This paper introduces and analyzes the Observer Locus Instability Problem, a structural constraint on how observers may move, reindex, or relocate perspective within a lawful closure-governed system. Rather than treating observers as freely mobile viewpoints or abstract labels, the paper formalizes observer motion as a Ψ-trajectory subject to admissibility conditions. An observer locus is shown to be viable only if its motion preserves global invariants, remains within class-dependent kinetic bounds, and does not violate anchor consistency under closure. When these conditions fail, attempted observer motion becomes inadmissible, leading not to new branches or worlds, but to instability or collapse of the observer locus itself. The analysis distinguishes Ψ-instability from familiar mechanisms such as measurement collapse, decoherence, or Everettian branching. Ψ-motion does not generate new records or outcomes; it reindexes perspective within a fixed closure structure. As a result, not all imaginable observer trajectories are physically or structurally allowed, even in principle. The paper provides: a formal definition of Ψ-admissibility, explicit criteria for observer locus stability, a structural explanation for forbidden observer paths, and a clarification of how observer identity persists or fails under global closure constraints. These results place fundamental limits on observer mobility, perspective-shifting, and identity traversal, with implications for cosmology, quantum foundations, consciousness studies, and the design of advanced AI systems. The work strengthens the case that observer behavior is governed not by epistemic choice or metaphysical freedom, but by law-level admissibility and closure consistency.