What Seismikon Currently Claims
A precise statement of operational claims, how each is measured, and which remain aspirational.
What Is Claimed
Seismikon makes the following operational claims about its live prediction service:
- Prospective prediction: the current S.A.F. issues forward 5-tuple predictive outputs specifying longitude, latitude, focal depth, magnitude, and time for significant California earthquakes up to 48 hours in advance.
- Defined evaluation protocol: every prediction is scored against a published, versioned, pre-fixed set of tolerances - location, magnitude, depth, time - so performance is independently verifiable.
- Skill above chance: the prediction record is evaluated against an empirical random baseline; the claimed skill score is published on the Track Record page and is computed client-side so it can be verified by any user.
- Retrospective research: Bikos (2024) reports ≥98% across approximate latitude, longitude, focal-depth, and magnitude parameters, with occurrence time-frame estimation also investigated. This is retrospective research, not live performance.
- Full transparency: every forward-issued prediction - including false positives and false negatives - is publicly published. No results are suppressed.
What Is Not Claimed
Seismikon explicitly does not claim:
- Certainty: no individual prediction is guaranteed to be correct. Earthquake prediction is probabilistic.
- Real-time safety alerts: Seismikon is not an emergency warning system and should not be treated as one.
- Global coverage: the current live service covers California only.
- Deterministic precision: predictions are evaluated against tolerances, not point matches.
- Publication-grade prospective validation: the prospective record is still accumulating. Large-sample statistical significance has not yet been established across all metrics.
How Claims Are Measured
Each claim is tied to a measurable, publicly available metric:
- Skill above chance → empirical skill score on Track Record page (updated as results accumulate).
- Prospective prediction → public prediction feed with timestamped issuance before event window.
- Defined protocol → published Evaluation Protocol with version history.
- Retrospective validity → Bikos (2024) peer-reviewed paper, downloadable from this site.
- Full transparency → Data Downloads page with complete prediction and outcome records.
Aspirational Claims (Not Yet Operational)
The following represent development goals that are not yet operational claims:
- Multi-region coverage beyond California.
- Independently audited prospective performance by a third-party scientific body.
- Formal Molchan diagram or CSEP (Collaboratory for the Study of Earthquake Predictability) submission.