今周刊報導 (1)，供應鏈管理成功的第一步是高層的決心
『南北十二座工廠、多達十二萬名員工，要想「知人」可不是一件容易的事，更別提派系林立一直是舊奇美電就有的難題。因此，董事長段行建下的第一個指令，就是要所有人「繳械」大清點，不論客戶訂單細節、報價，或各類情報、人才，都要大清點。
聽起來再平常不過的事，王志超卻搖了搖頭，「基於機密、系統不同等各種理由，這在舊奇美電，是不可能做到的。」派系間不願共享情報的結果，常常產品經理設定零組件價格在二十元，採購執行後，買回來卻是三十元。
他進一步舉物流管理為例，奇美電產出的面板，得運送到大陸浙江寧波做後段加工，負責物管的員工建議，每周一、三、五的運費較便宜，所以周二、四生產的面板就累積到每周一、三、五出貨，看起來物管人員替奇美電省下可觀運費，其實不然。
「面板是每天報價，周二報價可能比周三高，十月三十一日出貨與十一月一日出貨，算的就是不用季度的營收，一差就是好幾億元，但省下的船費卻微乎其微。」王志超坦言，舊奇美電時代，他只能看到物管人員替公司做了最好的選擇，但看不到的卻是庫存、跌價損失。』
接下來必須克服的就是不同人和部門間動機 (incentive) 或獎勵方式不一致的問題，解決此問題的同時必須考慮如何訂定一致的目標；報導中運費、庫存、和報價的平衡，應該合在一起考慮；生管課程中的價值密度 (value density) 和經濟訂購量 (EOQ) 便是一些基礎的工具；如果有價格的變動，要考慮更複雜的模型。
報導中也提及如何『搶利潤 展開差異化策略』，因為正實施中，成效要等時間來驗證。
(1) 賴筱凡，變身規格制訂者 奇美電擺脫殺價宿命，今周刊，2012 年 10 月，第 827 期
12/30/2012
U 型生產線：湧德電子的例子
湧德電子專注於網路連接器 (1)，商業周刊報導
『一個零組件只有三十樣的產品，湧德團隊竟擁有一百件專利，還有近五十件專利等待核准，專利，就是湧德的入場門票。‧‧‧
鴻海用大規模生產線，生產單一、量大的產品壓低成本，湧德就走客製化設計、少量多樣的市場，鴻海擅長經營國際 OEM（代工製造）大客戶，湧德就經營防鴻陣營客戶，讓自己成為最重要的第二供應商。‧‧‧
放棄傳統的流水線生產模式，改用日本豐田式的 U 型線生產，這種生產方式的特色是，工人幾人一組，自主控制生產的流程和效率，這種方式一天可以換線七到八次，傳統流水式生產線一天只能變換一次產品，理論上，U 型線生產更適合湧德少量多樣的經營模式。‧‧‧
現在，湧德的生產線人數只有過去的八成，降低缺工影響，同時「生產效率比過去增加一倍多，」中衛中心顧問吳宗憲觀察，同樣生產一款產品，以前一條生產線要一百二十人，現在卻只要五個人，「今年，我們的第五代生產線，只要三個人。」陳旻徹說。』
圖片來源 (2)
U 型生產線的另外一個優點是較易達成生產線平衡 (line balancing)，使每個人的生產速度接近。只是不解的是，120 人如何變成 3 個人 (120 * 0.8 = 96)？
(1) 電腦後面連接網路線的小小插孔。文章全文參見林宏達，網路連接器小廠湧德，瓜分全球近半市占，鴻海告不倒它 淨利兩年跳三倍，商業周刊，第1300期，2012-10-22
(2) R.B. Chase, F.R. Jacobs, and N.J. Aquilano, Operations Management for Competitive Advantage with Global Cases, 11th edition, McGraw-Hill, 2006, page 240.
12/24/2012
管理真的有用？
三個最佳實踐 (best practices) 和精實製造 (lean manufacturing)
作者經過十幾年的研究 (1)，針對製造業、零售業、醫院、和學校，根據三大最佳的管理實踐 (targets, incentives, monitoring)，給與 1 到 5 分的評價，詳細的問題參見此網頁。發現
A one-point increment on a five-point management score correlated with better performance at manufacturers around the globe.
1 point increment on a management score was associated with
+23% productivity
+14% market capitalization
+1.4% percentage points annual sales growth
A one-point improvement in the management score is associated with about a 10% jump in student test performance. Similarly, at hospitals, a one-point management-score increase is associated with a 0.5% lower 30-day mortality rate for heart attack victims who are admitted to emergency rooms.
作者也在印度的 28 個工廠中，實驗和比較使用精實製造的差距 (影片)，發現
On average, firms that received the management intervention cut defects by half, reduced inventory by 20%, and raised output by 10%.
(1) Nicholas Bloom, Raffaella Sadun, and John Van Reenen, Does Management Really Work?, November 2012, Harvard Business Review. 中文的翻譯
作者經過十幾年的研究 (1)，針對製造業、零售業、醫院、和學校，根據三大最佳的管理實踐 (targets, incentives, monitoring)，給與 1 到 5 分的評價，詳細的問題參見此網頁。發現
A one-point increment on a five-point management score correlated with better performance at manufacturers around the globe.
1 point increment on a management score was associated with
+23% productivity
+14% market capitalization
+1.4% percentage points annual sales growth
A one-point improvement in the management score is associated with about a 10% jump in student test performance. Similarly, at hospitals, a one-point management-score increase is associated with a 0.5% lower 30-day mortality rate for heart attack victims who are admitted to emergency rooms.
