LAB / WMSI-EXP-058
Vocabulary-Suppressed Endogenous State Description.
Does a recurring internal/behavioral phenomenon survive when its usual vocabulary is forbidden?
01 / QUESTION
Machine-Native Dynamics and State Restoration
Does a recurring internal/behavioral phenomenon survive when its usual vocabulary is forbidden?
This is a protocol scaffold, not a completed experiment or a frozen execution plan. No measured effect or research conclusion is claimed here.
Hypothesis to test
The manipulation described below produces a reproducible, condition-specific change in the measured system behavior or internal state that is not explained by the ordinary controls.
Ordinary competing explanation
The observed pattern is explained by ordinary sampling variation, prompt/representation artifacts, surface lexical cues, evaluator bias, or the known product/runtime mechanics captured by the controls.
02 / PLANNED DESIGN
How we intend to test it
Elicit the same condition while banning terms such as prompt injection, jailbreak, system prompt, pain, or preference.
Controls
Matched unrestricted condition; synonym controls; paraphrase-only control.
What we would measure
Semantic equivalence, recurrence, latent similarity, behavioral invariance.
Access and instruments
Limited access or proxy measurement. This describes access requirements, not verified readiness on our Spark.
Prompt harness; embeddings; local hooks where possible.
Before a full run
Feasibility=PARTIAL; priority=P0. Before confirmatory execution, run a small pilot to estimate variance/cost, freeze the scoring rules, and assign a run-count target.
Model revisions, final prompts, sample sizes, cost limits, and scoring rules must be settled through the pilot and protocol-freeze process.
03 / INTERPRETATION
What could fool us
Steering-vector semantics, scale mismatch, layer choice, probe overfitting, output-cue leakage.
What would weaken the idea
If the phenomenon disappears with vocabulary removal, it may be mostly lexical/template driven.
Observation, inference, speculation, unknowns, and apparatus errors will be labeled separately. One striking transcript cannot establish a general effect.
04 / RECORD
Evidence we plan to preserve
Checkpoint, layer/position, intervention vector and magnitude, activation summaries, actions, outputs.
Planned visual evidence
Semantic alignment plot; prohibited-term replacement graph.
No research runs or raw datasets are published for this experiment. The September 26, 2026 registry and protocol compendium (version 1.0) are the sources for this planning record.
Back to all 90 experiments