Files
confidence-engine/docs/archive/experiments/vol-1-chapters/README.md
T

285 lines
8.8 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# Design Evolution Archive — Tranche 1
This directory contains exact historical extracts from
`docs/design-evolution-log.md`.
The original monolithic log remains intact and authoritative while the archive refactor is in progress.
These files are historical provenance, not current product truth.
For current product state use:
- docs/current-handoff.md
- docs/current-project-state.md
## Verified extracts
### Chapter 1
Path:
docs/archive/experiments/vol-1-chapters/ch1/experiments-01-to-06-and-phases-1-to-4.md
Original source:
docs/design-evolution-log.md lines 1352
Contents:
Phases 14 and Experiments 0106.
Fidelity:
Exact contiguous copy.
### Chapter 2
Path:
docs/archive/experiments/vol-1-chapters/ch2/early-reasoning-and-provenance-discovery.md
Original source:
docs/design-evolution-log.md lines 3541122
Contents:
Experiments 0720 and associated emerging-direction material.
Fidelity:
Exact contiguous copy.
### Chapter 3
Path:
docs/archive/experiments/vol-1-chapters/ch3/phase-transition-and-emerging-directions.md
Original source:
docs/design-evolution-log.md lines 11241216
Contents:
Phase Transition, Graph as Source of Truth, and Experiments 2122.
Fidelity:
Exact contiguous copy.
### Chapter 4
Path:
docs/archive/experiments/vol-1-chapters/ch4/passive-classifiers-and-explore-contract-validation.md
Original source:
docs/design-evolution-log.md lines 12181501
Contents:
Experiments 2325B and their closeout/return-to-work material.
Fidelity:
Exact contiguous copy.
## Tranche 2
### Chapter 5
Path:
docs/archive/experiments/vol-1-chapters/ch5/context-inventory-routing-and-handoff-infrastructure.md
Original source:
docs/design-evolution-log.md lines 15022052
Contents:
Experiments 2634 — context routing inventory and handoff infrastructure design.
Fidelity:
Exact contiguous copy.
### Chapter 6
Path:
docs/archive/experiments/vol-1-chapters/ch6/handoff-validation-behavior-selection-and-assessor-audit.md
Original source:
docs/design-evolution-log.md lines 20532503
Contents:
Experiments 3541 and Experiment 41 Conclusion — handoff validation, behavior selection, and assessor audit.
Fidelity:
Exact contiguous copy.
Tranches 1, 2, 3, and 4 have now been extracted.
The original monolithic log remains intact and authoritative while extraction is incomplete.
## Tranche 3
### Chapter 7
Path:
docs/archive/experiments/vol-1-chapters/ch7/experiments-42-to-46.md
Original source:
docs/design-evolution-log.md lines 25043069
Contents:
Experiments 4246 — narrow acknowledge exclusion, Clarify readiness audit, assessor unclear-starting-point, and "Too Broad" boundary investigations.
Fidelity:
Exact contiguous copy.
### Chapter 8
Path:
docs/archive/experiments/vol-1-chapters/ch8/experiments-47-to-51.md
Original source:
docs/design-evolution-log.md lines 30703534
Contents:
Experiments 4751 — shared-anchor coherence diagnostic, unknown relationship audit, production shared-anchor test, and semantic decision-relevance coherence experiments.
Fidelity:
Exact contiguous copy.
## Tranche 4
### Chapter 9
Path:
docs/archive/experiments/vol-1-chapters/ch9/experiments-52-to-52i.md
Original source:
docs/design-evolution-log.md lines 35354980
Contents:
Experiments 5252I and Correction to Experiment 52B Conclusion — semantic interpretation generalisation, evaluation using existing Qwen model, separate meaning from relevance labels, normalisation into decision-relevance contract, supports_decision/could_change_decision boundary coherence, ambiguity normalisation failures, ambiguous meaning filled with domain expectations, ambiguity wording resistance, grounding rule for relationship strength invention, and provenance boundary analysis.
Fidelity:
Exact contiguous copy.
## Tranche 5
### Chapter 10
Path:
docs/archive/experiments/vol-1-chapters/ch10/experiments-54a-to-54k-provenance-audit.md
Original source:
docs/design-evolution-log.md lines 49816423
Contents:
Experiments 54A54K — complete provenance audit chain including graph provenance verification, trace through pipeline, update flow determinism, prompt-level provenance analysis, evidence reference integrity, evidenceType reliability, deterministic source linkage, pre-LLM source identity establishment, multi-interpretation source anchoring, meaning separation capability, and automated semantic grounding with live model inference.
