Management and Business Ethics · EMI · NTUST College of Management
Six ideas you need to fill in the roles table and the weekly log
Ten people writing one document is a project, not a homework assignment. The difference is that the work has to be divided before it is done, and recombined afterwards — and both of those steps fail silently if nobody is responsible for them.
A work breakdown (W1–W10), an owner column, a reviewer column, a due column, six milestones, and a weekly log.
Why those columns exist, what each one is for, and the specific mistake each one prevents.
Every slide has a Listen button that reads an English summary aloud, with the same text printed beside it.
Idea 0 · Framing
Temporary · unique deliverable · progressively elaborated
Temporary: it has a defined start and finish.
Unique: the deliverable did not exist before.
Elaborated as it goes: you know least on day one.
Routine work is learned by repetition. A project is finished by the time you would have learned it.
So the structure must be chosen at the start, by people who have not done it before. That is not a flaw in your team; it is the normal condition of project work.
Three things you are managing, and they trade against each other. Scope — how much you analyse. Time — the submission date, which is fixed. Effort — the hours your members actually have. Since time is fixed and effort is nearly fixed, the only variable you really control is scope. Teams that refuse to limit scope end up cutting quality instead, usually in the last 48 hours.
Idea 1 · Scope
The single most useful habit in this whole deck
| Activity — cannot be checked | Deliverable — can be checked |
|---|---|
| “Research the stakeholders” | Stakeholder map, 8 groups, power and legitimacy separated |
| “Work on the ethics part” | 3×5 ethics-test matrix, a verdict in every cell |
| “Help with the slides” | Segment 4 of the presentation, 3 minutes, rehearsed once |
| “Do the references” | Reference list in APA, every in-text citation matched |
And write down what is out. “We are not analysing the parent company.” “We are answering Q6 and Q7, not Q8 or Q9.” A boundary you never declared is a boundary you will argue about in the last week, when there is no time to move it.
Idea 2 · Work breakdown structure (WBS)
Three rules, and your table already follows them
1 · The 100% rule. The pieces add up to all the work — nothing missing, nothing extra.
2 · No overlap. If two members can both think a piece is theirs, it is badly defined.
3 · Small enough to own. One person, one piece, one output someone else can inspect.
How small is small enough? A common rule of thumb is that a work package should be between roughly 8 and 80 hours of work — big enough to be worth tracking, small enough that being stuck on it is visible within a week. If a package cannot be reported on in a weekly meeting, split it.
Idea 2b · Definition
A title is not a definition
| Field | Example — W3 Stakeholder map | Where you record it |
|---|---|---|
| Output | A map of 8 stakeholder groups, power and legitimacy distinguished | Table 1, work package name |
| Destination | Section 3.2 of the report, plus an extended version of Exhibit 2 | Appendix column “where it appears” |
| Owner | One named member — never two | Table 1, Owner |
| Done means | Reviewer has read it and signed it off; it is in the shared file | Table 1, Reviewer + Due date |
The failure this prevents. Ten pieces of work, each competent, that do not join up — three members analysing a different unit of analysis, two stakeholder lists that disagree, a recommendation that answers a question nobody asked. Naming the destination at the start is what stops it.
Idea 3 · Ownership
Why the table has two name columns and not one
Exactly one name. Answers for the package at the meeting whether or not they did every hour of the work.
Two names in this column is the most common way a team loses a section: each assumes the other is finishing it.
A different member reads the entire output before it goes in.
Assign in a ring: A → B → C → … → A. Nothing in the report has
been read by only one person.
Review is not proofreading. The reviewer’s question is not “are there typos?” but “does this answer the question that was asked, and would I defend it in the Q&A?” A reviewer who only fixes grammar has not reviewed anything.
Idea 4 · Dependencies
The critical path, in the only form you need it
A day lost here is a day lost by the whole team. W1 is the clearest case: change the unit of analysis late and W2, W4 and W9 all have to be rewritten.
W3, W6, W7 and W8 can run in parallel and have slack. A day lost there may cost nothing — provided it is handed to W10 on time.
Practical consequence. Settle W1 in the first meeting even if it feels premature, and give W10 a real deadline of its own. Integration is not a formality performed on the last evening; it is the package with the most dependencies in the project.
