Reported evidence
Shared-endpoint audit → measures → Repeated measurement
Within this audit, both same-window ranking repetition and next-day replay fell below their preregistered thresholds. The second statistic's full name is unavailable in the visible excerpt, so it must not be treated as the same measure as Spearman correlation. The findings do not identify the mechanism responsible.
Read the evidence
A preregistered study recorded, across 52,988 audited request attempts, a Spearman value of 0.400 for agreement between repeated rankings within the same time window, below the preset 0.90; the agreement value for next-day replays of byte-identical requests was 0.78, also below the preset 0.99 [S6]. Within the scope of this shared-endpoint audit, neither same-window repeated rankings nor next-day replay met the study's preset repeatability standards.Read the full research