title: "Gather — "What do we NOT know?"" source: "tasks/TFW-38__quality_enforcement/research2/gather.md"

Gather — "What do we NOT know?"¶

Parent: HL-TFW-38 Goal: Evidence for domain-agnostic review stages and review modes.

Findings¶

G1: External — Universal Quality Gate Stages (Domain-Independent)¶

ISO 9001 and industry QA frameworks identify three universal quality activities:

Activity	Question	Domain-Independent?
Verification	"Are we building the product right?" = process check	✅ Yes — applies to code, docs, research, design
Validation	"Are we building the right product?" = outcome check	✅ Yes — applies everywhere
Conformance	Does it match standards/requirements?	✅ Yes — universal

Industry standard quality gate progression: 1. Requirements Gate → does output match spec? 2. Quality Gate → does output meet internal standards? 3. Acceptance Gate → does output fulfill end-user needs?

These are NOT code-centric. "Requirements" could be TS acceptance criteria, design requirements, or content specs. "Quality" could be code conventions, writing style, or pedagogical standards.

G2: Natural Stages in Real TFW REVIEW Files¶

Analyzing the best REVIEW files, natural stages emerge:

HD-3 PhaseF REVIEW (code task, 103 lines): 1. Read — checklist with evidence (§1) 2. Acceptance Criteria Audit — line-by-line AC verification (§2) 3. Tech Debt Collection — triage observations (§3) 4. Process Verdict — scope budget, artifacts, discipline (§4) 5. Fact Candidates (§5) 6. Final Verdict

HD-2 Phase0 REVIEW (document task, 76 lines): 1. Read — checklist with N/A for code-centric items (§1) 2. AC verification detail (inline) 3. Tech Debt (§3) 4. Traces (§4) 5. Fact Candidates — "none" (§5)

Pattern: Both follow the same structure but the CODE review has deeper verification (line numbers, formula checks). The DOC review marks code-centric items as N/A.

G3: Current 9-Point Checklist — Domain Analysis¶

#	Check	Code	Docs	Research	Design
1	DoD met?	✅ always	✅ always	✅ always	✅ always
2	Code quality	✅ core	N/A	N/A	N/A
3	Test coverage	✅ core	N/A	N/A	N/A
4	Philosophy aligned	✅ always	✅ always	✅ always	✅ always
5	Tech debt	✅ always	✅ always	✅ always	✅ always
6	Security	✅ conditional	N/A	N/A	N/A
7	Breaking changes	✅ conditional	N/A	N/A	N/A
8	Style & standards	✅ always	✅ partial	✅ partial	✅ partial
9	Observations collected	✅ always	✅ always	✅ always	✅ always

Universal items (all domains): #1 DoD, #4 Philosophy, #5 Tech debt, #8 Style, #9 Observations = 5 items Code-specific items: #2 Code quality, #3 Tests, #6 Security, #7 Breaking changes = 4 items

The checklist is ~44% code-specific. For non-code tasks, reviewers write "N/A" for 4 items = friction.

G4: Review Stage Candidates (Domain-Agnostic Naming)¶

Mapping ISO verification/validation/conformance to TFW review context:

Stage	Cognitive Mode	What Reviewer Does	Parallel to Research
Comprehend	Passive understanding	Read RF, TS, HL. Understand what was done and why. Answer: "Do I understand the claims?"	Briefing
Verify	Active audit	Open deliverables. Spot-check RF claims against reality. Run one test/build if code. Check deliverable existence if docs. Answer: "Are the claims true?"	Gather
Assess	Analytical judgment	Apply domain-relevant checklist. Judge quality, standards, philosophy alignment. Answer: "Is the quality sufficient?"	Extract
Synthesize	Decision-making	Write verdict, collect tech debt, capture facts. Answer: "What's the verdict and what did we learn?"	Challenge + RES

This is a 4-stage model: - Comprehend = read and understand (current Step 1) - Verify = audit (the missing step from iteration 1) - Assess = checklist + judgment (current Step 2, but with domain-aware items) - Synthesize = verdict + traces (current Steps 3-7 compressed)

Checkpoint¶

Found	Remaining
ISO verification/validation/conformance model is universal	None
Current checklist is 44% code-specific — needs domain config	None
4-stage model: Comprehend → Verify → Assess → Synthesize	Needs classification of what "Verify" means per domain
Natural stages already emerge in best REVIEW files	None

Sufficiency: - [x] External source used? (ISO 9001, QA gate frameworks) - [x] Briefing gap closed? (Domain-agnostic stages designed)

Stage complete: YES → User decision: ___