Waterfall

The useful inheritance is commitment-sensitive assurance—not a fixed one-pass phase diagram.

Caricature and history

Royce used a staged sequence as a risky problem setup, then added feedback, preliminary design, a pilot, early testing, documentation, and customer involvement. “Royce prescribed one pass” is therefore inaccurate. The opposite claim—that all Waterfall criticism attacks only a straw man—is also inaccurate: contracts, approvals, organisational handoffs, and review structures produced real long-latency sequential practice.

Software V/V charts, the systems-engineering Vee, and Germany's V-Modell/V-Modell XT are distinct decision-relevant families. A shared V-shaped drawing does not establish common ancestry, fixed ordering, or deferred testing.

Selected sources and claim limits

Source	Supports	Does not establish
Benington, Production of Large Computer Programs	Large-system specifications, interfaces, testing, and records before Royce.	Controlled proof for retrospective prototype and cost claims.
Naur and Randell, 1968 NATO report	Pre-Royce feedback, simulation, prototypes, and acceptance-evidence discussions.	One prescribed method; the report records heterogeneous views.
Royce, Managing the Development of Large Software Systems	The bare sequence was called risky and supplemented with corrective steps.	Comparative causal evidence; the proposal remained strongly staged.
Bell and Thayer, Software Requirements: Are They Really a Problem?	An early published “waterfall” naming hinge and persistent requirements change.	The complete textbook transmission chain or absolute one-pass endorsement.
Boehm, 1979 V&V guide	An early software V chart pairing definition products with evaluation and feedback.	That all V-shaped systems share one genealogy.
ISO/IEC/IEEE 15288:2023 and 12207:2026	Lifecycle processes and outcomes that can be tailored and applied iteratively.	Universal causal benefit or automatically non-bureaucratic use.
NASA, Systems Engineering Handbook	Recursive systems engineering, interfaces, V&V, and technical management.	A universal consequence profile.
NIST, SSDF SP 800-218	Security obligations integrated into varied lifecycle models.	Validation of Waterfall, Agile, or one lifecycle.

Surviving assurance properties

The surviving state is authoritative but revisable intent, assumptions, configuration identity, interfaces, and change provenance.
Working evidence that retires uncertainty before hard commitment.
Evidence includes multi-level verification, intended-use validation, qualification boundaries, acceptance authority, and residual-risk custody.
Evidence depth selected by consequence, coupling, observability latency, external authority, recovery difficulty, and test scarcity.
Traceability only where a decision or consequence needs reconstruction—not a universal manual trace matrix.

Rejected caricatures and ceremonies

Fixed one-pass sequencing, complete up-front requirements, document-completion gates, fixed review meetings, universal central change boards, exhaustive manual trace matrices, and a universal exponential late-change cost curve are rejected or not adopted.

Documentation, configuration, traceability, independent challenge, and review survive only at the lowest-cost form that truthfully protects a real consequence. A report, board, baseline label, or model tool is not the property.

Native route

IMPLEMENTAUDIT expresses these properties through owner/source, exact identities, evidence plans, phase acceptance, independent review, rollback, publication authority, and final audit. The Evolved-LAW synthesis reconciles assurance depth with reversibility and cheap exploration. No Waterfall mode or fixed lifecycle was added.