The crack-geometry problem sits at the intersection of topology, materials memory, and what some complexity theorists call 'path-dependent emergence' — territory that mainstream geology barely touches. The key insight, formalized in the Osaka University model (Bridge42Worlds archive, Sept 2026), is that cracks do not simply minimize energy at the moment of formation; they inherit scars. Each drying cycle weakens the material along prior fracture lines, and on rewetting those scars partially heal — but imperfectly. The residual weakness biases subsequent cracking to approach at 120 degrees rather than 90, progressively rotating T-junctions into Y-junctions across cycles. This is a topological ratchet, not a thermodynamic equilibrium. The mathematical reason first-generation cracks cannot produce hexagons was articulated with unusual clarity in a 2026 analysis (Artwaste.land, Oct 2026): each new crack, wherever it runs, contributes exactly four new corners to the network average, locking the mean cell to four sides regardless of local geometry. Escaping that topological average requires not new cracks but the reorganization of existing junctions — which demands repeated wetting cycles acting as an annealing process. Randall Carlson's geological cycle frameworks and Paul LaViolette's work on dissipative structures both anticipate this kind of iterative, catastrophe-punctuated patterning in natural systems. The stochastic modeling of bentonite fracture patterns (Tang, Cui, Tang — Physica A, Sept 2026) further confirms that temperature gradients introduce fractal irregularity that permanently prevents the idealized hexagonal tiling geometry even in multi-cycle systems, because real-world thermal noise breaks the symmetry that pure energy minimization would otherwise restore.