好萊塢或將步入寒冬

尼克·比爾頓

Hollywood may be the last creative industry not to be hollowed out1 by technology. What happens when Silicon Valley2 starts replacing screenwriters with bots， and your favorite actress with C.G.I.3？ In some corners of Los Angeles， the future is already here.好萊塢或許是最后一個尚未被科技顛覆的創意產業了。假如硅谷開始用機器人來替代編劇，用電腦成像來替代你最喜歡的女演員，那會發生什么？在洛杉磯的某些角落，這種未來已經到來。

Shortly after midnight on December 8， 1962， on West 43rd Street in Manhattan， right at the epicenter of the world， hundreds of people who worked in the newspaper industry walked out of the New York Times building and into the cold streets to begin a strike that would last 114 days. About 17，000 newspaper employees， from pressmen to paper handlers， elevator operators to reporters， joined the picket lines4. “At its core， the New York newspaper strike was a battle over technology，” Scott Sherman wrote in Vanity Fair. “The 1950s and 1960s saw the emergence of computerized typesetting systems that would revolutionize the newspaper composing room. Newspapers that prohibited unions， such as the Los Angeles Times， rushed to install cutting-edge computers such as the RCA 301. Newspapers with union contracts， including those in New York City， faced tempestuous5 resistance from labor leaders， who could easily see that automation would cost jobs.”

Of course， the newspaper strike didnt stop the future. If anything， it simply hastened the demise6 of numerous newspapers. What the union leaders at the time didnt understand （or want to accept） was that technology cannot be stopped， slowed， or avoided in an industry that is being disrupted. Either you find a way to adapt， or someone else does.

Hollywood， which has mostly avoided the dislocations affecting the rest of the media industry， now faces its own moment of reckoning7. Unlike journalism， where technology has decimated profit margins， Tinseltown8 is still living high on the hog9， sustaining an entire ecosystem of agents and middlemen who do nothing， add little value， and get paid incredibly well for their services.

Earlier this week， as I was lying in bed， I felt our house rattle very briefly. I quickly Googled “earthquake Los Angeles，” and found myself reading an article that had just been written on the L.A. Times Web site， explaining that a tremor had occurred a few miles from my house， its level on the Richter scale， and a few other pertinent10 details. Yet， at the end of the article， instead of a byline， I learned that the piece had been written by an algorithm. As far as I could tell， no human intelligence had been involved. The same is true for stories about stocks， sports， and anything else that involves numbers and data that can be created by a computer rather than a person.

It wont be long before machine learning11 begins to impact jobs in Hollywood， too. A few years ago， on a tour at the Massachusetts Institute of Technology， professors explained to me how they were exploring A.I. that could be used to write screenplays and edit films. Now， algorithms are actually co-writing screenplays. Computer algorithms are the stars of our shows， too. How long will it be before a film studio， after fighting with an actor about their pay for a sequel， simply animates them instead？

More frightening is a future in which our smart TVs can watch us back. Years ago， I saw a security demo12 for a program that could be used by spy agencies to track peoples eyes as they read words on a screen. The primary purpose was to ensure that only the intended recipient could read a decrypted message， but the software could also monitor pupil movement and dilation， facial movements， and heartbeat speed to determine if viewers were engaged， nervous， excited， and so on. Could that same technology be applied to an audience watching a TV show？ Already， Netflix tracks which show icons you linger on， and adjusts how they are presented to maximize the likelihood that you will click. Now imagine a world in which Netflix and its ilk can literally see when you get bored， if a joke works or falls flat， if a scene isnt exciting enough， and so on.

In some ways， that future is already here. Last year， after crunching the numbers13 on Arrested Developments disastrous fourth installment14， Netflix decided to recut the entire season， shortening the episodes and re-arranging scenes so that they were chronological rather than based around a single character. The remix wasnt perfect， but it was comparatively well received. Perhaps more important， it served as a proof of concept for the idea that content doesnt need to be fixed， but can evolve to suit an audience. In the case of Arrested Development， series creator Mitchell Hurwitz stepped in to recut the show. Soon， however， shows may be able to recut themselves. During a recent meeting at one of the big studios， I heard about a technology that is analyzing the way award-winning movies and TV shows are edited， and exploring if a computer can edit content with the same precision. You come home from work and say， “Hey， Alexa， make me a personalized comedy with a female lead set in New York that is 16 minutes long so I can watch it before dinner.” Then， like magic， thats what youre watching.

When you compare what is happening in Hollywood to what has happened in every other industry， maybe Hollywood unions have a chance to ensure that the people who make a living here dont get screwed over like Uber drivers and journalists did. As one writer explained to me， “When guilds of any kind have been disrupted， you see a complete degradation for the quality of work and pay for the people who work in those industries： they make less money， have no health insurance， and certainly have less power. Disruption has not been good for labor， even if it has been good for consumers.”

Most Americans think Hollywood has already been disrupted by technology. They look at Netflix， Amazon Prime， and the fact that they can stream Game of Thrones on their iPhone as a sign that Hollywood has gone through its inevitable technology changes already. But that quixotic thinking is akin to the newspaper unions of 1962 seeing the upcoming RCA 301 as the only disruption that would happen to newsrooms in our lifetime. In reality， that was nothing. The same is true with the streaming-video platforms today. All these companies have really done is change how we get our content; on the production side， creative work is frustratingly messy and entirely human. Hollywood， whether it believes it or not， is still ripe for disruption. The question is： will it be people who do the disrupting， or a computer somewhere in Silicon Valley？

