fpf.tools
Library / First Principles Framework (FPF) - Core Conceptual Specification
Jump to passage
In this reading
Source · lines 105646–105668

Link to current text

Published source confirmed at last check

Source changed 2026-10-03 15:41:27 UTC · snapshot created 2026-10-03 15:43:46 UTC · last check 2026-10-03 15:55:10 UTC

F.1:4.6 - Self-checks

For finding:

  • Question-preservation test. Could a changed or omitted query term change what would answer the original question? Restore that condition when inspecting candidates.
  • Reach test. What material could the available route actually inspect, and which consequential blind spot remains? Use a complementary route only if it could repair that gap.
  • Candidate-return test. Can the receiver obtain the inspected passage, see why it matters and distinguish missing context or an unavailable original?
  • Stop test. Does the next use already have sufficient material, or is its exact missing basis returned? Do not turn a scoped no-hit result into a claim of absence.

For source selection:

  • Answer-change test. What can each retained source change in the answer or action? If nothing, exclude it or state the missing role.
  • Rival-and-limit test. Is a known rival explanation, counterexample, or transfer limit still hidden? Add the source that makes it inspectable or return the source gap.
  • One-source test. Does one already identified source close the ordinary question? If yes, stop without additional F.1 steps. If the cut will support a SoTA claim, take this exit only when that source is a current critical synthesis of the serious alternatives and no known action-changing rival or counterexample remains hidden.
  • SoTA-role test. Has each retained source been classified under one of the comparison roles defined in E.8:11? If not, classify it there before using the cut for a SoTA claim; do not recreate the role meanings in F.1.
  • Rank test. Did official status, popularity, maintenance state, date, or a catalogue check raise a source’s SoTA role? Remove that inference; keep the check only as source identity/currentness evidence.
  • Gap test. If the best-known line cannot be established, does the note return the missing rival, counterexample, or synthesis need as a source gap rather than naming the newest available source?
  • Locality test. Have two sources been treated as saying the same thing merely because they use the same word? If so, use the ordinary F.0.1 branch on the exact passages before deciding their roles, then return to selection.
  • Search-policy test. Did a score decide membership before source claims were inspected? Undo that decision.
  • Memory test. Can the receiver hold the cut and each source’s role in view? Remove non-changing material or split genuinely different questions.
  • Reopen test. Can the note say what future change would require selection again?