained prior state can participate in changing which permitted behaviour is selected while the reference path and resulting decision evidence remain inspectable.

Status: ✅ Live retained-state selection and customer-safe evidence chain complete.

Commercial Status:

Ready for controlled managed evaluations, paid pilot discussions and licensing conversations.

Not yet public SaaS, multi-tenant production infrastructure or a finished customer-hosted deployment.

Protected under Inappropriate Media Limited / Collapse Aware AI.
Proprietary Crown and selector internals remain private.

Information icon

We need your consent to load the translations

We use a third-party service to translate the website content that may collect data about your activity. Please review the details in the privacy policy and accept the service to view the translations.