The Rubicon Index measures distance from the point of no return. Falling = danger rising.
—
Why the index sits here
Four drivers converge toward a war-permissive configuration. Each bar is the driver's current level; the % is its weighted share of convergence.
The collapse gates
Perceptual collapse — the moment leadership stops seeing alternatives to force — requires both gates open at the same moment. A huge shock alone does nothing if the structure hasn't converged (that's why Taiwan 1996 stayed quiet in our blind tests).
Trajectory
Rubicon Index per assessment cycle. Shaded bands are alert tiers; dashed verticals are named shock events. A grey dashed opening segment is hindcast backfill — assessed retrospectively with knowledge of outcomes, shown for context and never scored into the track record. Click any point to inspect that cycle.
Forecasts — the track record
Tetlock-style accountability: every forecast is a resolvable question with a deadline, explicit resolution criteria, and an outside-view base rate. Every revision is logged below the moment it is made; resolved questions are Brier-scored against a base-rate-only benchmark — misses included.
Shock watch
Discrete events that hit the system — detected by the continuous news watch, corroborated across blocs before injection.
Sub-threshold activity
Observable actions below the threshold of war — the actor shaping the space.
Analyst log
What the AI analyst read and concluded, verbatim — every value on this page traces to an entry here. Guardrails: changes capped per cycle, low-confidence assessments discarded.
Get alerted when the index moves
Free weekly digest of every monitored region. Instant alerts on tier crossings, shocks and forecast revisions are coming with Pro — free while we build our public track record.