溝通至上 你會做網上性格分析嗎

“Lucy is a gifted communicator, makes occasional gut decisions against logic, and prefers big ideas over details.”

“露西有出色的溝通能力，偶爾會憑直覺、而非邏輯做出決定，她喜歡奇思妙想，而不愛關注細節。”

As a description of me, this is pretty good. I have spent my working life communicating, so I jolly well ought to be good at it by now. I have just made a decision in my private life which I know is not only illogical but financially catastrophic, yet I’m doggedly going with my gut. And as for details, unless they suit my argument, I have no fondness for them at all.

這段性格分析對我的描述挺恰當。在我的職業生涯中，我一直在不停地溝通，如今我當然應該很善於溝通。在我的個人生活方面，我剛剛做了一個決定，我知道這個決定不僅不合理，在財務上還將是災難性的，但我固執地憑着直覺決定那麼做。至於細節，除非它們符合我的觀點，否則我一點也不喜歡它們。

Yet what is so disturbing about the above character analysis is that it does not come from someone who knows me; in fact, it doesn’t come from a person at all. It was thrown together in three seconds flat by an algorithm developed by some computer scientists who have worked out how to trawl the internet for all public information about a person and turn it into a potted portrait of a personality.

以上的性格分析真正讓人不安的地方是，它並非來自某個認識我的人；事實上，這段分析根本不是人做出的，而是由一個算法在短短3秒鐘裏得出的。開發這個算法的計算機科學家們研究出了一種方法，能夠蒐集互聯網上關於某個人的全部公共信息，然後利用這些信息歸納出這個人的性格特徵。

Having let analyse me, I then typed in my colleagues’ names and found they were summed up uncannily well too. A fellow columnist, whom I know to be impatient, bold and creative, was judged to be just that by Crystal. Some of its verdicts were a bit off the mark — yet in each case, the app tells you how confident it is about being right; the cases in which accuracy was poor were mostly those where there wasn’t much data to go on.

用分析了我自己以後，我又輸入了一些同事的名字，發現對他們的性格的總結也出奇地準確。一位寫專欄的同事是個不怎麼有耐心、大膽、有創意的人，Crystal對他的判斷也正是如此。有些判斷與實際有些出入，但在每個案例中，這款應用都讓你覺得它很自信它的判斷是對的；判斷不夠準確大多是因爲沒有足夠的數據可供分析。

This is all good fun, but the main selling point of Crystal is not to provide uproarious entertainment. It is to help you communicate better. You simply link the app to Gmail and every time you start writing a message it tells you what sort of approach will go down best with the recipient.

這很有意思，但Crystal的主要賣點不是讓人捧腹大笑，而是爲了幫你更好地溝通。你只需要把這款應用連接到你的Gmail上，每次你開始寫信的時候，它就會告訴你什麼樣的風格最適合你的收件人。

I have just emailed Drew D’Agostino, the founder of Crystal Projects. “Be brief,” warned a little green button at the bottom of my screen, telling me that Drew doesn’t like wordy. Had I been writing to someone else, it might have advised: be casual or be formal.

我剛剛給Crytal Projects的創始人德魯•達戈斯蒂諾(Drew D’Agostino)寫了一封郵件，屏幕下方的綠色小按鈕警告我要“簡短”，提示我德魯不喜歡囉嗦。如果我給其他人寫信，它給出的建議可能是讓我寫得隨意或者正式一些。

According to Mr D’Agostino, Crystal is the biggest improvement to email since the spell check. But so far tetchy bloggers are not impressed. It’s been called the “stalking app” and “creepy” and “sinister”. Some people have protested that it doesn’t feel good to know perfect strangers are looking you up without your permission, and forming instant views of you.

據達戈斯蒂諾說，Crystal是繼拼寫檢查功能之後對電子郵件的最大改進。但敏感易怒的博主們迄今並未被打動。人們把Crystal叫做“跟蹤應用”，說它“可怕”、“邪惡”。一些人說，完全陌生的人不經你的允許查找你的信息，並立即得出有關你是個什麼樣的人的觀點，這種想法讓人感到不舒服。

Yet none of this strikes me as particularly creepy — all the data are already in the public domain. And I think I’d rather that decisions about me were made based on some sort of system rather than on hunch, prejudice and ignorance, as at present. Unlike most apps, Crystal solves a genuine problem. There are no common rules on email. Every time we write an email we are in the dark as to what is going to work best.

