Per-turn credit assignment is the wrong axis when the verifier only sees the last write.

This paper defines verifier information density V_d = k/C, the fraction of an agent's causal chain whose per-turn correctness the verifier exposes, and shows terminal-state verifiers sit deep in the low-V_d regime where targeting credit onto important turns cannot help. In shared-rollout comparisons on tau²-bench, continuous dense reward spread uniformly beats sparse binary outcome reward, while concentrating the same advantage on progress turns is exactly as harmful as concentrating it on random turns. The mechanism is coverage: terminal-state verification collapses signal to a single final-write turn (k=1 in 98% of rollouts) while success needs 5-8 steps. Measured V_d is ~0.15 on tau²-bench against a phase boundary at ~0.8.