# IL 101 — Project Charter Rubric：行為錨點 / Behavior Anchors

> 本文件只補充現行六個評分面向的「行為錨點」說明，權重與及格門檻沿用網站既有設定（見課程頁「評分項目與
> 權重」），不更動任何計分邏輯。
> / This document only adds behavior anchors for the six existing scoring dimensions; the weights and
> pass threshold are unchanged from the site's existing configuration (see the course page's
> "Scoring Criteria and Weights"), and no scoring logic is modified here.

共同四級量尺 / Common 4-level scale: 0 = 無法判定 / cannot be judged,
1 = 待改善 / needs improvement, 2 = 合格 / adequate, 3 = 優良 / strong.

## 中文

| 面向（權重） | 0 | 1 | 2 | 3 |
|---|---|---|---|---|
| 問題與範圍品質（20%） | 問題陳述空泛，無法判斷改善對象 | 有問題陳述，但範圍模糊或缺少現況流程 | 問題具體、有現況流程，in/out scope 清楚 | 同左，並能說明範圍界定如何避免專案中途擴大 |
| 價值邏輯（20%） | 沒有價值假設或邏輯不通 | 有價值假設，但缺少計算依據 | 依 SDD 公式計算，標記估計／實測 | 同左，並揭露限制與重複計算風險 |
| KPI／證據設計（20%） | 沒有可衡量指標 | 指標存在但缺單位／分子分母／來源 | 指標卡六欄位齊全（見 Module 3） | 同左，並有多個指標互相對照佐證 |
| 可行性與準備度（15%） | 未提及依賴或風險 | 列出依賴但未評估影響 | 依賴清單具體，說明未到位的影響 | 同左，並已有初步緩解計畫 |
| 人機責任分工（15%） | 未區分 AI 與人的角色 | 有區分但人的角色僅是「核准」 | 人機分工具體到情境層級（見 Module 4） | 同左，並明確保留人的最終責任與定期審查機制 |
| 主管溝通品質（10%） | 使用大量術語，主管無法理解 | 基本可讀，但缺乏決策所需重點 | 主管可在不懂技術細節下判斷是否投資 | 同左，並能預先回答常見追問 |

## English

| Dimension (weight) | 0 | 1 | 2 | 3 |
|---|---|---|---|---|
| Problem/scope quality (20%) | Vague problem statement; the target for improvement cannot be judged | Has a problem statement, but scope is unclear or the current workflow is missing | Concrete problem, current workflow present, in/out scope is clear | As left, and explains how the scoping prevents scope creep |
| Value logic (20%) | No value hypothesis, or the logic does not hold up | Has a value hypothesis but no calculation basis | Calculated per the SDD formula, labeled estimate/validated | As left, and discloses limitations and double-counting risk |
| KPI/evidence design (20%) | No measurable metric | A metric exists but is missing a unit/numerator-denominator/source | All six metric-card fields present (see Module 3) | As left, with multiple metrics cross-checking each other |
| Feasibility/readiness (15%) | No dependencies or risks mentioned | Dependencies listed but their impact is not assessed | Concrete dependency list with the impact of each gap explained | As left, with an initial mitigation plan already in place |
| AI/human responsibility split (15%) | AI and human roles are not distinguished | Distinguished, but the human role is only "approval" | The split is concrete down to the scenario level (see Module 4) | As left, with the human’s final responsibility and a periodic review mechanism explicit |
| Executive communication quality (10%) | Heavy jargon; a sponsor cannot understand it | Basically readable but missing the points needed for a decision | A sponsor can judge whether to invest without understanding technical detail | As left, and pre-answers common follow-up questions |