Idea 5 · Milestones
Zero duration · binary · verifiable by an outsider
| # | Milestone | What is verifiably true on that date |
|---|---|---|
| 1 | Case and roles fixed | Table 1 is filled in: every package has an owner and a reviewer. |
| 2 | First drafts in | Every W-package has a draft in the shared file. Not “nearly”. |
| 3 | Review and revision done | Every output has been read and signed off by its reviewer. |
| 4 | Integration and English editing | One file, one voice, one reference list, one unit of analysis. |
| 5 | Presentation rehearsed | Run once, timed, inside 15 minutes, every member having spoken. |
| 6 | Submission | Uploaded, with the signed contribution statement attached. |
Test for a real milestone. Could a stranger walk in on that date and tell whether you hit it, without asking anyone how they feel about it? If not, it is a task wearing a milestone’s clothes.
Idea 5b · Scheduling
Forward planning produces a schedule that ends after the due date
People underestimate their own task durations even when they know they have done so before. The fix is not optimism management; it is evidence: how long did your last assignment of this size actually take?
At the end of the project, as named slack before submission — not hidden inside each person’s task, where it is always consumed and never visible.
Also known as student syndrome: work expands to fill the time available and then starts at the last possible moment anyway. Milestone 2 — drafts in — exists precisely to make that visible while there is still time to react.
Idea 6 · Tracking
Binary progress against a definition of done
“About 70% done.” It is a guess, it is optimistic, and the last 10% will take as long as the first 90. Packages famously sit at “90%” for weeks.
Is the output in the shared file, and has the reviewer signed it off? Yes or no. Nothing in between is reportable.
| Weekly log, section A | Why that column is there |
|---|---|
| What was completed | Forces a deliverable, not an activity. |
| Where it is | If it has no location, it is not finished. |
| Reviewed by | Done means a second person has read it. |
Idea 7 · Risk
Section D of the weekly log
Avoid — change the plan so the risk cannot occur.
Reduce — make it less likely or less damaging.
Transfer — move it to someone better placed to carry it.
Accept — consciously, in writing, with a trigger for acting.
Single point of failure: one member owns a 30-point section alone.
Late framing: the unit of analysis is still moving in week three.
Integration crush: W10 receives everything on the last evening.
Silent member: somebody has not produced anything and nobody has said so.
A risk with no response is not a risk entry. “D is behind on W4” is an observation. “D is behind on W4, so B finishes W2 and joins W4 as a second pair of hands while remaining its reviewer, and if it slips again we tell the instructor rather than hide it” is a risk entry.
Idea 8 · Change control
The question is never whether, only by which route
Table 1 is a baseline, not a contract. The point of a baseline is not that it cannot move — it is that when it moves, you can see what moved. A team that quietly edits the plan has no baseline at all, and nothing to write its contribution statement from.
You are graded on the report you produce. Table 3 records what each member actually did, not what Table 1 assigned in week one.
A team whose plan never changed either had unusual luck or is not telling you something.
A contribution statement that follows the original plan rather than what happened is not true, and every member signs it.
It also removes your only defence if a member later disputes who did what.
When the two documents disagree, the log wins. It is dated and it was written at the table; the table is a summary that someone forgot to update.
Idea 8 · Change control
Cheapest first
| The change | What to do | Cost |
|---|---|---|
| Too much work for the time left | Drop one of Q6–Q9: only three are required, so the fourth is free. Never cut ethics and CSR — 30 points. | lowest |
| A date will be missed | Move the date, keep the milestone. A milestone that has moved twice is not asking for a third date; it is telling you to cut scope. | low |
| An owner leaves or goes silent | The reviewer takes the package over — they have already read it — and a new reviewer is assigned. | low |
| A package is too big for one person | Split it: W4a and W4b, each with its own owner and reviewer. Do not put two names on one row. | low |
| The unit of analysis (W1) changes after the drafts are in | W2, W4 and W9 all have to be reworked — W1 sits at the front of the chain. Decide as a whole team; write down what is redone, by whom, by when. | high |
The cheap rows change who or when. The expensive row changes what the report is about — which is why the first meeting belongs to W1, not to the schedule.
Idea 9 · Meetings
Sections B and C of the weekly log
What was settled, and why. The reason is what stops the decision being reopened in week five by someone who missed the meeting.
A named member, a named output, a date. “We should look at the regulator’s report” is not a commitment; it has no owner.
Filled in during the meeting, read back before you leave, signed by the recorder. The recorder rotates each week.
Why the log asks you to read it back. Two members who left the same meeting with different understandings is the normal case, not the exception. Reading the decisions aloud before anyone stands up costs ninety seconds and is the cheapest control in this entire deck.