然而我並不覺得以上所說的這些特別可怕——這些信息本來就是公開的——而且我認爲，與其像現在這樣，人們基於直覺、偏見和無知做出關於我的決定，我寧願這些決定是基於某種系統做出的。和大多數應用不同，Crystal解決了一個真正的難題。寫電子郵件沒有統一的規則。我們每次寫電子郵件的時候，都不知道哪種書寫風格效果最好。

You could say that a post-Crystal world, in which all messages arriving in my inbox were written in the same prescribed style, would be a dull one. Yet it couldn’t be as dull as the current arrangement whereby I read dozens of emails every day that begin with windy small talk like “I hope this email finds you well”. If people knew how much I hated that, they would desist. That would save time for them and save me from irritation. Win, win.

你也許會說，在一個“後Crystal”的世界裏，我的收件箱裏每一封郵件都使用軟件所建議的統一文風，這聽起來很無趣。但怎麼也不會像現在這麼無趣吧。現在我每天都要閱讀幾十封用空洞的寒暄開頭的郵件，比如“當你看到這封信的時候，希望你一切都好”。如果人們知道我多麼討厭這種套話，他們就不會這麼寫了。這不僅能節省他們的時間，我也不用惱火了。雙贏局面。

A more philosophical complaint is that, if all the judging of character were done by machines, we would lose our own — more nuanced — ways of assessing people. And because we are suggestible, we don’t question the apps’ results, even when they are based on not much information. Confirmation bias, and all that.

還有一種更偏哲學層面的抱怨，如果對性格的判斷全都是由機器完成的，我們就不再會用我們自己的更細緻入微的方法去評判人了。因爲我們很容易輕信，我們將不會質疑這款應用得出的結果，即使這些結果是在信息不多的情況下得出的。這也就是證實傾向(Confirmation bias)之類的問題。

There may be something in this, but my main objection to Crystal is more basic. It doesn’t work well enough.

這麼說或許有些道理，但我反對Crystal的主要理由更加簡單。它的效果還不夠好。

I should be the easiest person in the world for Crystal to analyse given that I have published a personal column every week for 21 years. Even though the app has more or less got my character right, its suggestions on how to approach me are worse than useless. “Use emoticons,” it begins. Is this a joke? I have never knowingly used an emoticon in my life, and instantly knock a couple of points off the IQ of anyone who deploys them. Even more outrageously, it says that I don’t mind if people are late. Again: could not be more wrong. I am obsessively punctual.

對於Crystal來說，我應該是世界上最容易分析的人了，因爲我21年來每週都發表一篇個人專欄。就算這款應用對我性格的判斷或多或少是對的，它在如何給我寫郵件方面給出的建議卻有害無益。建議一開始就說“使用表情符號”。這不是開玩笑吧？我這輩子從來沒有有意地使用過一個表情符號，看到任何使用表情符號的人，我都會立即在心裏給他們的智商評價扣掉幾分。更讓人無法容忍的是，建議中說我在約會中不介意對方遲到。這也大錯特錯。我是個極其守時的人。

Emoticonphobia and obsessive timekeeping are not things that an algorithm needs to deduce about me. They are phobias that I have written about explicitly and often. Thus my problem with Crystal is not that it is stalking me, but that it doesn’t stalk nearly enthusiastically enough.

我對錶情符號的反感和對守時的執着不需要一個算法來推斷。我經常公開地在文章裏寫到這些特質。所以我不滿Crystal不是因爲它跟蹤我，而是因爲它對我的跟蹤還不夠“狂熱”。

In my brief email to Drew, I asked about accuracy. “Hi Lucy. Thanks for reaching out,” he emailed back, thus committing two email bloopers before he had even got going. He then reassured me that accuracy is getting better, and soon users will be able to correct mistakes and supply preferences themselves.

在我寫給德魯的簡短的郵件中，我問到了準確性的問題。“你好露西。感謝來信，”他回郵件說。在還沒進入正文之前，他就已經犯了兩個錯誤。接着他讓我放心，說準確性會越來越好，很快用戶就能自行修正建議中的錯誤，補充自己的喜好。

In the meantime here is my character report on Crystal: great idea; must try harder.

以下是我對Crystal出具的性格分析報告：想法不錯；但還得加把勁。