作者也在印度的 28 個工廠中，實驗和比較使用精實製造的差距 (影片)，發現
On average, firms that received the management intervention cut defects by half, reduced inventory by 20%, and raised output by 10%.
(1) Nicholas Bloom, Raffaella Sadun, and John Van Reenen, Does Management Really Work?, November 2012, Harvard Business Review. 中文的翻譯
12/23/2012
哈佛商業評論三篇大資料 (Big Data) 的文章
因為出現了大量的、及時的、和多樣的資料，企業可以更精準地瞭解其顧客行為或改進內部的營運。根據作者的研究 (1)，正確使用此資料的優點為
In particular, companies in the top third of their industry in the use of data-driven decision making were, on average, 5% more productive and 6% more profitable than their competitors.文中舉航空業 (PASSUR) 和零售業 (Sears) 為例說明之，也陳述了管理的挑戰 (leadership, talent management, technology, decision making, company culture) 和開始的方法
- Pick a business unit to be the testing ground. It should have a quantfriendly leader backed up by a team of data scientists.
- Challenge each key function to identify five business opportunities based on big data, each of which could be prototyped within five weeks by a team of no more than five people.
- Implement a process for innovation that includes four steps: experimentation, measurement, sharing, and replication.
- Keep in mind Joy’s Law: “Most of the smartest people work for someone else.” Open up some of your data sets and analytic challenges to interested parties across the internet and around the world.
12/09/2012
即品網：降價再降價
商業周刊報導『即品網與時間作戰，越不景氣它越紅』：
『通路商和供應商間有「允收期」的規定，超過一半到三分之二的可用效期後，通路就不願收貨了。
這塊餅不是沒人要吃，陳德仁說，通路商不願收即期品，就是不願承擔庫存壓力，任何人想靠它賺錢，都必須與時間作戰。
但與時間作戰，即品網如何打贏勝仗？靠的是少量多樣，推陳出新，快速週轉。
而且隨著期限越來越短，價格越來越低，讓顧客們養成來挖寶的習慣，回購率竟超過了六成。
不過，「要在彈性中求規模化，是很兩難的事情，」一位大型食品商業務員分析，彈性進貨與銷貨很花人工，一旦規模化，管理庫存、顧客關係維持與快速銷貨能力，都將面臨挑戰。』
快過期商品和時裝一樣，藉由降價將貨品售出 (1)，以增加公司的收入；屬於營收管理 (Revenue Management) 的範疇。一般而言，必須根據消費者的需求曲線，以決定降價的時間和幅度，譬如第三天降一半、第五天降成四分之三，整個問題可以表示成一個最佳化的問題。推陳出新的商品可以利用類似商品的銷售資料推估，等待有真正的銷售產生後，可以修正需求曲線的參數。
管理大規模的庫存必須藉由軟體系統的輔助，這牽涉到主事者的決心、適當的人才、和資訊系統的投資報酬率。如果有了資訊系統，也可以從事複雜的行銷活動，例如推薦系統、銷售組合等等。一般的步驟可以參考此書。
(1) 雖然兩者的金額大小不同，沒銷售完的時裝有殘值 (salvage value)；但是，在數學建模 (mathematical modeling) 上而言，兩者是相同的。
『通路商和供應商間有「允收期」的規定，超過一半到三分之二的可用效期後，通路就不願收貨了。
這塊餅不是沒人要吃，陳德仁說，通路商不願收即期品，就是不願承擔庫存壓力，任何人想靠它賺錢，都必須與時間作戰。
但與時間作戰，即品網如何打贏勝仗？靠的是少量多樣，推陳出新，快速週轉。
而且隨著期限越來越短，價格越來越低，讓顧客們養成來挖寶的習慣，回購率竟超過了六成。
不過，「要在彈性中求規模化，是很兩難的事情，」一位大型食品商業務員分析，彈性進貨與銷貨很花人工，一旦規模化，管理庫存、顧客關係維持與快速銷貨能力，都將面臨挑戰。』
快過期商品和時裝一樣，藉由降價將貨品售出 (1)，以增加公司的收入；屬於營收管理 (Revenue Management) 的範疇。一般而言，必須根據消費者的需求曲線，以決定降價的時間和幅度，譬如第三天降一半、第五天降成四分之三，整個問題可以表示成一個最佳化的問題。推陳出新的商品可以利用類似商品的銷售資料推估，等待有真正的銷售產生後，可以修正需求曲線的參數。
