Answer
No, for this system. THEORY-275 (Active), from Das, Manikandan and Banerjee (LIT-947, read in NOTE-788): two colloids coupled through a symmetric hydrodynamic tensor, with recorded noise moving one trap. Time-delayed mutual information and transfer entropy run drive → driven → other, with backflow at moderate drive, in experiment and in numerics. In numerics of the fitted model only, the flow reads one-way under strong drive. So a directional measure can report a hierarchy where the coupling is reciprocal.
The answer is a counterexample, which ADR-049 counts as an answer to a question of the form “must X?”. It is not a theorem about directional measures in general; that generalization is the record’s own claim, CLAIM-396 (Proposed), and its test in a second system is QUESTION-186.
What stays with the parent
QUESTION-063 asks what measurement or intervention tells shared drive, coupling among members and distinct internal routes apart. THEORY-275 removes one candidate for the observational case and says nothing of the rest: the drive here is recorded and on one unit, while QUESTION-063’s case is an unrecorded drive that may be shared. THEORY-129 and THEORY-130 keep their partial answers to the parent.
Why it is a split
THEORY-275 answered QUESTION-063 “in part, negatively”, as the parent’s
body already said. Under ADR-049 a partly answered question is split: the
answered part is filed as a child worded to exactly what the Active entry
settles, and the answering entry’s answers link moves to it. The owner
asked for the splits in the exchange of 2026-10-23 (U9: “can we split any
QUESTIONs into an answered part and an unanswered refinement?”).