亞馬遜前首席科學家 Andreas Weigend：展望大數據與未來

最新 07-31

這是《黑獅全球》欄目的第二篇文章。

黑獅全球——我們將在全球範圍內，邀請一些頂尖的企業家、行業專家、創業新銳來黑獅國際欄目訪談，構建1000位全球化的CEO會員網路！

本期來《黑獅全球》做客的是來自美國的大數據鼻祖Andreas Weigend，他曾擔任亞馬遜首席科學家、阿里巴巴全球顧問等職務。Andreas Weigend 還是多家創業公司、世界經濟論壇的顧問。

採訪 |羅旭、姚尚文撰文 |丁雪聰

Andreas Weigend在擔任亞馬遜首席科學家期間，幫助亞馬遜創建了數據戰略和構建了以客戶為中心的體系。在此之後他成為了阿里巴巴集團全球顧問，併兼顧企業技術諮詢，數據相關研究及寫作出書等職務。

Andreas同時還是諸多創業公司的顧問，合作過的公司包括高盛、漢莎航空、SAP（思愛普）、SingTel（新加坡電信）、湯森路透、百思買、以及世界經濟論壇等。

近年來，中國科技公司在全球化的道路上發展迅猛，湧現出多家獨角獸公司。黑獅全球欄目組連線Andreas Weigend，與他一起暢談了他過去的職業經歷、對大數據未來發展的展望，以及表達了對中國科技公司全球化的一些看法。

工作經歷

黑獅：In your job as a Chief Scientist in amazon, can you talk about some interesting experience in past career?

您曾在亞馬遜公司任職首席科學家，能否聊聊過去職業生涯上一些有趣的經歷？

Andreas Weigend：When I was at home that night, the phone rang and I was asked if I would like to work at amazon with Jeff Bezos. Actually, I was not interested in entering a big company before, because my thinking on career is different from that of most people. But still, I went to amazon and I was honored to work with bezos.

I have spent most of the time studying big data during the last ten years. I"ve also written a book, Data for the People. At the same time, I was working with some large corporations, medium-size and small businesses to help them building up strategy base on data. These things seem very fragmented and irrelevant, but the core is: I only do what I want to do and remain curious. What I really care about is human, including human behavior that we can understand through data.

The global total now doubles every year and a half, and the process of data collection used to take 30 years can now be done in a second. The point is not just speed, but the ubiquity of data: wherever you go, whatever you do, you have left data that can be collected and analyzed. My book is not called Data of the People, Data by the People, but Data for the People, which means I hope Data can be better used to service people.

那晚我在家，電話響起，有人問我願不願意進入亞馬遜與傑夫·貝佐斯（Jeff Bezos，亞馬遜創始人及現任CEO）一起工作。其實我之前對進入大公司不感興趣，因為我對職業的思考跟大多數人不同。但後來我還是進入了亞馬遜，也很榮幸與貝佐斯工作。

過去十年我都在大數據領域摸索：我也寫了一本書——《Data for the People》；同時還與眾多大公司、中小企業與新創團隊合作。這些事情看起來很碎片化，似乎互不相關，但核心在於：我只做自己想做的, 保持好奇心。而我真正關心和好奇的是人，包括我們能夠透過數據去了解到的人類行為。

如今全球的總數據量每一年半就會翻倍，過去搜集過程需要三十年的數據，現在只要一秒，這些轉變充滿了戲劇性。重點並不只是速度，更是數據的無所不在：你去過的所有地方、你做過的任何事，都留下了能被他人所分析的數據。我的書不叫Data of the People、Data by the People，而是Data for the People，代表的含義是我希望數據能更好地為人所用。

Andreas Weigend （左二）

黑獅全球特邀記者羅旭（左四）

黑獅：Then you came to China and served as a global adviser to jack ma. What was your experience and what was your main job?

