Why this page exists為什麼有這頁
Reading regulation closely produces mistakes. Paragraph counts get miscounted on first read, sub-point references get the layer wrong, dates drift across edits, terminology that was correct in one phase becomes wrong in the next. The honest position is not to claim this site is error-free; it is to make every correction visible.
認真讀法規會犯錯。段數第一次讀會數錯，子條的層級會引錯，日期會在多次編輯之間漂移，某個階段對的術語在下個階段變成錯的。誠實的立場不是宣稱本站從不出錯；是讓每一處修正都看得到。
EU law itself uses this convention. The Official Journal publishes corrigenda when it finds errors in regulations after publication. A reference site about EU regulations should adopt the same discipline. This page is the corrigenda log for CRA Notebook.
歐盟法規本身就用這套做法。OJ 在公告後發現錯誤時、會發 corrigenda（更正）。一個關於歐盟法規的參考網站應該採用同樣的紀律。本頁就是 CRA Notebook 的 corrigenda log。
What gets logged here這裡記錄什麼
Only corrections that affect a reader's understanding of the CRA, or that would mislead a reader who relied on an earlier version. Concretely: wrong Article references, wrong sub-point layers, wrong paragraph counts, wrong dates, wrong deadlines, terminology corrections that change meaning, conclusions that flipped. Site-wide editorial changes (numbering, layout, navigation, design system) are not logged here; they belong in the changelog. Polish-level edits to in-progress essays are not logged anywhere; that is what the Working status is for.
只記錄會影響讀者對 CRA 內容理解、或讀者依舊版做事會被誤導的修訂。具體來說：條號錯、子條層級錯、段數錯、日期錯、期限錯、改變語意的術語修正、結論翻盤。site 層級的編輯改動（編號、版面、導覽、設計系統）不記在這裡；那些屬於 changelog。在 書寫 狀態的文章做 polish 不記錄在任何地方；這就是 書寫 狀態本身的意義。
Log紀錄
E001
2026-04-29
Substantive實質錯誤
Article 14 sub-point references in PSIRT routing table
PSIRT 路由表中的第 14 條子條引用
- What was wrong原錯
-
The PSIRT routing table on
article-17.html labelled the 24h, 72h, and 14-day reporting deadlines as Article 14(1), 14(2), and 14(3). This collapses the OJ's actual structure: 14(1) is the trigger duty for vulnerabilities, and the three timing sub-points sit under 14(2) as (a), (b), (c). 14(3) is a separate trigger duty for severe incidents.
article-17.html 上的 PSIRT 路由表把 24 小時、72 小時、14 天三個通報時點分別標成 第 14(1) 條、14(2)、14(3)。這把 OJ 真實層級壓平了：14(1) 是「應通報弱點」的觸發義務本體，三個時點 (a)、(b)、(c) 都在 14(2) 之下。14(3) 是另一條觸發義務（嚴重事件通報）。
- What it is now現修為
-
Sub-points relabelled
14(2)(a), 14(2)(b), 14(2)(c) for the vulnerability track. A new row added for the severe-incident parallel track at 14(4)(a)–(c), with the 1-month final-report clock distinguished from the 14-day vulnerability clock.
弱點軌道的子條改標為 14(2)(a)、14(2)(b)、14(2)(c)。新增嚴重事件平行軌道列、條號 14(4)(a)–(c)、把 1 個月最終報告時鐘跟 14 天弱點時鐘明確區分。
- Source出處
-
Verified against OJ L 2024/2847, Article 14, p.36–38.
對照 OJ L 2024/2847 第 14 條、p.36–38 字面確認。
- Why it matters為什麼重要
-
Anyone using the table to design a PSIRT runbook would have written the wrong sub-point references into their internal documentation. The deadlines themselves were correct; the citation layer was wrong. Easy to spot in legal review, easy to fix once spotted.
任何用這張表設計 PSIRT runbook 的人會把錯的子條寫進內部文件。時點本身是對的，引用的層級錯。法律審查時容易抓到、抓到後容易修。
Files affected: 影響檔案：
article-17.html
E002
2026-04-29
Substantive實質錯誤
Article 13 paragraph count in act-stub metadata
第 13 條 act stub metadata 段落數
- What was wrong原錯
-
The metadata strip on
article-13.html showed Paragraphs · 24. Article 13 of OJ L 2024/2847 has 25 paragraphs.
article-13.html 的 metadata 框寫 段落數 · 24。OJ L 2024/2847 第 13 條共有 25 段。
- What it is now現修為
Paragraphs · 25 / 段落數 · 25.- Source出處
-
Verified against OJ L 2024/2847, Article 13, p.35–38.
對照 OJ L 2024/2847 第 13 條、p.35–38 字面確認。
- How it survived為何漏
-
A site-wide grep sweep on the previous build picked up "24 paragraphs" / "Twenty-four paragraphs" / "24 段條文" and similar phrasings, but did not match the metadata-strip pattern
· 24. The metadata template was a separate component the grep didn't account for.
前一版的 site-wide grep sweep 抓到「24 paragraphs」、「Twenty-four paragraphs」、「24 段條文」之類寫法，但沒抓到 metadata 框的 · 24 pattern。metadata 模板是另一個元件、grep 沒涵蓋。
Files affected: 影響檔案：
article-13.html
E003
2026-04-29
Substantive實質錯誤
Article 14 paragraph count in act-stub metadata
第 14 條 act stub metadata 段落數
- What was wrong原錯
-
The metadata strip on
article-14.html showed Paragraphs · 9. Article 14 of OJ L 2024/2847 has 10 paragraphs.
article-14.html 的 metadata 框寫 段落數 · 9。OJ L 2024/2847 第 14 條共有 10 段。
- What it is now現修為
Paragraphs · 10 / 段落數 · 10.- Source出處
-
Verified against OJ L 2024/2847, Article 14, p.36–38.
對照 OJ L 2024/2847 第 14 條、p.36–38 字面確認。
- How it survived為何漏
-
Same blind-spot as E002: the metadata template was outside the scope of the previous grep sweep.
與 E002 同一盲點：metadata 模板在前次 grep sweep 範圍之外。
Files affected: 影響檔案：
article-14.html
Editorial discipline編輯紀律
When a substantive error is found:
發現實質錯誤時：
- The fix is made and the change pushed in the next build.
- 立即修正，下一版 build 推上線。
- A new entry is added here, with a permanent ID (
E001, E002...), the date, the severity tag, what was wrong, what it is now, the source consulted, and how it survived earlier review.
- 在這頁新增一筆紀錄，帶永久 ID（
E001、E002……）、日期、嚴重度標籤、原錯內容、現修內容、引用來源、為何在先前的審查中漏掉。
- The entry is never removed or rewritten — only superseded if a later entry corrects it. This is what permanence means.
- 紀錄永不刪除、永不改寫——僅能被後續紀錄超越（如果後續修正動到同一處）。這就是 permanence 的意思。
- If the error appeared in a syndicated copy of the content (LinkedIn, Medium, etc.), a correction note is also posted there with a link back to this errata entry.
- 如果該錯誤出現在內容的同步版本（LinkedIn、Medium 等），也在該處發布更正說明、回鏈本勘誤紀錄。