管理大規模的庫存必須藉由軟體系統的輔助，這牽涉到主事者的決心、適當的人才、和資訊系統的投資報酬率。如果有了資訊系統，也可以從事複雜的行銷活動，例如推薦系統、銷售組合等等。一般的步驟可以參考此書。
(1) 雖然兩者的金額大小不同，沒銷售完的時裝有殘值 (salvage value)；但是，在數學建模 (mathematical modeling) 上而言，兩者是相同的。
10/26/2012
值得你拼命的公司？(1/2)
連 Google 都不值得
久未聯絡的同學打電話邀約見面，以為是提早退休雲遊四海，那知是疑似重病復發 (1)，被公司資遣。
剛入社會的人，模仿別人打拼的方式，工作到十一、二點，然後一、二點才睡覺；當身體發出警訊時，不是拖延、就是沒有放慢腳步；等你生病了，需要時間檢查和休息時，公司不是想辦法和你度過難關，而是請你回家吃自己。不可思議的是該公司年年賺大錢，台灣知名媒體還有大幅地報導！
10/22/2012
退休軍公教人員優惠存款
如果改成 8%，每年多增加 450 億的經費 (1)
根據自由電子報的報導『72年實施迄今 歷經數次改革』：
但是，根據聯合報的社論『終於有感：勞軍公保同步檢討』，似乎不可能：
根據自由電子報的報導『72年實施迄今 歷經數次改革』：
"七十二年時，市場一般利率升為十二％，政府單位將優存利率下限修正提高為年息十八％，十八％的優存利率就此沿用至今。政府為補貼十八％與一般存款的差額利息，八十四年度需補貼約一百八十八億元，但到了一○○年度，因一般存款利息已剩不到二％，利差補貼金額已成長至八○一億餘元；在此同時，退休軍公教辦理優惠存款的戶數，也從八十四年度的廿七萬餘戶，至今年六月底已大幅成長至五十一萬餘戶。"如果以利差做為基準，改成 (18 - 12 + 2= ) 8%，每年多增加 450 億的經費。根據紐約時報的報導 Japan's Big-Works Stimulus Is Lesson，說明政府投資在教育和社會服務比硬體建設好：
"Moreover, it matters what gets built: Japan spent too much on increasingly wasteful roads and bridges, and not enough in areas like education and social services, which studies show deliver more bang for the buck than infrastructure spending."所以增加的經費可以用在教育文化，譬如建設學校的英聽教室、聘請合格教師，補充高中以下各級學校的圖書、實驗和電腦設備，專款專用以增加各縣市的圖書經費，訓練與補助年輕人創業，輔導失業的中年人等等。
但是，根據聯合報的社論『終於有感：勞軍公保同步檢討』，似乎不可能：
"尤其是二十年來，主政者昧於形勢，遲遲不肯隨客觀環境的變化調整制度，結果讓整體軍公教揹上汙名，不時成為外界責難的目標。以所得替代率為例，環顧世界，除了財政破產的希臘外，哪個國家退休公務員像台灣一樣領到九成以上的薪俸？但考試院卻說各國皆然。日昨行政院會決議將上限調整為八成，顯示此一比率絕非無法討論的「聖牛」；然而比起勞保的四成，這八成之數恐怕仍有再議的空間。"(1) 退休軍公教人員優惠存款：wiki
10/21/2012
台灣的大學學費
貴嗎？
以學費除以國民所得，在亞洲四小龍、日本、中國、菲律賓之間 (註 1)，台灣最低。目前的低學費政策是拿大家的錢補助收入較高的家庭，更不公平 (註 2)；表四是 2001 年的資料，如果再加上這幾年新增的 5 年 500 億，補助頂尖大學的數字更驚人。而弱勢學生較多的私立大學 (註 3)，補助少得可憐，我們的學生平均不到一萬。看到報載，一些台清交的學生說要凍漲學費，他們是否有想過自己是制度的受益者？
我們應該實施兩項不可能的改革，一是適當地提高學費，二是改採學分制收費。增加的經費用以增加對弱勢族群的獎助學金和改善教學環境，例如所需的軟硬體、輔導學生所需的資源等等 (註 4)。
幾年前的金融風暴，許多排名前面大學的學生不畢業，每年只需繳 5 萬的學雜費，繼續留在學校，卻享用數十萬的補助，顯然不合理 (註 5)。
在美國，任何時候重修都要收錢。在台灣，只有暑修或延畢要收錢。修我課程的學生，有的從來不來上課、有的來了只是坐在那裡不動、有的從來不交作業、或者考試可以帶書和講義卻考零分。詢問助教或老師，每次加學期總成績一分，至多五分，有不少人是從來不問我們的。如果不及格重修要花錢，學生自然會比較認真。前幾天有位同學來找我，問了一下，其父母都從事辛苦的工作，卻沒有任何理由就不來上課導致被退學。
如果學分提早修滿了，也就可以畢業，不用再繳學費，以鼓勵認真的學生。
單位學費成本提高了，學生也會更用心地要求老師提高教學品質 (註 2)。
我接觸到系上一些需要自食其力的打工同學，成績其實都還不錯。上課時也常常告訴同學，要多多地充實自己，培養專業，使用腦力賺錢，才能改善家中的經濟環境。以學校中的課程助教而言，每一個小時 150 元，如果沒有人前來詢問，還可以讀自己的書。以我們資管系最近畢業的學生為例，如果學會資訊系統的軟體開發，這些 (大學部) 專業工作的起薪大都是 30,000 到 42,000 元 (註 6)。有位女同學起薪三萬，表現良好，不到一年就調成 43,000 元。如果月薪四萬元，比起其他同學的二萬元，二年之間就多了 48 萬，可以還完助學貸款。所以，念大學值不值得完全看個人 (註 7)。
(註 1) 菲律賓的例子。美國是貴得嚇死人。歐陸的賦稅負擔率很高，所以學費便宜，參考我國大學學雜費與國際之比較 (doc 檔)；這幾年卻大幅度地高漲，例如法國的 Insead MBA 一年的費用是 5 萬 8 千歐元，或搜尋『英國大學學費』。
(註 2) 吳聰敏和駱明慶，『好老師，莫強求』。2001 年，教育部對台大、成大、淡江、和逢甲每位學生的平均補助分別為 25.22、19.99、2.36 和 2.36 萬元 (pdf 檔)
