Empty Is Not a State

In his latest article, Marco Somma argues that developers often make a critical error by collapsing various system outcomes—such as failures, outages, or genuine absence of data—into a single 'None' or empty value. This practice obscures the distinction between different failure domains, leading to poor control-flow decisions. Somma proposes a more robust architecture that separates observation from adjudication. By recording detailed metadata about every request and comparing it against historical baselines, developers can create an explicit taxonomy of outcomes. This allows systems to distinguish between temporary network issues, contract failures, and legitimate empty states. Ultimately, the author emphasizes that 'empty' is not a state in itself, but an observation that requires context. By implementing a deterministic policy layer, developers can ensure that state transitions are based on evidence rather than ambiguous defaults, preventing systems from incorrectly updating their internal logic based on incomplete or misleading data.