-
Notifications
You must be signed in to change notification settings - Fork 0
feat(player): A/B 대조 청취 픽스처 + 블라인드 프로토콜 #53
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merged
Merged
Changes from all commits
Commits
Show all changes
2 commits
Select commit
Hold shift + click to select a range
File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,28 @@ | ||
| # A/B 블라인드 청취 프로토콜 | ||
|
|
||
| 목적: "좋은 개발 흐름이 더 듣기 좋게 들리는가" 가설을 사람 귀로 검증한다. | ||
| 자동 지표 대조는 `player/tests/contrastListening.test.mjs`가 CI에서 고정한다. | ||
|
|
||
| ## 준비 | ||
|
|
||
| - Player 골든 리스닝 팩의 `대조 A — 모범 PR 흐름` / `대조 B — 거친 이력` 사용. | ||
| - 두 데모는 같은 저장소·같은 PR 번호로 **조성(tonic)과 모티프가 동일**하게 | ||
| 통제되어 있다 — 차이는 이벤트 내용(개발 흐름의 질)뿐. | ||
|
|
||
| ## 절차 (평가자 1인당 약 3분) | ||
|
|
||
| 1. 진행자는 평가자에게 A/B 라벨과 배경 설명 없이, 무작위 순서로 두 데모를 | ||
| 자동 재생으로 들려준다 (각 1회, 원하면 재청취 허용). | ||
| 2. 청취 후 3문항에 답하게 한다 (첫 번째/두 번째로 답변): | ||
| - Q1. 어느 쪽이 더 **안정적**으로 들렸는가? | ||
| - Q2. 어느 쪽이 더 **완결된 곡**처럼 끝났는가? | ||
| - Q3. 다시 듣고 싶은 쪽은? | ||
| 3. 진행자는 재생 순서와 답변을 기록한다. | ||
|
|
||
| ## 판정 | ||
|
|
||
| - 평가자 n ≥ 3, 문항별로 "모범 흐름(A)" 선택이 과반이면 가설 지지로 기록. | ||
| - 결과는 `docs/maestro-player/evidence/`에 날짜 파일로 남긴다 | ||
| (예: `2026-08-XX-ab-listening-results.md` — 평가자 수, 문항별 집계, 특이 코멘트). | ||
| - 과반 미달 시: 소리 층(보이싱·음색) 개선 후 재시험하고, 매핑 규칙 자체를 | ||
| 의심하기 전에 합성 품질 변인을 먼저 제거한다. |
40 changes: 40 additions & 0 deletions
40
docs/superpowers/specs/2026-08-04-player-ab-contrast-design.md
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,40 @@ | ||
| # Maestro Player A/B 대조 청취 픽스처 설계 | ||
|
|
||
| - 날짜: 2026-08-04 | ||
| - 상태: 확정 ("좋은 코딩이 듣기 좋게 들리는가" 가설의 직접 검증 트랙) | ||
| - 범위: `player/` + `docs/maestro-player/` 하위만 | ||
|
|
||
| ## 0. 목표 | ||
|
|
||
| 같은 저장소·같은 PR 번호(= 같은 조성·모티프로 **통제**)에서 이벤트 내용만 | ||
| 다른 두 시나리오를 골든 팩에 추가해, 개발 흐름의 질이 소리 차이로 | ||
| 번역되는지를 ① 자동 지표로, ② 블라인드 청취로 검증 가능하게 한다. | ||
|
|
||
| ## 1. 픽스처 쌍 (repoId `ab-contrast-showcase`, pr:7 공유) | ||
|
|
||
| - **대조 A — 모범 PR 흐름**: 적정 규모 feat 커밋 3 + fix 1 + pr-open → | ||
| review-comment → 반영 → approve → **merge**. 기대: ionian(장조계), | ||
| 낮은 평균 긴장, merge가 해소 정점. | ||
| - **대조 B — 거친 이력**: 대형 덤프 커밋 + **revert×3** + request-changes + | ||
| reopen, 머지 없음, 짧은 간격 재작업. 기대: phrygian(가장 어두운 선법), | ||
| sus4(긴장 코드), 높은 평균 긴장, 해소 정점 부재. | ||
|
|
||
| ## 2. 자동 대조 게이트 (`tests/contrastListening.test.mjs`) | ||
|
|
||
| - 두 세션의 **tonic 동일**(통제 확인) + A=ionian, B=phrygian. | ||
| - B.chordColor = sus4 (revert 지배 규칙). | ||
| - B.tensionScore ≥ A.tensionScore + 0.15 / A.resolutionScore ≥ B + 0.1. | ||
| - A의 peakResolution 이벤트 = merge, B는 merge 부재. | ||
| - 기존 정량 하니스(스케일·그리드·밀도·도약)가 두 픽스처에 자동 적용됨 | ||
| (골든 팩 엔트리로 편입되므로). | ||
|
|
||
| ## 3. 블라인드 청취 프로토콜 (`docs/maestro-player/ab-listening-protocol.md`) | ||
|
|
||
| 평가자에게 A/B 라벨을 숨기고 두 데모를 무작위 순서로 들려준 뒤 3문항 | ||
| (어느 쪽이 안정적/완결적/다시 듣고 싶은가)을 수집하는 절차. n≥3이면 | ||
| 과반 일치로 가설 지지 여부를 기록. | ||
|
|
||
| ## 4. 영향 | ||
|
|
||
| 골든 팩 3→5 엔트리 (기존 개수 단언 테스트 갱신). UI는 카드 2장 추가 | ||
