Three different failures can share a shape.
An immune response loses its brake, and defense begins injuring what it protects. A market discovers that alarm is profitable, and warnings feed the acceleration they were meant to slow. Memory is not a file stored apart from the body; when the three-dimensional organization supporting it unspools, access and continuity can fail.
These are not one mechanism. Together, they reveal the same design question: what remains safe when a system can no longer stop itself?
Imagine the right to stop as infrastructure: a room, protocol, object, or relationship whose pause cannot be bought, overridden, or made conditional on recovery. It would not demand proof that someone will return to their former shape. It would hold a small area steady while the larger system remains unstable.
At human scale, this might mean a paper plan, stored water, a familiar object, or a practiced boundary that ends incoming demands. At institutional scale, it might mean an independent pause and a fallback that does not depend on the failing system.
This is a proposal, not a proven law. Its test is concrete: when alarm rises, does the anchor reduce dependence, preserve choice, and allow rest without requiring performance?
When the map dissolves, keep one piece of ground that can still say: stop.