A direct US-China military incident in the western Pacific
GEO-02 · probability 38% (confidence 62%, ±15 pts) over 2026-2031 · horizon NEXT 3-5Y · domain geopolitics. Probabilistic simulation, not advice.
The reading
Not full war — a collision, a shoot-down, a blockade probe that produces casualties and forces both capitals to the brink before de-escalating. Allison found that 12 of 16 rising-vs-ruling power transitions ended in war; the modern tripwires are Taiwan (which fabricates ~90% of leading-edge chips) and the Second Thomas Shoal.
What would prove this wrong
This projection is WRONG if through 2031 there is no US-China military contact producing casualties AND PLA air incursions into Taiwan's ADIZ fall back below their 2021 tempo for a sustained year.
Trigger events tracked
- A PLA "quarantine" or customs-inspection blockade of Taiwan framed as law enforcement
- A collision between Chinese coast guard and Philippine or US vessels at Second Thomas Shoal
- US naval escort of Taiwanese shipping, crossing a declared Chinese red line
- Large-scale PLA exercises (Strait Thunder / Joint Sword series) transitioning to no-notice posture
Causal chain
Historical precedents
If it happens
Field notes
2026-08-28: Adjudication clarification: the entry is graded on whether the stated event occurred by the resolve date — the falsifiability line is the pre-registered strong-refutation test, not a second event definition. If no qualifying incident occurs by 2031, this scores as NOT OCCURRED even if the incursion-tempo clause is unmet. The sealed probability is unchanged.
Sources
Subjects this belongs to
More on this desk
Browse the War & Geopolitics desk, the full projection registry, the calibration ledger, or the record by subject. Scoring is explained in how a prediction is made.