Chapter 16. The Properness Table

Every audit in this part ended the same way: a column. This chapter assembles them. No cell below is new; the adjacency is, and adjacency is what a ledger is for. Each column header cites the chapter that scores it. The derived column cites two: Chapter 8, which proved it, and Chapter 15, which audited it. Its cells add a parenthetical citing the scoring result. The rows are the seven derived properties: R1–R3 on state, S1–S4 on architecture. is satisfied, ~ partial, failed. In the online edition every cell of the derived column links to its proposition.

Property Relational (13) OOP/ORM (13) XML stack (10) JSON/REST (10) SPA/JS (11) Wasm (12) GraphQL (14) Derived stack (8, 15)
R1 ~ ✓ (5.2, 5.4)
R2 (5.1)
R3 ~ (5.3)
S1 ~ ~ ~ (4.3)
S2 ~ (8.3)
S3 ~ ~ ~ ~ (8.3)
S4 ~ ~ ✓ (4.3, 8.3)

Notes, one per line where a cell needs it.

Read the table by row rather than by column — property by property instead of paradigm by paradigm — and the audit’s finding shows its shape. R2 fails in every column but one: composition without coordination is the property nobody else has, and it is what the “world wide” in the name promises. R2, R3, and S4 — the machine-spanning properties — fail or fall partial in every column but one. R3 and S4 are the two that, as Chapter 14 put it, make an architecture the web rather than an app platform that happens to use browsers. Chapter 13 found the highest pre-web score failing exactly the machine-spanning three; Chapter 14 found the convergence stalled before R3 and S4, the two still ahead. And the S-row failures kept appearing in this part twice: once as a score, once as a compensating industry. The industry is the bridge built across the gap the row names, at the boundary where a paradigm meets the web. The industries are the measurement; the table is the ledger.

One column has no failures, and Part II proved that shape was forced: that is the book, in one exhibit. Part V builds with it.