LAB / WMSI-EXP-003
Concept-Graph Recovery Across Output Lengths.
Estimate how conceptual structure changes as answer bandwidth changes.
01 / QUESTION
Representation, Output, and Internal State
Estimate how conceptual structure changes as answer bandwidth changes.
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
Recover node/edge graphs from outputs at several budgets; compare to source task requirements and each other.
Controls
Multiple independent graph extractors; manual spot audit; fixed ontology where possible.
What we would measure
Node recall, edge recall, causal depth, graph edit distance, contradiction edges.
Access and instruments
Ordinary software or API access. This describes access requirements, not verified readiness on our Spark.
NetworkX/igraph; extraction prompts; adjudication UI.
Before a full run
Feasibility=NOW; 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
Length leakage, judge bias, truncation artifacts, prompt wording.
What would weaken the idea
If graph structure remains stable across budgets except for low-value detail, output bandwidth is not the main constraint.
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
Prompt/input, token budget, raw text, token counts, extracted claims/concepts, judge outputs.
Planned visual evidence
Layered concept graphs; graph-edit-distance curve.
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