Explaining Why Teams Beat Sequential Investigation

A colleague says: "I don't see why agent teams are worth the extra tokens for debugging. If I have five hypotheses for a bug, I could just have one Claude Code session, or a series of sequential subagents, work through all five in order — cheaper, and it gets to the same five answers eventually."

1. Explain precisely what anchoring is in this context, and why a single agent working through five hypotheses in order is structurally vulnerable to it in a way that a team of five parallel, adversarial teammates is not.
2. Your colleague's claim that sequential investigation "gets to the same five answers eventually" assumes investigating each hypothesis produces the same finding regardless of what order it's investigated in. Explain why that assumption is exactly what anchoring breaks, using a concrete example.
3. Acknowledge the real cost tradeoff honestly: under what circumstances would your colleague actually be right that sequential investigation (or even a single hypothesis-free investigation) is the better choice?