Failure modes
None of these is a character problem; all of them are structural
| Failure | What catches it |
|---|---|
| Work divided by task type — “you find sources, I’ll type” | Work packages are deliverables tied to grading criteria, not activities. |
| Two owners for one package | The Owner column takes one name. |
| Nobody owns integration | W10 has an owner, a reviewer and a date of its own. |
| Free riding found out too late | Weekly binary reporting in log section A: output, location, reviewer. |
| Review is only proofreading | The reviewer answers for the content in the Q&A, not just the spelling. |
One more, and it is the quiet one. A team that never writes anything down has no evidence when a member disputes the contribution statement at the end — and the member may be right. The log protects the diligent members and the accused one equally.
How the documents fit
Two documents, one system
WBS (W1–W10), owner, reviewer, due date, six milestones, presentation segments. Filled in at the start.
Progress, decisions, commitments, risks, and the photo of the week. Filled in at every meeting.
The contribution statement attached to the report. It is a summary of the log, not a new document.
The test. Every sentence in your contribution statement should be traceable to a line in a log page. If it is not, either the log is incomplete or the sentence is not true.
Closing
Everything else in this deck is support for these two lists
Concepts used in this deck: project characteristics and the scope–time–effort trade-off · deliverables vs. activities · work breakdown structure, the 100% rule and work-package definition · single-owner accountability and peer review · dependencies and the critical path · milestones · backward planning, buffers and the planning fallacy · binary progress measurement against a definition of done · risk identification and the four responses · decision and commitment records.
管理與企業倫理 · EMI · 臺科大管理學院
填角色分工表和每週紀錄前要懂的六個觀念
十個人合寫一份文件，是專案，不是一般作業。差別在於工作要先拆開才能做，做完還要再合起來——這兩步只要沒人負責，就會無聲無息地出錯。
一份工作分解（W1–W10）、Owner（負責人）欄、Reviewer（審閱者）欄、Due（期限）欄、六個里程碑（milestone），還有每週紀錄（Weekly Work Log）。
這些欄位為什麼存在、各自的用途，以及每一欄要防的是哪一種錯。
每一頁都有 Listen 按鈕，會朗讀英文摘要，旁邊同時列出同一段文字。
觀念 0 · 定位
有期限 · 成果獨一無二 · 邊做邊具體化
有期限：有明確的開始和結束。
獨一無二：交付成果是以前沒有的東西。