(註 3) 這學期全校學生辦理就學貸款三千多人，超過 3 成。導生班 50 人中，20 人辦理就學貸款，達 4 成；大部份是家境因素而需要辦學貸，有的則是希望靠自己的力量完成學業。
(註 4) 以 UT 為例，工學院有 7700 位學生，學院專屬的就業輔導室人員就有七位，可以幫助學生看履歷表和實施模擬面談等等 (www 1, 2)
(註 5) 5 萬元是學雜費，但是，根據現行的規定，通常不用繳到 5 萬元。許多排名前面學校的延畢生達二成。詳見自由電子報的報導『使用者付費 台大、政大及清大延畢生增收學雜費』
(註 6) 該起薪 42,000 元同學的成績普通：但是，畢業專題的指導教授和業界有顧問關係，同學畢專的訓練是該公司所需，所以起薪高。
(註 7) 除此之外，還有各式各樣免費的通識和外系課程、專業與職場相關的演講、訓練、和活動。至於大學教育是否應該強調就業是個有爭議的問題，參見書籍『大學教了沒』。
以學費除以國民所得，在亞洲四小龍、日本、中國、菲律賓之間 (註 1)，台灣最低。目前的低學費政策是拿大家的錢補助收入較高的家庭，更不公平 (註 2)；表四是 2001 年的資料，如果再加上這幾年新增的 5 年 500 億，補助頂尖大學的數字更驚人。而弱勢學生較多的私立大學 (註 3)，補助少得可憐，我們的學生平均不到一萬。看到報載，一些台清交的學生說要凍漲學費，他們是否有想過自己是制度的受益者？
我們應該實施兩項不可能的改革，一是適當地提高學費，二是改採學分制收費。增加的經費用以增加對弱勢族群的獎助學金和改善教學環境，例如所需的軟硬體、輔導學生所需的資源等等 (註 4)。
幾年前的金融風暴，許多排名前面大學的學生不畢業，每年只需繳 5 萬的學雜費，繼續留在學校，卻享用數十萬的補助，顯然不合理 (註 5)。
在美國，任何時候重修都要收錢。在台灣，只有暑修或延畢要收錢。修我課程的學生，有的從來不來上課、有的來了只是坐在那裡不動、有的從來不交作業、或者考試可以帶書和講義卻考零分。詢問助教或老師，每次加學期總成績一分，至多五分，有不少人是從來不問我們的。如果不及格重修要花錢，學生自然會比較認真。前幾天有位同學來找我，問了一下，其父母都從事辛苦的工作，卻沒有任何理由就不來上課導致被退學。
如果學分提早修滿了，也就可以畢業，不用再繳學費，以鼓勵認真的學生。
單位學費成本提高了，學生也會更用心地要求老師提高教學品質 (註 2)。
我接觸到系上一些需要自食其力的打工同學，成績其實都還不錯。上課時也常常告訴同學，要多多地充實自己，培養專業，使用腦力賺錢，才能改善家中的經濟環境。以學校中的課程助教而言，每一個小時 150 元，如果沒有人前來詢問，還可以讀自己的書。以我們資管系最近畢業的學生為例，如果學會資訊系統的軟體開發，這些 (大學部) 專業工作的起薪大都是 30,000 到 42,000 元 (註 6)。有位女同學起薪三萬，表現良好，不到一年就調成 43,000 元。如果月薪四萬元，比起其他同學的二萬元，二年之間就多了 48 萬，可以還完助學貸款。所以，念大學值不值得完全看個人 (註 7)。
(註 1) 菲律賓的例子。美國是貴得嚇死人。歐陸的賦稅負擔率很高，所以學費便宜，參考我國大學學雜費與國際之比較 (doc 檔)；這幾年卻大幅度地高漲，例如法國的 Insead MBA 一年的費用是 5 萬 8 千歐元，或搜尋『英國大學學費』。
(註 2) 吳聰敏和駱明慶，『好老師，莫強求』。2001 年，教育部對台大、成大、淡江、和逢甲每位學生的平均補助分別為 25.22、19.99、2.36 和 2.36 萬元 (pdf 檔)
(註 3) 這學期全校學生辦理就學貸款三千多人，超過 3 成。導生班 50 人中，20 人辦理就學貸款，達 4 成；大部份是家境因素而需要辦學貸，有的則是希望靠自己的力量完成學業。
(註 4) 以 UT 為例，工學院有 7700 位學生，學院專屬的就業輔導室人員就有七位，可以幫助學生看履歷表和實施模擬面談等等 (www 1, 2)
(註 5) 5 萬元是學雜費，但是，根據現行的規定，通常不用繳到 5 萬元。許多排名前面學校的延畢生達二成。詳見自由電子報的報導『使用者付費 台大、政大及清大延畢生增收學雜費』
(註 6) 該起薪 42,000 元同學的成績普通：但是，畢業專題的指導教授和業界有顧問關係，同學畢專的訓練是該公司所需，所以起薪高。
(註 7) 除此之外，還有各式各樣免費的通識和外系課程、專業與職場相關的演講、訓練、和活動。至於大學教育是否應該強調就業是個有爭議的問題，參見書籍『大學教了沒』。
10/04/2012
培育 (軟體) 科技人才：美國公司的作法
因為美國大學生念資工的人太少，不足以應付工業界的需求
"There are likely to be 150,000 computing jobs opening up each year through 2020, according to an analysis of federal forecasts by the Association for Computing Machinery, a professional society for computing researchers. But despite the hoopla around start-up celebrities like Mark Zuckerberg of Facebook, fewer than 14,000 American students received undergraduate degrees in computer science last year, the Computing Research Association estimates. And the wider job market remains weak."
