The original code assumed dyadic-sob>? is negation of dyadic-sob<=?, but

it doesn't hold when two sobs are neither subset of each other.