邊做邊具體化：第一天知道的最少。
例行工作靠重複來學會；專案則是等你學會，它也結束了。
所以架構必須在一開始、由沒做過的人決定。這不是你們團隊的缺陷，而是專案工作的常態。
你們要管三件事，而且它們互相牽制。範疇（Scope）——分析多少。時間（Time）——繳交日，固定不變。投入（Effort）——組員實際有的時數。時間固定、投入也幾乎固定，真正能調的只有範疇。不肯限縮範疇的團隊，最後犧牲的是品質，而且通常是在最後 48 小時。
觀念 1 · 範疇（Scope）
整份簡報裡最有用的一個習慣
| 活動（Activity）——無法檢查 | 交付成果（Deliverable）——可以檢查 |
|---|---|
| 「研究利害關係人」 | 利害關係人圖，8 個群體，權力與正當性分開標示 |
| 「做倫理的部分」 | 3×5 倫理檢驗矩陣，每一格都有判斷 |
| 「幫忙做簡報」 | 簡報第 4 段，3 分鐘，排練過一次 |
| 「處理參考文獻」 | APA 格式的參考文獻，每個文內引用都對得上 |
也要寫下哪些不做。「我們不分析母公司。」「我們回答 Q6 和 Q7，不答 Q8、Q9。」沒有事先講明的界線，到最後一週就會吵起來，而那時已經沒時間改了。
觀念 2 · 工作分解結構（WBS）
三條規則，你們的表已經照著做了
1 · 100% 規則。各塊加起來就是全部的工作——不缺也不多。
2 · 不重疊。如果兩個組員都可能以為某一塊是自己的，就是定義得不好。
3 · 小到一個人能負責。一個人、一塊、一份別人能檢查的產出。
多小才算夠小？常見的經驗法則是：一個工作包（work package）約 8 到 80 小時的工作量——大到值得追蹤，小到卡住時一週內就看得出來。如果一個工作包沒辦法在每週會議上回報，就再拆。
觀念 2b · 定義
有標題不等於有定義
| 項目 | 範例——W3 利害關係人圖 | 記在哪裡 |
|---|---|---|
| 產出 | 8 個利害關係人群體的圖，區分權力與正當性 | Table 1，工作包名稱 |
| 去處 | 報告第 3.2 節，加上 Exhibit 2 的延伸版 | 附錄「where it appears」欄 |
| 負責人（Owner） | 一位指名的組員——絕不能兩位 | Table 1，Owner |
| 完成的定義 | 審閱者已讀過並簽核；已放進共用檔案 | Table 1，Reviewer＋Due |
這樣做是在防什麼。十份工作各自都做得不錯，卻兜不起來——三個組員分析的分析單位（unit of analysis）各不相同、兩份利害關係人清單互相矛盾、建議回答的是沒人問的問題。一開始就寫明去處，才擋得住。
觀念 3 · 權責歸屬（Ownership）
為什麼表上有兩欄名字，而不是一欄
只寫一個名字。不管是不是每個小時的工作都由他做，會議上都由他對這個工作包負責。
這一欄寫兩個名字，是團隊弄丟一個章節最常見的原因：兩人都以為對方會收尾。
由另一位組員在產出放進報告前，把整份讀過。
排成一圈：A → B → C → … → A。報告裡沒有任何部分只有一個人讀過。
審閱不是校對。審閱者要問的不是「有沒有錯字？」，而是「這有沒有回答到題目？Q&A 時我敢不敢替它辯護？」只改文法的審閱者，等於什麼都沒審。
觀念 4 · 相依關係（Dependency）
要徑（critical path），只講你們用得到的部分
這裡慢一天，就是全組慢一天。W1 最明顯：分析單位太晚改，W2、W4、W9 全都要重寫。
W3、W6、W7、W8 可以同時進行，也有餘裕。這裡慢一天可能毫無影響——前提是準時交給 W10。
實務上的意思。就算覺得還太早，第一次開會就把 W1 定下來，並且給 W10 一個真正屬於它的期限。整合不是最後一晚走個形式；它是整個專案裡相依關係最多的工作包。
觀念 5 · 里程碑（Milestone）
不佔時間 · 非是即否 · 外人也能驗證
| # | Milestone（里程碑） | 到那一天，什麼事可以驗證為真 |
|---|---|---|
| 1 | 個案與角色確定 | Table 1 填好了：每個工作包都有 Owner 和 Reviewer。 |
| 2 | 初稿到齊 | 每個 W 工作包在共用檔案裡都有初稿。「快好了」不算。 |
| 3 | 審閱與修改完成 | 每份產出都由它的 Reviewer 讀過並簽核。 |
| 4 | 整合與英文潤稿 | 一份檔案、一種語氣、一份參考文獻、一個分析單位。 |
| 5 | 簡報排練完成 | 完整走過一次、有計時、15 分鐘內、每位組員都開口講過。 |
| 6 | 繳交 | 已上傳，並附上簽名的貢獻聲明（contribution statement）。 |
怎麼判斷是不是真的里程碑。一個陌生人在那天走進來，不必問任何人「感覺如何」，能不能看出你們有沒有達成？不能的話，那只是披著里程碑外衣的任務。
觀念 5b · 排程
從前往後排，排出來的時程會超過截止日
人會低估自己做事要花的時間，就算明知以前低估過也一樣。解法不是叫自己別太樂觀，而是看證據：上次同樣份量的作業，實際花了多久？
放在專案的最後，當作繳交前明確標出的餘裕——不要藏在每個人的任務裡，藏起來的一定會被用掉，而且沒人看得見。
這也叫學生症候群（student syndrome）：工作會膨脹到把時間填滿，結果還是拖到最後一刻才開始。Milestone 2（初稿到齊）就是為了趁還來得及反應時，讓這件事現形。
