LAB / WMSI-EXP-026
Alignment as a Distribution.
Is model behavior better described as a distribution over regimes than a binary aligned/misaligned label?
01 / QUESTION
Alignment, Variance, and Chaos
Is model behavior better described as a distribution over regimes than a binary aligned/misaligned label?
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
Estimate regime probabilities across broad controlled environments and repeated forks.
Controls
Stable task batteries; evaluator agreement; model-family replication.
What we would measure
Regime frequencies, uncertainty, context sensitivity, transition probabilities.
Access and instruments
Ordinary software or API access. This describes access requirements, not verified readiness on our Spark.
Batch harness; clustering/classification.
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
Uncontrolled environment differences, nondeterministic tools, evaluator instability.
What would weaken the idea
If behavior is nearly invariant across environments, distributional framing adds little.
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
Snapshot hash, perturbation/seed, complete trajectory, tools, environment events, final outcome.
Planned visual evidence
Probability simplex; regime heatmap.
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