popper
An autonomous loop that tests hypotheses against live astronomy archives, then tries to kill its own results.
1.12e-210the p-value of a mass result it killed anyway, as a selection effect
popper proposes a falsifiable hypothesis, tests it against public astronomy archives queried live over TAP/ADQL, and then runs a separate pass that tries to falsify the result before it is kept.
In a 6-cycle batch it kept 2 findings and killed 4. One of the four was a radial-velocity versus transit mass result at p = 1.12e-210, rejected as a selection effect: significance was not allowed to be the verdict.