Another question on the lossy nature of observation. Re-stating the final paragraph of section 1.1, from an information-theoretic perspective:

1. Let the initial system \\(X\\) have information content \\(Ix\\)
2. Subject \\(X\\) to a lossy observation \\(f\\) producing system \\(Y\\) with information content \\(Iy\\\)
3. By definition, \\(Iy < Ix\\)

Assuming that's valid, is it reasonable to interpret generative effects as *(re)discovery*? In other words: we are re-discovering latent properties of \\(X\\) that were elided by \\(f\\). An important consequence would seem, therefore, that the information resulting from those generative effects is bounded, pre-determined, and equivalent to \\(Ix - Iy\\).