V5 showed that structure beats optimization — the purple spine mattered more than adding buses. V6 asks a different question: if you give some buses an advantage (faster boarding, higher capacity), does that change the attractor shape, or just who leads the cluster?
Heterogeneous fleet: Large blue circles are Express buses — they board 2x faster and hold more passengers. Small grey circles are Struggling buses — they're slow. Standard buses are in between.
The hypothesis: Express buses don't escape bunching. They become the front of the cluster. The attractor is robust to agent quality. The only intervention that actually changes the shape is the dispatch rule — headway-based holding that operates above the individual agent level.
Toggle between modes to see if the hypothesis holds. If it does: agent quality is mostly noise. System architecture is signal.