而且夠資格的 (高中) 老師也不足
"Finding capable computer science teachers is also hard. Few other industries are as good as the technology business in its ability to divert would-be educators into far more lucrative corporate jobs. Mr. Edouard graduated from the University of Florida in 2011 and considered enlisting in Teach for America, but he also had multiple offers from technology employers."
所以微軟、Google 和一些公司便派員工前往高中教授軟體開發，員工還有額外的津貼
"Most educators believe that for students to be excited about computer science, it is critical to introduce them to it at an early age. Yet support for the subject at cash-short K-12 schools has faded (1). In almost every state, computer science is taught as an elective, rather than a core requirement. The percentage of graduates who earned credits in high school computer science classes fell to 19 percent in 2009 from 25 percent in 1990, making it the only subject among science, technology, engineering and mathematics courses to experience such a drop, according to a report by the Education Department.
Google, for one, holds a programming summer camp for incoming ninth graders and underwrites an effort called CS4HS, in which high school teachers sharpen their computer science skills in workshops at local universities.
But Microsoft is sending its employees to the front lines, encouraging them to commit to teaching a high school computer science class for a full school year. Its engineers, who earn a small stipend for their classroom time, are in at least two hourlong classes a week and sometimes as many as five. Schools arrange the classes for first thing in the day to avoid interfering with the schedules of the engineers, who often do not arrive at Microsoft until the late morning.
This year, only 19 of the 110 teachers in the program are not Microsoft employees. In some cases, the program has thrown together volunteers from companies that spend a lot of their time beating each other up in the marketplace."
微軟的工程師和專業教師協同教學，以提高成效
"One of the biggest concerns about Microsoft’s effort is that most of its volunteers have little teaching experience. To comply with district licensing requirements and to help engineers with classroom challenges like managing unruly teenagers, a professional teacher is also in the room during lessons. One of the program’s tenets is that Microsoft engineers need to teach the teachers, alongside students, so that those instructors can eventually run an
engaging computer science class on their own."
原文 N. Wingfield, Fostering Tech Talent in Schools, New York Times, September 30, 2012
"There are likely to be 150,000 computing jobs opening up each year through 2020, according to an analysis of federal forecasts by the Association for Computing Machinery, a professional society for computing researchers. But despite the hoopla around start-up celebrities like Mark Zuckerberg of Facebook, fewer than 14,000 American students received undergraduate degrees in computer science last year, the Computing Research Association estimates. And the wider job market remains weak."
而且夠資格的 (高中) 老師也不足
"Finding capable computer science teachers is also hard. Few other industries are as good as the technology business in its ability to divert would-be educators into far more lucrative corporate jobs. Mr. Edouard graduated from the University of Florida in 2011 and considered enlisting in Teach for America, but he also had multiple offers from technology employers."
所以微軟、Google 和一些公司便派員工前往高中教授軟體開發，員工還有額外的津貼
"Most educators believe that for students to be excited about computer science, it is critical to introduce them to it at an early age. Yet support for the subject at cash-short K-12 schools has faded (1). In almost every state, computer science is taught as an elective, rather than a core requirement. The percentage of graduates who earned credits in high school computer science classes fell to 19 percent in 2009 from 25 percent in 1990, making it the only subject among science, technology, engineering and mathematics courses to experience such a drop, according to a report by the Education Department.
Google, for one, holds a programming summer camp for incoming ninth graders and underwrites an effort called CS4HS, in which high school teachers sharpen their computer science skills in workshops at local universities.
But Microsoft is sending its employees to the front lines, encouraging them to commit to teaching a high school computer science class for a full school year. Its engineers, who earn a small stipend for their classroom time, are in at least two hourlong classes a week and sometimes as many as five. Schools arrange the classes for first thing in the day to avoid interfering with the schedules of the engineers, who often do not arrive at Microsoft until the late morning.
This year, only 19 of the 110 teachers in the program are not Microsoft employees. In some cases, the program has thrown together volunteers from companies that spend a lot of their time beating each other up in the marketplace."
微軟的工程師和專業教師協同教學，以提高成效
"One of the biggest concerns about Microsoft’s effort is that most of its volunteers have little teaching experience. To comply with district licensing requirements and to help engineers with classroom challenges like managing unruly teenagers, a professional teacher is also in the room during lessons. One of the program’s tenets is that Microsoft engineers need to teach the teachers, alongside students, so that those instructors can eventually run an
engaging computer science class on their own."
原文 N. Wingfield, Fostering Tech Talent in Schools, New York Times, September 30, 2012
9/23/2012
同一桌中，餐廳如何一致地出菜
一家公司的作法
由西北大學商學院三位教授所執筆的 blog，針對報章或生活中生管的問題，有精湛的評論。在此篇中，引述Wired 雜誌，說明 Lyfe Kitchen 如何使用 RFID、流程分析、和軟體以一致地出菜；可供臺灣的餐廳參考
"Unlike the sit-down bistros where gourmet food is generally prepared and served, Lyfe sees each brussels sprout as merely a cog in a vast clockwork, a system that is set into motion as a customer approaches the counter, gives their name, and places an order. Once that order is sent electronically to the kitchen, a cashier hands the customer a coaster. RFID strips beneath every table pick up the signal from the coaster and send it back to the kitchen. That’s how the runner — someone other than the person who took your order — knows where you are sitting, what you have ordered, and your name.
Now that the order has gone into the kitchen, the software-based cooking system kicks in. It’s smart enough to separate the elements of your order and send each of them to the monitor that hangs above the relevant food-prep station. The flatbread maker sees flatbread orders; the pantry chef, who makes all the salads and desserts, sees the salad order; the rotisseur at the broiler station—you get the picture. So everything everyone needs to cook shows up in a queue, and the chefs each hit a plastic button beneath the screen to signal that they have begun. When they’re done, they press a button that “bumps” the food order to the “quarterback,” who gathers the finished product and puts it on a plate with all the other stuff you want to eat.
