Meteorites
Negative ControlSecond negative control. Meteorites don't trigger each other — and our method agrees.
Decision Chain: Meteorites
Ingest Meteoritical Bulletin meteorite fall records. Witnessed meteorite falls with recorded dates.
Meteorite fall records ingested.
Fit self-exciting point process. Meteorite falls are random events — one meteorite cannot trigger another.
Branching ratio 0.005. Near-zero self-excitation, as expected for random physical events.
Compare with the fireball negative control (era-corrected BR 0.007). Two independent datasets both show near-zero.
Meteorites (0.005) and fireballs (0.007) independently confirm: the method does not hallucinate self-excitation.
Second negative control: PASSED
Meteorite falls are genuinely random — one meteorite entering Earth's atmosphere cannot trigger another. The branching ratio of 0.005 is effectively zero, confirming that our Hawkes process does not produce false positive self-excitation. Combined with fireballs (0.007), we now have two independent negative controls. When we report high branching ratios for datasets like NUFORC (0.959) or BFRO (0.988), we can be confident that's real contagion, not a method artifact.