A Gaussian sky keeps its phases secret, and the CMB's phases have been caught talking. For a statistically Gaussian random field, the kind inflation delivers, the Fourier phases of the fluctuation pattern are independent and uniformly random: all cosmological information lives in the amplitudes (the power spectrum), and any structure in the phases, correlations between modes, clustering of phase angles, coherent phase relationships across scales, is forbidden information the field should not contain. Dedicated phase-statistics analyses, from Chiang and Coles's foundational work through Naselsky's WMAP studies and successors on Planck maps, have repeatedly found non-random phase behavior at the largest scales: phase correlations among low multipoles, phase coupling aligned with the known anomaly directions, and cross-scale phase coherence that survives foreground cleaning choices while remaining entangled with them, the chronic condition of large-angle CMB science.
The finding is the anomaly family in its most fundamental dress: the Axis of Evil, the hemispheric modulation, and the parity preference are all, at bottom, statements that the sky's phases are organized, since aligned multipoles and modulated power are what phase correlations look like when projected onto more familiar statistics. ΛCDM's position is uniform: phases are random, full stop, so every detected correlation must be foreground residue, masking artifact, or chance, defenses that must hold across every estimator family simultaneously, because the model has no parameter that could make phases speak.
The standing is the deepest version of the isotropy question: phase statistics are where Gaussianity and isotropy are tested together, the low-multipole sector remains unresolved between cosmology and contamination, and polarization phase statistics from LiteBIRD-class data will retest the correlations in a nearly independent channel.