Two things to add, one correcting the framing of the proposed follow-up test, one pushing on where the approval boundary can actually do work.
On the proposed test. The paper already contains a close version of it. Irregular ran a planning probe of 160 plans and varied what the environment...