A test suite that passes on typical inputs can miss bugs that a handful of edge cases would catch immediately. Why are the edges so much more productive to test than the middle?

B. Because bugs cluster where behaviour changes, while typical inputs share one path.

Ten typical inputs usually walk the same route, so the tenth tells you almost nothing the first did not, where each edge probes somewhere different: empty tests the zero-iteration path, one element tests initialisation, the maximum tests overflow. Property-based tools generate inputs systematically and find edges humans do not think of, which says something about how far intuition gets you.