What the Monte Carlo Tested and What It Cannot Test: A Methodological Commentary on the Independent Verification of Paleopole Orientation Clustering

This commentary is a methodological response to the pre-registered independent verification of paleopole orientation clustering in the Buildreps database. The verification used Monte Carlo permutation as its primary statistical instrument and reached a contradictory conclusion: it initially confirmed the clustering at two proposed pole latitudes as a real feature of the data, then introduced post-hoc controls that eliminated the statistical significance of that same clustering. This commentary addresses three fundamental problems. First, the Monte Carlo procedure is structurally incapable of testing the chronological sequence predicted by the model, because it mixes structures from different pole periods into a single permutation pool and destroys the temporal order by construction. Second, the test demonstrated its own insensitivity by failing to detect the current geographic pole (90°N), a known and verifiable signal that the geometric method successfully identifies using randomly selected modern buildings. Third, the verification abandoned its own pre-registered protocol when the initial results were favorable to the model, and used unregistered post-hoc controls to reverse those results. The central finding is that the initial, pre-registered Monte Carlo test did not produce a null result. It produced a highly significant result at the exact latitude predicted by the hemispheric intersection. The test confirmed the model's central geographical prediction. The negative conclusion was manufactured afterward, by abandoning the protocol and tightening the null models until the signal disappeared. The model predicts a sequence. The test cannot see a sequence. The confirmation of clustering is therefore a partial result, consistent with the model but not a test of it. A proper test of the sequence requires treating the geometry and the Antarctic ice core record as the primary evidence, and a null model that preserves the temporal structure rather than destroying it.