Take that free-range chicken dish. Just as in a high-end kitchen in Chicago or San Francisco, the roisseur here cooks the chicken breast at a different spot from the one the legumier uses to saute the brussels sprouts, squash, and cranberries. Of course, the two elements—the chicken and the vegetables—have different cooking times. Lyfe’s software is ready for such complication, though, and sequences the timing. In the case of salmon and potatoes, say, the fish takes five minutes while the tubers take two. So the roisseur receives the salmon order three minutes before the potatoes order appears on the legumier’s screen. Same for the chicken and brussels sprouts. It’s all finished at the same time and plated together by the quarterback. Then the runner picks it up, heads to your table, and says, “Hi, Fred.”
由西北大學商學院三位教授所執筆的 blog，針對報章或生活中生管的問題，有精湛的評論。在此篇中，引述Wired 雜誌，說明 Lyfe Kitchen 如何使用 RFID、流程分析、和軟體以一致地出菜；可供臺灣的餐廳參考
"Unlike the sit-down bistros where gourmet food is generally prepared and served, Lyfe sees each brussels sprout as merely a cog in a vast clockwork, a system that is set into motion as a customer approaches the counter, gives their name, and places an order. Once that order is sent electronically to the kitchen, a cashier hands the customer a coaster. RFID strips beneath every table pick up the signal from the coaster and send it back to the kitchen. That’s how the runner — someone other than the person who took your order — knows where you are sitting, what you have ordered, and your name.
Now that the order has gone into the kitchen, the software-based cooking system kicks in. It’s smart enough to separate the elements of your order and send each of them to the monitor that hangs above the relevant food-prep station. The flatbread maker sees flatbread orders; the pantry chef, who makes all the salads and desserts, sees the salad order; the rotisseur at the broiler station—you get the picture. So everything everyone needs to cook shows up in a queue, and the chefs each hit a plastic button beneath the screen to signal that they have begun. When they’re done, they press a button that “bumps” the food order to the “quarterback,” who gathers the finished product and puts it on a plate with all the other stuff you want to eat.
Take that free-range chicken dish. Just as in a high-end kitchen in Chicago or San Francisco, the roisseur here cooks the chicken breast at a different spot from the one the legumier uses to saute the brussels sprouts, squash, and cranberries. Of course, the two elements—the chicken and the vegetables—have different cooking times. Lyfe’s software is ready for such complication, though, and sequences the timing. In the case of salmon and potatoes, say, the fish takes five minutes while the tubers take two. So the roisseur receives the salmon order three minutes before the potatoes order appears on the legumier’s screen. Same for the chicken and brussels sprouts. It’s all finished at the same time and plated together by the quarterback. Then the runner picks it up, heads to your table, and says, “Hi, Fred.”
9/08/2012
自強號跑輸復興號
花 441 元坐自強號，竟然比 284 元的復興號慢，這又貴又慢的自強號，讓乘客坐得好悶，點進去龜速的 224 次自強號查看，從樹林到花蓮，數一數，竟然狂停了 19 站，684 次的復興號開到台東，也不過才停 10 站。
北花自強號除了「跑輸」復興號，同樣是自強號，快慢也差很多。最快的太魯閣號只須兩小時，最慢的竟要三小時十三分。台鐵解釋，六八四、六八五次主要是為疏解北東自強號的需求，才會推出停站少、便宜票價的復興號吸引旅客；上午七時卅分從台北出發的二七二次自強號，沿途有很高的通勤需求，加上同時段替代性列車少，才會停站較多。台鐵說，北花自強號平均約兩小時四十分，「並沒有真的差很多」。
鐵路排班是一個複雜的問題，必須考慮駕駛員、火車數量、旅客需求、運輸時段、和兩站間的運輸時間，不適合憑著直覺來處理。但是，非常適合使用作業研究來處理，以時間的限制式為例，根據不同時段旅客上下車的速度，使用適當的統計分析和加上緩衝時間，可以計算每班次的時間。根據顧客需求分析，得知自強號必須比復興號快多少時間，形成另外一組限制式。同樣是自強號，其時間差距也不可以太大，形成另外一組限制式。
以荷蘭的例子為例 (註 1)，為了因應歐盟自由化的要求，政府只負責維護鐵路系統，旅客服務由數家公司經營：
Until 1995, Netherlands Railways (Nederlandse Spoorwegen, or NS) was a state-owned company, operating passenger and freight services and building and maintaining the railway infrastructure. Because of European Union regulations and liberalization of the railway market, NS was split into several companies during the period 1995–2002. The state maintains ownership of the infrastructure because of its strategic value. ProRail, a nonprofit organization owned by the state, is responsible for maintaining and allocating the infrastructure. Several companies, including NS, now operate the passenger services; there are also a number of freight operators.
使用作業研究來處理後，其成效是增加了旅客搭乘人數和公司收入：
The new resource schedules and the increased number of passengers resulted in an additional annual profit of E40 million ($60 million); the additional revenues generated approximately E10 million of this profit. We expect this profit to increase to E70 million ($105 million) annually in the coming years.
