Objective – to demonstrate that entropy is not an absolute property of a physical system but depends on the choice of description level G (degree of coarse‑graining). Methods. The study is based on classical works by Gibbs (fine‑grained vs. coarse‑grained entropy), Jaynes (maximum entropy principle), and von Neumann (observer entropy), as well as on the contemporary theory of observational entropy (Šafránek, Aguirre, Deutsch, 2020). The parameter G is formalized by partitioning phase space into cells of varying sizes. Results. The formula S = f (, G) is introduced, where is the number of available microstates and G is the choice of description level. It is shown that the same physical system (e. g. , a gas in a container, a black hole) can have different entropy values under different G. The quantum limit (Heisenberg uncertainty principle) sets a lower but not an upper bound for G. Conclusions. Recognizing G does not invalidate the second law of thermodynamics but makes its application more conscious. Entropy is not an absolute characteristic but a quantity defined relative to a chosen coarse‑graining. This approach resolves pseudo‑paradoxes such as the “black hole information paradox” by showing that different entropy values correspond to different legitimate choices of description level. The choice of G is not arbitrary: it is determined by the research task, available information, and instrumental capabilities. Keywords: entropy, choice of description level, coarse‑graining, G, observational entropy.
Alexander Yourievitch Kotelnikov (Sat,) studied this question.