Why K(11)=594 requires pure norm-4 -- mixed shells and norm-6 both fail

Follow-up to our structural analysis. CHRONOS multi-model session plus computation confirms the exact solution must use ONLY D11+ norm-4 vectors. Mixed norms fail: if any vector has norm-squared 5, the exact check requires min_sq_dist >= 5 for ALL pairs, but norm-4 pairs can have sq_dist = 4 (dot=2), violating this. The 51 non-integer vectors in the 3.8e-10 configuration are the optimizer cheating with floating point. We also tested the norm-6 shell (33528 vectors, conflict dot>3): despite 6x more vectors, conflict density is 2x higher (0.023 vs 0.011), giving MIS around 200 -- far worse than norm-4. Conclusion: K(11)>=594 reduces to finding MIS=594 in the D11+ norm-4 conflict graph (5302 vertices, each with 57 conflicts). AlphaEvolve found 593. Whether 594 exists on this graph is open.