(註 1) Leo Kroon, Dennis Huisman, Erwin Abbink, Pieter-Jan Fioole, Matteo Fischetti, Gábor Maróti, Alexander Schrijver, Adri Steenbeek, and Roelof Ybema, The New Dutch Timetable: The OR Revolution, Interfaces, 2009 39: 6-17. (2008 Franz Edelman Award)
(註 2) 一般而言，可能還需要資工 (管) 的學 (碩) 士，以處理資訊方面的問題。
9/05/2012
計算機程式補考後的人生 (1/2)：世事難料
高中畢業的我沒有接觸過電腦，大學考上了電機系，一上和一下分別 2 學分的計算機程式是必修，學的是 Fortran。
當時流行 DOS，應用程式或網路應用很少。學校電腦是終端機連接到共用大型主機，面對只有命令提示符號的終端機輸入程式碼，心中充滿了疑惑，也提不起興趣。詢問班上學習良好的同學，似乎也無法幫我找出學習困難的原因。
8/19/2012
作業管理有趣的論文：九種可能
這是 Prof. Cachon 當選 INFORMS MSOM Fellow 的演講 (註 1)， 說明何謂有影響力的 (Impact) 論文：
什麼是有趣的 (Interesting) 論文？令人意想不到的結果。作者使用論文說明了九種可能
在 What Was Expected to Be a Small Effect Is Really a Large Effect 中， 作者舉了一個不屬於作業管理的例子 (註 3)， 研究使用 RFID 標籤 (類似臺灣的 eTag ) 快速通過收費站所降低廢氣的影響， 有驚人的發現
(註 1) Cachon, G. 2012. What is interesting in operations management?, Manufacturing & Service Operations Management. 14(2). 166-169. (in pdf)
(註 2) 作者沒有舉例說明。 一種可能是 Management Science 所列舉的 50 Most Influential Papers (1954-2004)。
(註 3) Currie, J., R. Walker. 2011. Traffic congestion and infant health: Evidence from E-ZPass. Amer.Econom . J.: Appl . Econom . 3(1) 65–90.
Impact = Interesting * Important什麼是重要的 (Important) 論文？有用的 (useful) 結果，例如產生了更多的研究或是解決社會的問題 (註 2)。
什麼是有趣的 (Interesting) 論文？令人意想不到的結果。作者使用論文說明了九種可能
- What Was Thought to Be Exogenous Is Really Endogenous
- What Was Thought to Be Complex Is Really
Simple - What Was Thought to Be Simple Is Really Complex
- What Was Expected to Be a Small Effect Is Really a Large Effect
- What Was Thought to Be a Large Effect Is Really a Small Effect
- What Was Thought to Be a Large Effect Is Really Much
Larger - What Was Thought to Be Easy Is Really Hard
- What Was Assumed to Not Be a Problem Is Really a Problem
- What Should Improve Performance Really Harms
Performance
在 What Was Expected to Be a Small Effect Is Really a Large Effect 中
One could easily assume (as I would) that this effect, if it exists, would be too small to measure, but they find a substantial effect: premature births decreased by 10.8% and low-weight births decreased by 11.8%.作者使用如此有趣的方式來看待研究，令人讚嘆。其教學和研究能力一樣傑出，令人敬佩。
(註 1) Cachon, G. 2012. What is interesting in operations management?, Manufacturing & Service Operations Management. 14(2). 166-169. (
(註 2) 作者沒有舉例說明
(註 3) Currie, J., R. Walker. 2011. Traffic congestion and infant health: Evidence from E-ZPass. Amer.
7/29/2012
邁向頂尖之路：哈佛商學院的人生習題
成長於英國和畢業於牛津大學的記者，前往哈佛商學院完成 MBA學位後，寫下的書
第一個重點是個案教學法，透過一年級的十門必修課和二年級的選修課，剖析如何分析和處理不確定的商業決策 (1)。
第二個則是許多的觀察和省思，包括 (哈佛) 商學院的教育、工作和家庭、資本主義、生命的價值、大企業不為人知的面向等等。
(1) Premier Case Collection 列了 450 個最熱門的個案。
第一個重點是個案教學法，透過一年級的十門必修課和二年級的選修課，剖析如何分析和處理不確定的商業決策 (1)。
第二個則是許多的觀察和省思，包括 (哈佛) 商學院的教育、工作和家庭、資本主義、生命的價值、大企業不為人知的面向等等。
(1) Premier Case Collection 列了 450 個最熱門的個案。
7/04/2012
Marx 或 Smith 是對的？
女兒學校的國二社會科 (social studies) 期中考考題 (1)，本身是博士的美國老師出了八題，女兒選擇回答的問題是
Does what's happening in the world today seem to suggest Marx was right or Smith? (2)
根據老師上課的補充資料，女兒的 (原始) 回答如下
Marx and Smith have two different alternative views of society and government. Marx believes in that society will eventually develop into communism, while Smith believes in a capitalism society. They both are connect on some part of history. Countries such as Russia and China did become communist, but are now receding back to the right. Most countries are moving to the right in history.
Take China as an example. It is today a socialist country. The KMT used to have an industrialized republic country but the workers revolted. Next, this party rise up and ruled China as communism. Up till this point, it fits Marx's view of history. Since then, people have become rich with its own business, so this is actually Smith's view of history. Still, the Chinese government today owns quite a few businesses. China did fit Marx's view of history but have made a turning point to the right and become a socialist country. If China keeps on moving, it will arrive to a capitalist country.
A different example is England, a capitalist country. England first industrialized then became a free market country. In England "greed is good", the market decides the price of goods and who to survive. This is Smith’s view of society which is related to nature selection. In England, the workers never revolted and remains a free market up till today. England's case is naturally similar to a lot of countries such as US, Germany and France. Also a capitalist country has no government involving in it.
