Lehman’s Software Classification

Lehman’s Classification of Software (S-/P-/E-type) and Laws of Software Evolution

SPE Classification, Lehman’s SPE Taxonomy

Software whose problem can be fully and formally specified. Correctness is provable against the specification. The world outside the program is irrelevant to its validity. Example: a sorting algorithm, a compiler for a formally defined language.

Software that addresses a real-world problem which cannot be fully specified, only approximated. The specification is a model of the real problem; the solution must be validated against the actual problem, not just the spec. Example: a chess-playing program, a weather forecasting system.

Software that becomes part of the real world and changes it through its use. This use in turn changes the environment and thereby the requirements — creating a feedback loop. Most business and organizational software is E-type. A "correct" E-type system drifts out of correctness over time even if nothing in it changes.

Lehman derived eight laws from observing E-type systems, including Continuing Change (E-type systems must be continually adapted or become progressively less satisfactory), Increasing Complexity (complexity increases unless work is done to reduce it), Self-Regulation, Conservation of Organisational Stability, Conservation of Familiarity, Continuing Growth, Declining Quality, and Feedback System.

Meir M. Lehman ("Programs, life cycles, and laws of software evolution", Proceedings of the IEEE, 1980); later developed with Laszlo Belady ("Program Evolution: Processes of Software Change", 1985).