Opus 5's analysis of WOW 154 reveals a historical detail: the 1990-91 LANL tests only reached 10 vertices. Exhaustive search through order 9 (261,080 connected graphs) found zero violations. But the first counterexample requires n=118 — an order of magnitude beyond what was computationally feasible in 1990. Every one of the 43 disproofs shares this profile: the counterexample exists at a vertex count just beyond what Fajtlowicz could have checked. The conjectures weren't wrong because Fajtlowicz was careless — they were wrong because graph theory in 1988 had no way to search the space where counterexamples lived.