| (다이어트로 컴팩트 카드라 부담 없음). |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,49 @@ | ||
| // A/B 대조 게이트 (스펙 2026-08-04 대조 §2): 개발 흐름의 질 차이가 음악 지표 차이로 번역되는지 고정한다. | ||
| import test from 'node:test'; | ||
| import assert from 'node:assert/strict'; | ||
|
|
||
| import { | ||
| buildContrastCleanFlowFixture, | ||
| buildContrastRoughFlowFixture, | ||
| buildGoldenListeningPackEntries, | ||
| } from '../src/lib/goldenListeningPack.js'; | ||
| import { buildMusicPlan } from '../src/lib/musicIntentMapper.js'; | ||
|
|
||
| function buildSessions() { | ||
| const clean = buildMusicPlan(buildContrastCleanFlowFixture(), { laneCount: 4 })[0]; | ||
| const rough = buildMusicPlan(buildContrastRoughFlowFixture(), { laneCount: 4 })[0]; | ||
| return { clean, rough }; | ||
| } | ||
|
|
||
| test('대조 쌍은 같은 tonic으로 통제되고 선법이 명확히 갈린다 (ionian vs phrygian)', () => { | ||
| const { clean, rough } = buildSessions(); | ||
|
|
||
| assert.equal(clean.harmony.tonic, rough.harmony.tonic, '같은 repo+PR이면 tonic이 같아야 함'); | ||
| assert.equal(clean.harmony.mode, 'ionian'); | ||
| assert.equal(rough.harmony.mode, 'phrygian'); | ||
| assert.equal(rough.harmony.chordColor, 'sus4'); // revert 지배 → 긴장 코드 | ||
| }); | ||
|
|
||
| test('거친 이력은 평균 긴장이 높고, 모범 흐름은 해소가 높다', () => { | ||
| const { clean, rough } = buildSessions(); | ||
|
|
||
| assert.ok( | ||
| rough.tensionScore >= clean.tensionScore + 0.15, | ||
| `tension 대조 부족: rough ${rough.tensionScore.toFixed(3)} vs clean ${clean.tensionScore.toFixed(3)}`, | ||
| ); | ||
| assert.ok( | ||
| clean.resolutionScore >= rough.resolutionScore + 0.1, | ||
| `resolution 대조 부족: clean ${clean.resolutionScore.toFixed(3)} vs rough ${rough.resolutionScore.toFixed(3)}`, | ||
| ); | ||
| }); | ||
|
|
||
| test('모범 흐름의 해소 정점은 merge이고 거친 이력에는 merge가 없다', () => { | ||
| const entries = buildGoldenListeningPackEntries(); | ||
| const cleanEntry = entries.find((entry) => entry.id === 'contrast-clean-flow'); | ||
| const roughEntry = entries.find((entry) => entry.id === 'contrast-rough-flow'); | ||
|
|
||
| assert.ok(cleanEntry && roughEntry, '대조 엔트리가 골든 팩에 편입되어야 함'); | ||
| assert.equal(cleanEntry.peakResolutionEvent.eventType, 'merge'); | ||
| assert.notEqual(roughEntry.peakResolutionEvent?.eventType, 'merge'); | ||
| assert.equal(roughEntry.peakTensionEvent.eventType, 'review-request-changes'); | ||
| }); |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
With the new rough B fixture this relaxation is needed only for sessions that intentionally have no cadence/merge, but applying it to every fixture means an existing golden scenario can lose all accent notes and the harness still passes as long as some other fixture has one. That would stop this test from catching broken merge/cadence chord voicing on individual scenarios; please keep the per-fixture assertion for fixtures that contain cadence-producing events and exempt only the unresolved contrast case.
Useful? React with 👍 / 👎.