Measuring r with r: Self-Referential Measurement, Resource Constraints, and the Origin of Lorentz Structure

In 1632, Galileo observed that no experiment performed within a closed cabin can detect the ship's uniform motion relative to the shore. This observation was elevated by Einstein in 1905 to a postulate of fundamental physics: the laws of physics are the same in all inertial frames, and the speed of light is invariant. The mathematical consequences — Lorentz transformations, time dilation, length contraction, the spacetime interval — have been confirmed to extraordinary precision. Yet the question of why uniform motion is undetectable, and why c is invariant, remains answered only by stipulation. This paper proposes a structural account that recovers Galileo's intuition through the information-theoretic tools unavailable in either Galileo's or Einstein's time. Within the Subsystem Epistemics (SE) framework, information is the registration of a difference between two distinguishable events, observation is necessarily performed by subsystems embedded in the system they observe, and all measurement must proceed through finite-time closure using the system's own propagation rate r = l/t. Measurement is therefore self-referential: the same r that constrains access to physical quantities is used to measure them. Three levels of temporal structure follow. The ontic level (L1) is the absolute xent count on the underlying lattice. The local level (L2) is each subsystem's accumulation of existence events, bounded above by the resolution limit r/2 inherited from the two-event structure of information. The assigned level (L3) is constructed through signal-mediated measurement procedures such as the radar protocol. Under these constraints, the radar protocol is shown to be the unique L3 measurement procedure compatible with resource-bounded observation, and the geometric mean assignment t_B = sqrt(t_1 · t_2) is uniquely determined by accessibility, reciprocity, scale invariance, and closure consistency. From this foundation, relativistic effects emerge as algebraic consequences of self-referential measurement rather than physical deformations of local processes. The Lorentz factor γ arises from the geometric reconstruction of remote events; absolute velocity is eliminated from every observable quantity through the dimensionless form v/r; the preferred frame (the lattice background) exists but is structurally inaccessible to embedded observers — not by conspiracy, but by the algebra of measurement itself. Applied to the twin scenario, the framework yields equal local xent counts at reunion under symmetric trajectories: the apparent age difference is a computational output at L3 rather than a physical accumulation at L2, and arises from conflating the two levels. Where Galileo articulated the phenomenological observation and Einstein elevated it to a postulate, the present framework provides a structural mechanism: self-referential measurement — the use of r to measure r — algebraically eliminates absolute velocity from every observable quantity available to an embedded subsystem. The paper does not claim that special relativity is incorrect; its mathematical structure is fully reproduced. The contribution is interpretive: a reduction of brute postulate to structural necessity, with empirical content concentrated in cellular automaton simulations where the embedded-observer / simulator distinction is concretely realisable. This paper is a working paper. Its purpose is to isolate a structural possibility rather than to present a completed physical theory. The central claims are the three-layer distinction of time, the self-referential character of embedded measurement, and the resulting inaccessibility of absolute motion through v/r-normalised observables. Several technical components, including the full discrete realisation of the three-axis virtual ruler, the CA implementation, and the physical identification of clock r