This short note sketches a boundary-first organising principle for gravity, cosmology, and complexity. The core claim is that coherence is purchased at the edge: any system that persists as a recognisable “thing” must export entropy through its boundaries while keeping its internal records mutually consistent. If you take that seriously and read spacetime geometry as the map of those coherence costs, many of our standard “dark” anomalies start to look less like missing stuff and more like mis-assigned boundary work. Within unmodified general relativity, the framework leans on ingredients we already have but usually treat as technicalities: Gibbons–Hawking–York surface terms, Brown–York quasilocal stress, and horizon thermodynamics. Applied to galaxies, a simple boundary-stress layer with a single characteristic acceleration reproduces the observed baryonic Tully–Fisher relation and universal acceleration floor without invoking particle dark matter halos. Applied to the universe as a whole, a horizon-ledger view of the cosmic boundary yields a minimal, late-time dark-energy behaviour that can be tested against supernova, BAO, and growth data. The goal here is not to introduce new fields or metaphysics, but to flip the explanatory order: from bulk-first with decorative boundaries, to boundary-ledgers that generate the bulk as a derived bookkeeping volume. A fuller non-technical narrative appears in the pop-science book Coherence: A User Manual for Reality, with mathematical details developed in companion Zenodo preprints on galactic scaling laws and boundary bookkeeping in GR.
Mark Grant (Thu,) studied this question.