fpf.tools
Library / First Principles Framework (FPF) - Core Conceptual Specification
Jump to passage
In this reading
Source · lines 84904–84939

Link to current text

Published source confirmed at last check

Source changed 2026-10-03 08:25:59 UTC · snapshot created 2026-10-03 08:26:43 UTC · last check 2026-10-03 09:55:09 UTC

E.8:4.1.2 - Preserve Pattern Use Value Across Material Revisions

A revision is material when the actual change can alter what a working reader recognizes, does, obtains, or must stop doing, regardless of whether the change is labelled as cleanup, clarification, terminology repair, or ontology alignment. Treat the revision as material when it can change at least one of these values:

  • the primary EntityOfConcern, governed kind, direct relation, claim kind, or scope;
  • the recurring situation or practical question that lets a reader recognize the use;
  • a Solution action, action condition, result kind, first useful result, stop, return, risk disclosure, or stronger-neighbor handoff;
  • the definition, constraint, test, method, cited-pattern contribution, split, merge, relocation, or source/SoTA stance that changes what the reader may do;
  • the asserted commonality, member set, membership rule, order, or governing premise of a list; or
  • ordinary first-use affordability.

For this comparison, the earlier edition is the exact accepted pattern edition that this candidate is intended to replace for the declared use. A formatting correction, spelling repair, citation repair, exact mechanical rendering, or wording change is not triggered only when the smallest comparison of the earlier edition and proposed text shows that all these values are preserved. A clean comparison needs no additional positive ledger, evidence table, or pattern section. Physical line count, file size, section count, inventory rows, and the author’s label for the change do not establish materiality.

Use one bounded material-revision loop over the actual prose. Before treating a materially revised pattern as authored:

  1. Recover the useful earlier-edition use at idea level: the recognizable situation and intended reader, first admissible action or judgement, first useful result, action-changing boundary or stop, and any domain claim, example, or relation needed to perform that move. Classify a changing or disappearing earlier-edition use only as retained, a valid outcome whose defective mechanism is repaired, an explicitly authorized retirement with a corrected action or boundary, or unsupported residue.
  2. Draft the candidate’s positive practitioner path in domain-recognizable language before guards: governed subject, recurring problem, action the reader can take, first useful result, and next action-changing condition or stop.
  3. Compare the earlier edition and proposed text at comparable application effort. Preserve every useful earlier-edition move or deliberately replace it with an at-least-equally-usable action, result, or boundary; admit a candidate-only use only from an exact accepted decision, source/SoTA stance, finding, or working need.
  4. Apply F.19 to each changed natural span. Remove exactness intensifiers, invented counterreadings, role or process wrappers, formal identities, and assurance apparatus that fail its contribution test, while preserving every kind, relation, use, and action-changing detail. Keep ordinary pattern-use wording ordinary; open a deeper FPF route only for a genuinely unresolved value.
  5. Check that recognition, first action, and first useful result still precede optional modeling, evidence, conformance, and assurance work. F.19 is the common semantic pass over the changed span; E.10 is a cue and an exact route for residual FPF wording, not a second normal-pass algorithm.
  6. For every changed public or consumed interface—entry wording, input or result, field or position meaning, action order, stop, return, or reconsideration condition—repair each determinate stale ToC or README cue, example, relation, and true direct consumer in the same authoring increment. Find consumers by the meaning they teach or use; a shared word, identifier, or nearby reference is not enough.

Earlier-edition and candidate-only uses remain different bases, and both may be present in one revision. Compare that exact earlier edition with the candidate edition. An earlier-edition use keeps its earlier-edition basis and one of the four classifications above; a candidate-only use keeps its exact accepted basis. Do not classify a candidate-only use as an earlier-edition use or invent history for it. Treat a selected use as required when its loss changes action or boundary, and as optional when it demonstrates breadth only. Backward compatibility alone is not improvement, and a candidate-only promise is not improvement until the text supports its executable use. Use desk replay by default and escalate to a cold reader, AI-agent, or observed-work check only when ambiguity or consequence justifies it. If later independent review needs a recoverable note, use the smallest existing authoring source; do not create a card, score, universal schema, or one written row per idea.

Test first-use affordability by checking whether the positive Solution supports this short rendering:

recognizable situation -> proposed action or judgement -> first useful result -> next action-changing condition or stop

This rendering explains the pattern; it does not claim that actual work is linear. Use an optional local mantra only when it improves recall, and show one ordinary traversal only when several rows materially improve explanation; choose the smallest form that keeps the action, result, and boundary recoverable. Explanatory rows may fade as competence or task demand permits, but an independently action-changing condition or boundary may not. An explanation retained for reuse can concern the actual subject, question or proposed continuations before CGUS qualification; use C.2.1 when its episteme identity matters. A demonstrative slice about a qualified CGUS is a separate episteme under A.22.CGUS, whose use depends on that qualification. Put a subject-side check immediately before the continuation it changes, and keep authoring, review, quality, and release checks outside the subject Solution.

Resolve authoring lists with F.19. When a list can change pattern use, apply the same connected F.19 reading used for prose. E.8 keeps only the authoring effect: put the practitioner’s proposition or action before illustrative material; declare a genuinely normative closed set as closed under its governing rule; signal examples as non-exhaustive when a plausible reader could mistake them for a classification; and do not let a noun series or catalogue replace the Solution.

Do not add a second enumeration taxonomy or a per-member result form. E.10 may cue a suspicious head or series, F.19 decides its membership semantics and discourse load, and an exact subject pattern settles any unresolved kind, relation, or normative set.