Testing PL-27/PL-28 Order-Sensitivity Across Three Physical Domains: A Scope Limitation, Its Fix, and a Consistent Negative Result

PL-27 (block-shuffle null) and PL-28 (multiscale delay scan) were introduced in WCM-PL v2.1(2026-07-21) as domain-general tools for testing whether a sequential system’s geometricfingerprint depends on temporal order at a chosen scale, validated on classical-music spectralfeatures and human motion-capture data. This paper reports what happened when wedeliberately tried to apply these same tools, unmodified, to three physical domains where WCM-PL already has standing programs: protein self-assembly (α-synuclein DMD trajectories),earthquake catalogs (JMA, two independent M7+/M9 mainshocks), and epileptic-seizure EEG(CHB-MIT, 3 subjects / 5 seizures). We report two classes of finding. First, a structural scopelimitation: PL-27’s block-shuffle null is provably uninformative (exactly degenerate, confirmednumerically to 10 decimal places) when applied directly to data that is already a native multi-axispoint per timestep, because WCM-PL’s occupancy statistics ( Sensor.assign / shape_vec / cos_dist ) arepermutation-invariant by construction. We designed, implemented, and audited a fix (frame-concatenation embedding, native_concat_shuffle.py , plus a new arbitrary-dimensionCrossPolytopeSensor ) and confirmed it restores a genuine order-sensitive null. Second, havingconfirmed the diagnostic itself works correctly in all three domains, we report a consistentnegative result: no domain tested here shows a genuine, properly-controlled, order-dependentprecursor signal near a target event (mainshock, seizure onset, aggregation commitment).Multiple promising-looking leads appeared during the investigation and were each traced to aspecific, identifiable confound ‒ an outlier/foreshock-composition effect, a single-baseline-window bias, cross-normalization disagreement, and a tie-breaking artifact ‒ rather than genuinetemporal structure. We report these confounds explicitly because the debugging process is itselfthe main methodological contribution of this paper, in the tradition of this series’ priormethodological/negative-record contributions (Bridge Papers M, P, X Corrigendum). Finally, weoffer a bounded, explicitly speculative discussion (§7, not connection-audited) of what thispaper’s negative result, combined with Bridge Paper S’s music corpus and its one periodicity-linked exception, suggests about what distinguishes music’s temporal structure from the domainstested here. ※ The License for codes is AGPL-3.0-or-later.