Existence vs reachability
A valid endpoint can still exist after the current route has already died. That separation is the operational problem the diagnostics are built to measure.
This is the plainest statement of the distinction: the goal can stay real after the current route is gone.
The map can stay alive after the route dies.
Existence asks whether a valid endpoint is still out there somewhere. Reachability asks whether the current process can still get from its committed state to one of those endpoints. A system can satisfy the first condition while already failing the second.
This is why many postmortems feel unsatisfying. People keep saying the goal was still possible. That can be true and still miss the operational reality. If the route died first, “the goal still existed” is the wrong level of description.
The distinction matters most in forward, commitment-heavy work: search, scheduling, routing, optimization, planning, structured decoding, or any process where going backward is costly or impossible.
Plain-language test
If a team can still truthfully say “a valid answer exists” while also truthfully saying “our current process can no longer reach it from here,” you are looking at the existence–reachability split.
Three operational patterns
The route dies early, the symptom appears late.
A process can keep moving after its viable future is already gone. Local scores still look acceptable until the final trap becomes visible.
Healthy local metrics can still mislead.
A state can look locally fine while the remaining completion set has already narrowed past recovery. That is why local “health” and future viability separate.
Stronger methods do not just score better.
They can move the boundary itself. The real question is what they buy: more viable runway, fewer shallow traps, less opacity, or some combination.
For research
The split creates a measurable object beyond endpoint existence. It lets you study when process capability diverges from landscape capability, not only whether solutions exist.
For engineering
The split turns vague failure into a concrete diagnostic question: where did the route close, how deep is the trap, and what change would actually move the boundary?