後來您又來到了中國，擔任馬雲的數據顧問，這是一段怎樣的經歷，主要做些什麼工作呢？

Andreas Weigend：Jack used to hire me as his "data coach" because he wanted to understand data science. He was so obsessed with Data that he even said don"t call him "JackMa," but "Data Ma."

There are many ways to do data science: one is to write a program that will be interrupted in a few minutes by just a few people, but may be enough to come up with good answers. Or a team of production engineers, software engineers, to solve various Exceptionhandling problems. Different people have different preferences, and I prefer to cooperate with the latter.

As a data consultant of alibaba, I have been committed to the future development and application of data research. The speed of technological development is always beyond our expectation. We can therefore predict that data technology will also have better development and application in the future, even subverting our lifestyle.

之前馬雲僱用我作為他的「數據教練」，因為他想要了解數據科學。他對數據是如此痴迷，甚至還說不要稱呼他為JackMa，而是「Data Ma」。

做數據科學有很多種方式：一種是只靠幾個人寫出幾分鐘就會中斷的程序，但也許已經足夠分析出好的答案；或是一班生產工程師、軟體工程師解決各式「異常情況處理」問題。不同人會有不同的偏好，而我比較喜歡與後者合作。

我在擔任阿里數據顧問期間，一直致力於數據在未來的發展及應用的研究，科技發展的速度總是出乎我們意料。可以預見，數據科技在將來也會有著更好的發展和應用，甚至於顛覆和改變我們的生活。

關於大數據

黑獅：You keep mentioning the importance of data, so what does data mean in your definition?

您不斷提到Data（數據）的重要性，那麼在您的定義里，Data到底意味著什麼呢？

Andreas Weigend：In fact, the value of data is that it represents who we are -- we are made up of every decision we make, and gathering a lot of decision data allows us to make better decisions. For example, when you use a Google map, all users share their data, and Google helps you decide whether to take a detour by figuring out which road is blocked and which is smooth. People used to build complex traffic models and supercomputers to calculate them, but now they don"t need to, all they need to do is collecting data. In other words, we are not predicting, but just observing, and then we find the best decision according to the experience of observation.

事實上，數據的價值在於它表示了我們是誰——我們是由自己所做的每一個決定所組成，搜集大量決策數據可以讓我們做出更好的決定。舉個例子，當你使用谷歌地圖的時候，所有用戶分享他的數據，再由谷歌分析出哪條路塞車、哪條路順暢，進而幫助你決定是否要繞路。過去有人建立起複雜的交通模型以及超級計算機來計算，但現在不需要，只要搜集數據。換句話說，我們不是預測，只是觀察, 然後從累積的觀察經驗中找出最好的決策。

黑獅：Recently, the European Union has just issued the GDPR privacy policy. How do you think about the balance between data and privacy ?

最近歐盟出台了GDPR隱私條款，你覺得應該怎樣拿捏數據與隱私之間的平衡？

Andreas Weigend：We live in a post-privacy economy, and when privacy becomes a resource, how do we make money from it? In other words, how valuable is your data? You care so much about your privacy, but you give it to Google. It is almost impossible for us to give up sharing data, therefore people have invented a way to protect our privacy while sharing data.

Assuming that the government is considering whether to subsidize treatment for drug users, if drug users are directly investigated, they may choose to lie because of fear of arrest. How can we understand the composition of drug users without intruding on their privacy?

Those wise anthropologists come up with a method: you ask a person a question and ask them to lose a coin, and only they know the result. If he turns heads, answer honestly; If the opposite happen, the answer is yes. There are two possibilities for the person who says "yes" : either the person does use drugs, or he just turns the other way. Interestingly, when the number of samples is large enough, the probability of a "yes" person actually being a drug addict is close to 50 percent, so we divide the number of "yes" people by 2 to get a roughly accurate answer. I call this differential privacy: everyone can share their data without baring the risk of disclosure.