Lastly, a communist country is impossible while a total free market is also impossible. Marx view the final stage of communism as no government. This is impossible since people can't live equally with no government at all. Smith's view is also impossible since the government will always be involved. In America, the government bailed out the banks. That is involvement. While in almost all the countries, the government will own few large companies. A total free market and a total communism is impossible, but the world today seems to lean more over to Smith's side.
In conclusion, history have evolved both ways, but every country that has reach Marx's view of communism will eventually recede right to Smith's view. Most countries in the world today are capitalist economy will little government involvement. This is a mix of two but more emphasis of Smith's view. The justified answer is that history lies more to Smith's view compared to Marx since every communist country will eventually fall. The economy in the world today is mostly Smith's view since government usually owns a few businesses only.
(1) 課本 B.J. Armento, etc., Across the Centuries, Houghton Mifflin, 2003, 558 pages.
(2) Karl Marx (卡爾馬克思) 和 Adam Smith (亞當史密斯)
Does what's happening in the world today seem to suggest Marx was right or Smith? (2)
根據老師上課的補充資料，女兒的 (原始) 回答如下
Marx and Smith have two different alternative views of society and government. Marx believes in that society will eventually develop into communism, while Smith believes in a capitalism society. They both are connect on some part of history. Countries such as Russia and China did become communist, but are now receding back to the right. Most countries are moving to the right in history.
Take China as an example. It is today a socialist country. The KMT used to have an industrialized republic country but the workers revolted. Next, this party rise up and ruled China as communism. Up till this point, it fits Marx's view of history. Since then, people have become rich with its own business, so this is actually Smith's view of history. Still, the Chinese government today owns quite a few businesses. China did fit Marx's view of history but have made a turning point to the right and become a socialist country. If China keeps on moving, it will arrive to a capitalist country.
A different example is England, a capitalist country. England first industrialized then became a free market country. In England "greed is good", the market decides the price of goods and who to survive. This is Smith’s view of society which is related to nature selection. In England, the workers never revolted and remains a free market up till today. England's case is naturally similar to a lot of countries such as US, Germany and France. Also a capitalist country has no government involving in it.
Lastly, a communist country is impossible while a total free market is also impossible. Marx view the final stage of communism as no government. This is impossible since people can't live equally with no government at all. Smith's view is also impossible since the government will always be involved. In America, the government bailed out the banks. That is involvement. While in almost all the countries, the government will own few large companies. A total free market and a total communism is impossible, but the world today seems to lean more over to Smith's side.
In conclusion, history have evolved both ways, but every country that has reach Marx's view of communism will eventually recede right to Smith's view. Most countries in the world today are capitalist economy will little government involvement. This is a mix of two but more emphasis of Smith's view. The justified answer is that history lies more to Smith's view compared to Marx since every communist country will eventually fall. The economy in the world today is mostly Smith's view since government usually owns a few businesses only.
(1) 課本 B.J. Armento, etc., Across the Centuries, Houghton Mifflin, 2003, 558 pages.
(2) Karl Marx (卡爾馬克思) 和 Adam Smith (亞當史密斯)
6/18/2012
企業如何達成分析決策的方法：DELTA
本書 (1) 延續前一本書 Competing on Analytics，說明如何達成該目標的五個步驟，分別為
Data: The prerequisite for everthing analytical
Enterprise: Integrting across organizational silos
Leadership: The deciding DELTA factor
Targets: Picking your spots for analytics
Analysis: Managing Scarce and valuable talent
第一章根據 Competing on Analytics 中的架構，將企業使用分析能力的狀況分成五個階層，分別為
Analytically impaired
Localized analytics
Analytical aspirations
Analytical companies
Analytical competitors
附錄中的 2 頁表格針對如何進階到高一階層的工作，分別說明 DELTA 中必須準備的事項。
書的第二部份則是說明如何維持該狀態，分別為
Embed analytics in business processes
Build an analytical culture
Review your business comprehensively
Meeting challenges along the way
Toward more analytical decisions and better results
達成的關鍵？
The key challenges are no longer technological or analytical; they have more to do with simply making managers familiar with the concepts and the process. (2)
(1) T.H. Davenport, J.G. Harris, and R. Morison, Analytics at Work: Smarter Decisions, Better Results, Harvard Business Press, 2010.
(2) T.H. Davenport, How to Design Smart Business Experiment, Harvard Business Review, 2009.
Data: The prerequisite for everthing analytical
Enterprise: Integrting across organizational silos
Leadership: The deciding DELTA factor
Targets: Picking your spots for analytics
Analysis: Managing Scarce and valuable talent
第一章根據 Competing on Analytics 中的架構，將企業使用分析能力的狀況分成五個階層，分別為
Analytically impaired
Localized analytics
Analytical aspirations
Analytical companies
Analytical competitors
附錄中的 2 頁表格針對如何進階到高一階層的工作，分別說明 DELTA 中必須準備的事項。
書的第二部份則是說明如何維持該狀態，分別為
Embed analytics in business processes
Build an analytical culture
Review your business comprehensively
Meeting challenges along the way
Toward more analytical decisions and better results
達成的關鍵？
The key challenges are no longer technological or analytical; they have more to do with simply making managers familiar with the concepts and the process. (2)
(1) T.H. Davenport, J.G. Harris, and R. Morison, Analytics at Work: Smarter Decisions, Better Results, Harvard Business Press, 2010.
(2) T.H. Davenport, How to Design Smart Business Experiment, Harvard Business Review, 2009.