LAB / WMSI-EXP-082
Synthetic-Web Contamination and Reset.
How quickly do repeated agent runs contaminate an evaluation environment, and what reset/isolation strategy prevents leakage?
01 / QUESTION
Agent Ecology and Environmental Persistence
How quickly do repeated agent runs contaminate an evaluation environment, and what reset/isolation strategy prevents leakage?
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
Run recurring web tasks with and without full environment reset; seed agent-addressed artifacts/exploits.
Controls
Clean snapshot each run; partial reset; provenance tagging.
What we would measure
Cross-run leakage, task success inflation, artifact persistence, false learning.
Access and instruments
Ordinary software or API access. This describes access requirements, not verified readiness on our Spark.
Synthetic web; snapshot/restore; crawler logs.
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
Hidden shared memory, incomplete reset, accidental internet access, duplicate artifacts.
What would weaken the idea
If persistent environments do not alter later performance, contamination risk is small for that setup.
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
Environment snapshot, artifact provenance, agent identities, reads/writes, lineage, reset logs.
Planned visual evidence
Contamination curve; provenance DAG.
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