我們生活在「後隱私經濟」，當隱私成為一種資源，我們如何從中賺錢？也就是說，你的數據有多少價值？你這麼在乎你的隱私，卻把隱私給了谷歌。但人們不能因此放棄分享數據，而是發明了兩全其美的方式。

假設政府在考慮是否該補助吸毒者的治療措施，如果直接調查吸毒者，吸毒者會因為害怕被逮捕而選擇說謊。我們能夠如何了解吸毒的人口組成，而不去侵犯個人隱私？

聰明的人類學家想到一個方法：你問受訪者一個問題，並讓他丟硬幣，只有他知道結果。如果他翻到正面，就誠實回答；如果反面，則回答「是」。這樣回答「是」的人有兩種可能，要麼此人的確吸毒，要麼他就只是翻到反面。有趣的在於，當測試數量足夠多時，回答「是」的人當中的確為吸毒者的概率將接近50%，於是我們將回答「是」的人數除以2就能獲得一個大概準確的答案。我將此稱為「差別隱私【differential privacy】」：人人可以分享自己的數據，並免去隱私被揭露的危害。

中國公司全球化

黑獅：How are the products of Chinese Internet companies performing compared with the leading Internet products in the us?

與美國行業領先的互聯網產品相比，中國互聯網公司的產品表現如何？

Andreas Weigend：Chinese tech companies are certainly growing fast, but they"re not doing enough for the foreign user experience. Chinese consumers are more easy-satisfied than American and German consumers.

To compare the performance of Chinese and American technology companies, we can only objectively compare their quality of use, such as the searching results of Baidu and Google. Many have said that if the Chinese government did not support local companies by blocking Google"s services in China, Google would dominate the Chinese market as it did in the US and Europe. I admire Larry Page, but I also respect Robin Li, the CEO of baidu. What really matters is not the differences between China and the United States, no matter westerners or Chinese, people"s lives is completely different compare to a decade ago. We should focus on those things in common: whether it"s baidu and Google, Didi and Uber, WeChat and Facebook, they have fundamentally changed the way hundreds of millions of people communicates.

中國科技公司的確發展很快，但目前對外國用戶的用戶體驗做的還不夠好。相較於美國和德國，中國的消費者更容易滿足。

要比較中美科技企業的好壞，只能客觀的比較其使用質量，例如對比百度與谷歌的搜尋結果。很多人說過，如果中國政府沒有透過阻撓谷歌在中國提供服務來支持本土公司，谷歌會主宰中國市場，如同他在美國與歐洲的結果一樣。不過我既欣賞Larry Page（Google創始人），也喜歡百度的CEO李彥宏，他們都是我十分尊重的人。真正重要的並不是中美之間差異，而是不論西方人或中國人，他們的生活都跟十年前徹底不同了。我們應該關注那些共通點：不管是百度與谷歌、滴滴與優步、微信與臉書，都從根本上轉變了數億人溝通的方式。

黑獅：So do you think it would be hard for Chinese internet companies to do better in other markets around the world ?

所以在你看來中國互聯網科技公司打入全球其他戰場的難度大嗎？

Andreas Weigend：For now, Chinese software is hard to compete with U.S. companies without government protection, but China"s laws on information are so different from those of the rest of the world that Chinese companies would have to choose their own path to growth.

Although China"s Internet companies lag behind the United States in terms of globalization, the speed of its development is remarkable. There is an interesting question: which part in the world other than China where Chinese software is doing better than American companies? I would say, that could be Tencent"s WeChat. Many of the technologies developed by WeChat have clearly been copied by Facebook. Perhaps this is a beginning of the rise of the global presence of Chinese technology companies.

就目前階段而言，中國軟體在沒有政府的保護下，是很難和美國公司競爭的。但中國關於信息的法律與世界上其他國家有很大的不同，因此中國公司只能選擇適合自己的道路去發展。

羅旭姚尚文丁雪聰是黑獅全球特邀記者。Andreas Weigend是美國大數據鼻祖，曾擔任亞馬遜首席科學家、阿里巴巴全球顧問等職務。

美中跨境電商協會（總部矽谷）

Association of Crossborder E-commerce (ACE)

成立啦！