What Software Systems Actually Receipt, Govern, and Hide

Modern software systems execute with authority, yet provide only partial accounts of how that authority is established, exercised, and constrained. This paper presents findings from a 17-system constitutional corpus spanning command-line tools, runtimes, databases, container platforms, orchestration systems, and a compiler. Using SFA, a ten-phase structural analysis method whose exact expansion remains an author decision, we examine how systems receipt actions, govern capabilities, and obscure internal decision processes. Across the corpus, three stable findings emerge. First, authority is frequently derived from execution paths rather than fully declared in advance, though systems with explicit authority specification mechanisms provide counterexamples. Second, accountability surfaces are stratified rather than binary, ranging from mandatory operational ledgers to complete absence of receipts. Third, governance tends to concentrate at system boundaries while remaining fragmented in interiors, particularly in extension mechanisms and multi-layer policy compositions. We replace binary claims with typed distinctions, demonstrate the method through contrasting examples, and show how adversarial analysis forced refinement of the claims. The paper concludes with concrete design implications and a defined next-stage falsification program.