Methane 3D Satellite Tracker
Passed public review. v2.1.0 marks two Java coast plumes as threshold-sensitive and puts the attribution caveat beside every plume.
Three things happened in September that we want on the record: two products finished public review, one certificate was revoked after a public reviewer found a flaw, and the November cohorts opened. Everything below links to its evidence.
Passed public review. v2.1.0 marks two Java coast plumes as threshold-sensitive and puts the attribution caveat beside every plume.
The intrusion-detection pipeline is Verified. Rules, thresholds and the replayable test traffic are public; anyone can rerun the evaluation.
The method behind the tracker, with every data source, the thresholds and where it fails.
Moved to community review. Reviewers are reading the oversight policy and the model card.
Opened for public review. Every run log is published; rerun any task and comment.
Submitted for Institute review, v0.3.1.
AI-CRT-0002 was revoked on 2026-09-09 after a public reviewer showed test documents had leaked into training. The record stays public, with the reason. This is the system working, and we would rather show it than hide it.
A community reviewer asked us to replace the mean background in the methane flagging step with a median. It removed most of the false plumes near large sources. The change and the review that asked for it are in the changelog.
Ten weeks, ending in a working intrusion-detection pipeline reviewed in public.
See the course →Pass the field check by reviewing two past proofs against the rubric.
Become a reviewer →D. Santoso
Editorial Board · Y. Hakim
Y. Hakim · B. Siregar
Newsletter
Monthly. New proofs, review results, and course openings. No spam.