1962年12月8日剛過午夜，就在那次震動世界的運動中心，曼哈頓西43街，成百上千的報業人士走出《紐約時報》的大樓，走上寒冷的街道，開始了一場持續114天的罷工。約有1.7萬名報業雇員，從印刷工到紙工，從電梯工到記者，參加了這場罷工。“究其實質，紐約的報業罷工是一場針對技術的戰斗。”斯科特·舍曼在《名利場》雜志撰文寫道，“20世紀50年代和60年代，電腦控制的排版系統出現，給報業的排字房帶來了革命性的影響。禁止工會活動的報紙，例如《洛杉磯時報》，火速安裝了諸如RCA301型的最先進電腦。而那些簽了工會合同的報紙，包括紐約市內的那些報紙，則遭遇了來自工會領袖的激烈抵制，后者能夠輕易看出自動化會導致失業。”

當然，那場報業罷工并沒有阻止未來的到來。如果它起到什么作用的話，也僅僅是加速了無數報紙的死亡。那時的工會領袖們沒能理解（或者不想接受）的一點是，在一個正在被科技顛覆的行業里，沒有什么能夠阻止、延緩或者躲開科技。你只能想辦法調整適應，不然就有別人這么做。

好萊塢在很大程度上已經避開了影響其他媒體行業的那些混亂，但現在卻被迫直面自己難以逃避的命運。和新聞行業不同——在那里，科技的發展已經大大削減了利潤空間，好萊塢仍然紙醉金迷，支撐著包括代理商和中間人的一整套生態體系，這些人什么也不干，提供的附加值很少，卻因其服務獲得了難以置信的高額酬金。

本周早些時候的一天，我還在床上躺著時，突然感覺房屋短暫震顫。我迅速上谷歌搜索“地震、洛杉磯”，然后就看到《洛杉磯時報》網站剛剛貼出來的一篇文章，該文說明在距我家幾英里開外的地方有輕微地震發生，列出了里氏震級和其他幾個相關細節。然而，在文章末尾，我卻沒有看到作者的署名，而是得知這篇文章是由一種算法寫就的。根據我的判斷，沒有人類智慧參與這篇文章的書寫過程。那些關于股票、體育和其他任何涉及能由電腦而非人生成數字及數據的報道同樣如此。

要不了多久，機器學習也會對好萊塢的工作產生影響。我幾年前參觀麻省理工學院的時候，教授們給我介紹了他們正如何研發可以用于編寫劇本和剪輯電影的人工智能。如今，算法已經實際參與劇本的編寫。算法也成為娛樂業的明星。那么，電影公司將來如果與某演員就拍續集的片酬問題產生分歧，就可以直接選擇電腦合成其影像了吧——這種情況還要多久才會出現？

更可怕的是，未來我們的智能電視可能反過來觀察我們。多年前，我看過某一程序的安全示例代碼，情報機關可以用它來追蹤人們在閱讀屏幕上文字時眼睛的移動情況。其主要目的是為了確保只有目標接收者才能閱讀解密信息，但這個軟件還可以用來監視瞳孔的移動和擴張、面部動作和心跳速度，從而判斷觀看者是否投入、緊張、激動等等。這一技術手段能否應用在電視觀眾身上？奈飛公司其實已經開始追蹤觀眾會偏愛哪些劇集圖標，并調整它們的呈現方式以最大化地提高觀眾點擊的可能性。現在，想象一下這樣的世界：奈飛之類的公司真能“看”出來觀眾是否覺得無聊、某個笑話是否奏效、某個場景是否足夠刺激，如此等等。

在某些方面，那樣的未來已經來到眼前。去年，在對《發展受阻》災難性的第四季進行大量數據統計分析后，奈飛公司決定重新剪輯整個第四季，將每集縮短并重新安排場景，使其按時間順序排列，而非原本的聚焦于一位人物。這個重新剪輯版并不完美，但接受度相對較高。或許更重要的是，它證明了以下觀點：內容并不需要重新修訂，但可以通過演化來適應觀眾需求。在《發展受阻》這個例子中，劇集創作者米切爾·赫維茨親自出馬重新剪輯。然而，很快，或許作品就能自我剪輯了。在某個大電影公司近期的一次會議上，我聽說了這么一項技術，公司用該技術分析那些獲獎影視劇的剪輯方式，并探索電腦可否以同樣的精準度來剪輯內容。你下班回到家，只消說：“嘿，亞歷克莎，給我做個定制喜劇，主角要女的，地點設在紐約，16分鐘長，我要在晚餐前看。”于是，像變魔術一樣，你就會看到這么個作品。

將正在好萊塢發生的事和已經發生在別的行業的事做個比較，或許好萊塢的工會還有機會保證在此謀生的員工不會像優步司機和新聞記者那樣陷入窘境。正如一位作家向我解釋的：“一旦行業協會被顛覆，就會看到產品質量和從業人員收入整體下降：從業人員薪酬減少，沒有健康保險，擁有的權力自然也會減少。行業顛覆對勞動力而言從事都不是好事，雖然對消費者是好事。”

大多數美國人認為好萊塢已經被技術所顛覆。他們將奈飛和亞馬遜Prime會員，以及能在自己的蘋果手機上觀看《權力的游戲》這一事實，視作一個標志，說明好萊塢已經歷過它無法避免的技術改革。但是這種堂吉訶德式的想法類似1962年報業工會看到RCA301型計算機到來時的那種想法，天真地以為這會是我們此生經歷的對新聞業的唯一威脅。事實上，那只是小菜一碟。現在，流媒體播放平臺亦是如此。所有這些公司迄今做過的其實就是改變我們獲得內容的方式;在生產環節，創意作品仍然混亂到令人沮喪的程度，并且完全依靠人力。無論相信與否，好萊塢仍然有待進一步顛覆。問題是，帶來顛覆的將會是人類，還是硅谷某處的計算機？

（譯者單位：北京外國語大學）