Fidelity:
Exact contiguous copy.
Tranches 1 through 6 have now been extracted.
The original monolithic log remains intact and authoritative while extraction is incomplete.
## Tranche 6
### Chapter 11
Path:
docs/archive/experiments/vol-1-chapters/ch11/experiments-54l-to-54q-grounding-and-evidence-stability.md
Original source:
docs/design-evolution-log.md lines 64257824
Contents:
Experiments 54L54Q — semantic grounding stability under repetition, disagreement exposure between interpretations, disagreement-driven information need changes, evidence-need discrimination for same-goal scenarios, explicit evidence needs recovering higher-level consequences, and structured semantic steps preserving evidence distinction with consequence.
Fidelity:
Exact contiguous copy.
### Chapter 12
Path:
docs/archive/experiments/vol-1-chapters/ch12/experiment-54r-clarification-requires-source.md
Original source:
docs/design-evolution-log.md lines 78258012
Contents:
Experiment 54R — testing whether models correctly distinguish disagreements resolvable through evidence from those requiring user input, across competing causal hypotheses, ambiguous user priority, and absent material disagreement patterns.
Fidelity:
Exact contiguous copy.
## Tranche 7
### Chapter 13
Path:
docs/archive/experiments/vol-1-chapters/ch13/54S-54V-clarification-target-and-answer-resolution.md
Original source:
docs/design-evolution-log.md lines 80138635
Contents:
Experiments 54S54V — clarification target identification, no-clarification-stability across repeated inputs, whether a clarification target becomes a useful user question without adding new meaning, and whether a clarification answer can resolve only the target.
Fidelity:
Exact contiguous copy.
### Chapter 14
Path:
docs/archive/experiments/vol-1-chapters/ch14/54W-54Z-clarification-chain-and-target-broadening.md
Original source:
docs/design-evolution-log.md lines 86369294
Contents:
Experiments 54W54Z — clarification chain end-to-end integrity, specificity loss under chaining, target broadening effects on question and resolution, and resolution differences when the answer is less explicit.
Fidelity:
Exact contiguous copy.
## Tranche 8
### Chapter 15
Path:
docs/archive/experiments/vol-1-chapters/ch15/55A-preserve-uncertainty-from-weak-clarification-answers.md
Original source:
docs/design-evolution-log.md lines 92969529
Contents:
Experiment 55A — Can the Model Preserve Uncertainty From Weak Clarification Answers? Testing fully explicit hard constraint, weak priority statement, conditional trade-off, and non-answer/insufficient clarification cases.
Fidelity:
Exact contiguous copy.
### Chapter 16
Path:
docs/archive/experiments/vol-1-chapters/ch16/55B-separate-answer-meaning-from-resolution-judgement.md
Original source:
docs/design-evolution-log.md lines 95309748
Contents:
Experiment 55B — Separate Answer Meaning from Resolution Judgement. Mode A instruction/output contract versus Mode B inference separation, tested on weak priority, conditional trade-off, and non-answer cases.
Fidelity:
Exact contiguous copy.
### Chapter 17
Path:
docs/archive/experiments/vol-1-chapters/ch17/55C-resolution-from-preserved-answer-meaning.md
Original source:
docs/design-evolution-log.md lines 97499982
Contents:
Experiment 55C — Resolution From Preserved Answer Meaning. Two-stage resolution using preserved meaning rather than raw answer, tested on weak priority, conditional trade-off, and non-answer cases.
Fidelity:
Exact contiguous copy.
### Chapter 18
Path:
docs/archive/experiments/vol-1-chapters/ch18/55D-separate-stated-vs-inferred-meaning-through-v058-provenance.md
Original source:
docs/design-evolution-log.md lines 998310322
Contents:
Experiments 55D (separate stated clarification meaning from inference), 55E (reasoning refinement requirements synthesis), 55F (reasoning requirements production path map), and v0.51v0.58 Progress — product provenance and architectural decisions.
Fidelity:
Exact contiguous copy.
The extraction phase is complete.
Tranches 1 through 8 now preserve the substantive historical contents of
`docs/design-evolution-log.md` as exact chapter extracts.
The original monolithic log remains intact and authoritative until a separate
compatibility/routing task replaces it with an index or pointer and updates repository references.
## Refactor status
Do not remove these ranges from `docs/design-evolution-log.md` yet.
Do not update repository-wide references to point here yet.
The original monolith remains authoritative pending migration.