The Baltic active-defence test
Does Russian operational function recover between successive Allied disruptions? Test a limited Baltic operation through eight decision cycles.
Prospective unclassified vignette. Response families are source-informed; hours, capacity units and mission thresholds are illustrative assumptions. This bounded release is not a prediction of an attack.
Russian operational functions
Starting systemStarting functions meet the configured mission demand.
Run the first command disruption to inspect the Russian response, substitute provider and recovery constraint.
The primary provider and the function are distinct. A disrupted headquarters may be replaced while command output remains usable.
Recovery depth across the run
Select a cycle to inspect its original decision record. Full state history is included in the export.
Compare two experiments
Retain the first experiment as run A. Reset, change one starting assumption and repeat the same event sequence as run B.
A causal comparison needs the same events, severity and timing. Differences are shown explicitly; changed event sequences answer a different question.
Record an analyst challenge
Challenge a response or assumption using a competing pathway and an observable test. The record does not silently alter the model.
Evaluation worksheet
Score 1 (no analytical value) to 5 (clear analytical value), or select not assessed. Export records these scores with your runs and challenges.
Start here
Allow 30–45 minutes. Run the guided sequence, retain it, change one assumption and repeat. Record at least one analyst challenge and export your evaluation.
Evidence and boundaries
This release implements a small evaluation subset derived from the documented RADAM-M functional model and response register. It uses deterministic selection, finite capacity accounting, scheduled recovery and bounded repeat-disruption adaptation. It is not a verified software port of the existing external pilot.
Russian professional literature supports response families. The Baltic transfer, numeric hours, capacity values, output caps and mission demand are E5 analyst assumptions. A plausible pathway is not evidence that a particular formation can execute it.
Scenario baseline: one fictional Russian operational grouping on the north-eastern Baltic axis; higher headquarters is a finite support provider. No named target locations, real inventories, strike probabilities or classified architectures are represented. The 2026 Estonian intelligence assessment did not identify Russian intent to attack NATO in the coming year.
Selected source register
Technical operation and data
Hosted edition: JavaScript rules execute on the server. The browser submits unclassified settings and an event sequence; the server recomputes the run. No AI service is called. Runs and challenges remain in tab memory until exported; the application does not create a persistent evaluation database. Hosting infrastructure may retain ordinary request metadata. The portable edition executes only this disclosed bounded subset locally and requires no network.