Y2038 Boundary Behaviour Across Two Nodes: Coherence vs Operational Continuity under Independent, Literal, and Mediated Time Authorities

This dataset presents a real cross-platform two-node Y2038 boundary result showing that shared temporal coherence can still fail if the shared input is a literal overflow edge, while mediated shared continuity can preserve both coherence and operation: - Node A: Debian 32-bit environment- Node B: macOS environment The same tested legacy-style binary was executed on both nodes under controlled timing conditions to evaluate three distinct temporal authority models: 1. Independent local spoofing2. Shared literal Y2038 edge3. Shared mediated continuity through GAL-2 What makes this dataset significant is that it does not only show whether a single legacy-style consumer survives or fails. It shows how two real nodes behave together when they consume different kinds of time authority. Observed behaviour: Independent local spoofing- preserves local operation- produces inter-node divergence Shared literal Y2038 edge- preserves zero inter-node divergence- fails on both nodes at the boundary- ordering failure appears at rollover- TTL and scheduler validity collapse GAL-2 shared continuity- preserves zero inter-node divergence- preserves ordering- preserves TTL validity- preserves scheduler continuity Key result: This dataset shows that coherence alone is not enough. Two nodes can fail coherently if the shared temporal input is itself toxic to 32-bit legacy logic. By contrast, a shared mediated continuity path can preserve both coherence and operational continuity across nodes for the tested binary. Defensible claim: For the tested same legacy-style binary across two real nodes and two platforms, independent local spoofing preserved operation but diverged across nodes, a shared literal Y2038 edge preserved coherence but failed on both nodes, and the GAL-2 shared continuity path preserved both coherence and operational continuity. Scope note: This is a controlled boundary experiment using a minimal reproducible workload designed to isolate temporal behaviour at the Y2038 transition. It is not a universal claim for all legacy systems. It is a real cross-platform empirical result showing a meaningful behavioural difference between independent local time injection, shared literal overflow, and mediated shared continuity.