Why do the real constraints show up after we've already committed?

You and your pod are good at discovery. You ask the right opening questions, you map the system, you nod at the plan. Then weeks later, in the middle of the build, something buried surfaces — a dependency no one named, a limit no one tested, a stakeholder no one asked. It costs you time, and it costs you the team's trust in your process. Here is the honest reason. Your discovery is built to confirm that a plan can work. It is not built to find what will make it fail. Those are different questions, asked differently, and most teams only ask the first one. The second question feels like doubt, so people avoid it — and call the avoiding "efficiency." You are working with others, which makes this worse, not better. Each person assumes someone else checked the risky part. No one owns the hunt. Fix the ownership, and the timing fixes itself.