The SESOI stays at 0.01, justified by cost-benefit, and stated as what sizes K¶

Audit item 7. Current guidance on justifying a smallest effect of interest (Anvari and Lakens, JESP 2021; Lakens, Collabra 2022) prefers a practical or cost-benefit argument, anchor or consensus methods, and advises against taking the value from effects earlier studies reported. The seed-noise origin recorded on 2026-09-22 is withdrawn as a justification.

Justification. An organization that adopts the result trains and governs an adapter on its own review history. The smallest gain taken to repay that is one more exactly correct refinement per hundred review comments: 0.01 in exact match. For context only, not as the anchor: the gains the code-refinement literature reports as contributions are larger (CodeReviewer over CodeT5, +5.9 points on its benchmark, Li et al. FSE 2022), so 0.01 is the conservative end of what the field treats as worth reporting, and a larger value would make K smaller (K scales with 1/xi^2, and at 0.02 OpenStack's K falls to the floor of 10).

What it decides. Through XI = SESOI, it sets K, the number of runs at which two independent draws of partitions agree within it with probability about 0.95. Beside each cell it is reported and decides no verdict. The earlier comment that it "decides nothing" was wrong and is corrected in decomposition.py; the manuscript states that it sizes K.