From replication to a predictive test: the Lane–Seifert programme and its planned Strand 4, an out-of-sample transfer test on DES-Dovekie

Programme outline for an independent research programme on the supernova evidence for timescape cosmology (Lane et al. 2025; Seifert et al. 2025). The programme asks how strongly the supernova data support timescape and how far that statement depends on the analysis. It was designed from the outset in four stages: reproduce, compare, understand, test. Each stage builds on the completed result of the previous one. Stages 1 and 2 are completed; Stage 3 is registered. This record discloses the planned Stage 4: an out-of-sample transfer test on DES-Dovekie, the recalibrated five-year supernova sample of the Dark Energy Survey. It sets out the question and the planned components. These are a paired pipeline bridge, a prediction learned from Strand 3 and hash-timestamped before any DES test computation, and a calibration under known truth. It also states what the test can and cannot show. Stage 4 is planned and not yet registered; its concrete prediction is derived from the Strand 3 results. The detailed plan (German, version 1.5.2) is not published here. It is bound by the SHA-256 hashes in the hash-commitment file and will be disclosed unchanged.