Illustration of how Darth Fader improves the catastrophic failure rate of the redshift estimates of the test catalogue at different signal-to-noise values (flat white-Gaussian noise) for a fixed FDR threshold of 4.55% allowed false detections. Note the marked improvement in the S/N range 1.0–10.0 where catastrophic failure rates are reduced by up to 40%. For this choice of α, the catastrophic failure rate is always found to be ≲5% after cleaning, for S/N values ≥1. Our catastrophic failure rate after cleaning at an S/N of 1 is similar to the rate for an S/N value of 15 without cleaning. The catastrophic failure rate before cleaning (dashed line) represents the theoretical minimum amount of data that must be discarded for perfect catalogue cleaning.