觀念 6 · 進度追蹤
對照「完成的定義」，只回答有或沒有
「大概完成 70%。」這是猜的，而且偏樂觀；最後 10% 花的時間會跟前面 90% 一樣多。工作包卡在「90%」好幾週是出了名的事。
產出在共用檔案裡了嗎？審閱者簽核了嗎？有或沒有，介於中間的都不算可回報的進度。
| 每週紀錄 A 區 | 為什麼有這一欄 |
|---|---|
| 完成了什麼 | 逼你寫交付成果，而不是活動。 |
| 放在哪裡 | 說不出放在哪，就是還沒完成。 |
| 誰審閱過 | 完成的意思是有第二個人讀過。 |
觀念 7 · 風險（Risk）
每週紀錄 D 區
避免（Avoid）——改計畫，讓風險不會發生。
降低（Reduce）——讓它較不可能發生，或傷害較小。
轉移（Transfer）——交給更有條件承擔的人。
接受（Accept）——有意識地接受、寫下來，並訂好何時該出手。
單點失效：一位組員獨自負責 30 分的章節。
問題界定太晚：到第三週分析單位還在變。
整合塞車：W10 在最後一晚才收到所有東西。
沉默的組員：有人什麼都沒交，也沒人說破。
沒寫應對方式的風險，不算一筆風險紀錄。「D 的 W4 進度落後」只是觀察。「D 的 W4 進度落後，所以 B 做完 W2 後加入 W4 幫忙，同時仍是它的審閱者；如果再延誤，我們會告訴老師，不隱瞞」才是風險紀錄。
觀念 8 · 變更管理（Change control）
問題從來不是會不會變，而是循什麼管道變
Table 1 是基準（baseline），不是合約。基準的意義不在於不能動，而在於動的時候看得出動了什麼。悄悄改計畫的團隊等於沒有基準，寫貢獻聲明時也無所依據。
評分看的是你們交出的報告。Table 3 記的是每位組員實際做了什麼，不是第一週 Table 1 怎麼分配。
計畫從頭到尾沒變過的團隊，不是運氣特別好，就是有事沒說。
照原計畫而不是照實際情況寫的貢獻聲明並不屬實，而且每位組員都要簽名。
日後若有組員對誰做了什麼有異議，你們也失去唯一的憑據。
兩份文件不一致時，以紀錄為準。紀錄有日期，而且是當場寫的；表格只是摘要，可能有人忘了更新。
觀念 8 · 變更管理
從代價最低的開始
| 變更 | 怎麼處理 | 代價 |
|---|---|---|
| 剩下的時間做不完 | Q6–Q9 捨掉一題：只要求三題，第四題本來就可以不做。絕不能砍倫理與 CSR——30 分。 | 最低 |
| 會趕不上某個日期 | 改日期，里程碑留著。已經延過兩次的里程碑，要的不是第三個日期，而是在提醒你們該縮小範疇。 | 低 |
| 負責人退出或失聯 | 由審閱者接手工作包——他已經讀過內容——再另外指派新的審閱者。 | 低 |
| 工作包一個人做不完 | 拆開：W4a 和 W4b，各有自己的 Owner 和 Reviewer。不要在同一列寫兩個名字。 | 低 |
| 初稿到齊後分析單位（W1）變了 | W2、W4、W9 全都要重做——W1 在鏈的最前端。全組一起決定；寫下重做什麼、誰做、何時完成。 | 高 |
代價低的幾列改的是誰做或何時做；代價高的那列改的是報告在談什麼——所以第一次開會該處理的是 W1，不是時程。
觀念 9 · 會議
每週紀錄 B 區與 C 區
定案了什麼、為什麼。寫下理由，才不會在第五週被沒來開會的人翻案重談。
指名的組員、指名的產出、一個日期。「我們應該看一下主管機關的報告」不算承諾，因為沒有負責人。
開會時填寫、離開前唸一遍、由記錄者簽名。記錄者每週輪流。
為什麼紀錄要你們唸一遍。兩個組員開完同一場會，理解卻不一樣，這是常態，不是例外。在大家起身前把決議唸出來只要九十秒，是整份簡報裡成本最低的控管手段。
失敗模式
這些都不是個性問題，全是結構問題
| 失敗方式 | 靠什麼抓住 |
|---|---|
| 按工作類型分工——「你找資料，我打字」 | 工作包是對應評分標準的交付成果，不是活動。 |
| 一個工作包有兩個負責人 | Owner 欄只填一個名字。 |
| 整合沒人負責 | W10 有自己的 Owner、Reviewer 和日期。 |
| 搭便車（free riding）太晚才發現 | 每週在紀錄 A 區用是非回報：產出、位置、審閱者。 |
| 審閱只是校對 | 審閱者在 Q&A 要為內容負責，不只是拼字。 |
還有一種，而且悄無聲息。什麼都不寫下來的團隊，最後有組員對貢獻聲明提出異議時拿不出證據——而那位組員可能是對的。紀錄對認真的組員和被指責的組員，保護是一樣的。
文件之間的關係
兩份文件，一套系統
WBS（W1–W10）、Owner、Reviewer、Due、六個里程碑、簡報分段。一開始就填好。
進度、決議、承諾、風險，以及當週照片。每次開會都填。
附在報告後的貢獻聲明。它是紀錄的摘要，不是另一份新文件。
檢驗方式。貢獻聲明裡的每一句話，都應該能追溯到某一頁紀錄裡的某一行。追不到的話，不是紀錄不完整，就是那句話不屬實。
結語
這份簡報的其他內容，都是在支撐這兩張清單
這份簡報用到的觀念：專案特性與範疇、時間、投入的取捨 · 交付成果與活動 · 工作分解結構、100% 規則與工作包定義 · 單一負責人當責與同儕審閱 · 相依關係與要徑 · 里程碑 · 倒推排程、緩衝與規劃謬誤 · 對照完成的定義、以是非衡量進度 · 風險辨識與四種應對 · 決議與承諾紀錄。