Facebook上上周四召開的工程大會(huì)涉及了很多方面的內(nèi)容,。

首先，F(xiàn)acebook介紹了它怎樣對(duì)其社交網(wǎng)絡(luò)的基礎(chǔ)軟件進(jìn)行了升級(jí),。同時(shí),，谷歌和微軟也分別展示了自己在機(jī)器學(xué)習(xí)等熱門領(lǐng)域的進(jìn)步。

此次大會(huì)是在加州西部城市圣何塞舉辦的,。會(huì)上的有些發(fā)言聽起來(lái)簡(jiǎn)直就像科幻小說(shuō)——比如DNA能夠成為數(shù)據(jù)的儲(chǔ)存介質(zhì),。當(dāng)然，會(huì)上也討論了一些離現(xiàn)實(shí)生活更近的問題,，比如如何促進(jìn)科技界的多元化等等,。在Uber等硅谷企業(yè)頻頻爆出性別歧視和性騷擾丑聞的背景下，后一類話題顯然也是很有意義的,。

以下就是此次Facebook Scale大會(huì)上的部分精彩看點(diǎn),。

1.DNA或成為數(shù)據(jù)儲(chǔ)存介質(zhì)

DNA除了承載著人類的全部遺傳信息，它說(shuō)不定還能用于另一個(gè)用途,。

華盛頓大學(xué)計(jì)算機(jī)科學(xué)教授,、微軟研究員路易斯·塞茲介紹了科學(xué)家們對(duì)DNA作為數(shù)據(jù)存儲(chǔ)介質(zhì)的可能性研究。專家認(rèn)為DNA可以儲(chǔ)存大量數(shù)據(jù),，其容量甚至超過了現(xiàn)在的旋轉(zhuǎn)盤或閃存技術(shù),。

塞茲表示，華盛頓大學(xué)和微軟已經(jīng)成功地將音樂短片,、《人權(quán)宣言》和古騰堡工程的文學(xué)作品存儲(chǔ)進(jìn)了人工合成的DNA,。DNA相比傳統(tǒng)存儲(chǔ)介質(zhì)具有永不過時(shí)的優(yōu)點(diǎn)，不會(huì)像20年前的軟盤一樣被扔進(jìn)歷史的垃圾堆里,。

不過科研人員在這方面要做的工作還有很多,。目前，從DNA傳輸10兆的數(shù)據(jù)大約需要一周的時(shí)間,，大致跟上世紀(jì)70年代的調(diào)制解調(diào)器的速度差不多,。

2.Netflix背后的強(qiáng)大計(jì)算能力

每天都有幾百萬(wàn)人在視頻網(wǎng)站Netflix上觀看《紙牌屋》和《超膽俠》等美劇。所以網(wǎng)站必須要有強(qiáng)大的計(jì)算能力,，才能確保視頻不卡頓或崩潰,。

Netflix高級(jí)軟件工程師弗蘭克·圣米格爾表示，Netflix的流媒體視頻服務(wù)有亞馬遜數(shù)據(jù)中心里的30萬(wàn)個(gè)CPU支持,。這些服務(wù)器不僅負(fù)責(zé)播放視頻,，也負(fù)責(zé)支持視頻處理、字幕和后期制作等上百個(gè)應(yīng)用程序。

3.谷歌翻譯或成“下一個(gè)大事件”

谷歌對(duì)運(yùn)用人工智能翻譯技術(shù)有著極為濃厚的興趣,。谷歌負(fù)責(zé)機(jī)器學(xué)習(xí)和語(yǔ)言翻譯業(yè)務(wù)的高管巴拉克·圖洛夫斯基表示,，95%的谷歌翻譯用戶都在美國(guó)以外。該服務(wù)最大的用戶群反而是巴西,、印度,、泰國(guó)和印尼人。

谷歌翻譯為何如此流行呢,？首先是因?yàn)榛ヂ?lián)網(wǎng)上有一半的內(nèi)容都是用英文寫成的,。隨著發(fā)展中國(guó)家智能手機(jī)愈發(fā)普及，無(wú)線基礎(chǔ)設(shè)施日益完善,，非英語(yǔ)國(guó)家的網(wǎng)民越來(lái)越多,，很多人都存在借助谷歌翻譯閱讀網(wǎng)絡(luò)內(nèi)容的需求。

圖洛夫斯基：“比如印度有大量人口只會(huì)說(shuō)本地語(yǔ)言,，但是網(wǎng)絡(luò)上卻找不到用本地語(yǔ)言創(chuàng)作的內(nèi)容,。”

4.Nest和Slack高管談?wù)衅概c職場(chǎng)多元化

雖然此次大會(huì)的焦點(diǎn)是高科技,，不過職場(chǎng)即時(shí)通訊服務(wù)Slack和谷歌的家居自動(dòng)化業(yè)化Nest等公司的高管還是談到了在招聘員工上的挑戰(zhàn),。畢竟沒有一支精干的員工隊(duì)伍，你是做不出任何技術(shù)的,。

Slack的基礎(chǔ)建設(shè)工藝總監(jiān)朱麗亞·格雷斯談到了職場(chǎng)多元化問題,。她舉例道，如果一名求職者走進(jìn)面試現(xiàn)場(chǎng),，發(fā)現(xiàn)整間屋子坐著的都是同一個(gè)性別,、同一個(gè)民族或是具有其他相同特征的人，那么他們可能會(huì)擔(dān)心,，自己能不能在這樣一家公司獲得成功,。

格雷斯向聽眾問道：“如果你在公司里永遠(yuǎn)見不到跟自己相似的人，你是什么感覺,？那么從入職第一天起,，你就會(huì)問自己：‘我真的屬于這里嗎？我能在這里獲得成功嗎,？’”

Nest公司電子商務(wù)與網(wǎng)絡(luò)安全高級(jí)總監(jiān)朱莉·珀?duì)柋硎?，Nest的科技產(chǎn)品是面向普通消費(fèi)者的，所以公司必須確保它生產(chǎn)的聯(lián)網(wǎng)恒溫器等家居產(chǎn)品能夠適應(yīng)來(lái)自各種背景的消費(fèi)者,。因此,，Nest要想推出被更多人接受的產(chǎn)品，就要需要更多來(lái)自不同文化背景的員工,。

珀?duì)柋硎荆骸拔覀冊(cè)谡衅钢袝?huì)強(qiáng)調(diào)某種形式的多元化?，F(xiàn)在我正在認(rèn)真思考,，如何將它體現(xiàn)在我們所構(gòu)建的產(chǎn)品中?！保ㄘ?cái)富中文網(wǎng)）

譯者：賈政景

Facebook's latest engineering conference for techies on Thursday covered a lot of ground.

Facebook described how it overhauled the underlying software for its social network. Meanwhile, Google and Microsoft revealed their latest advancements in hot areas of data crunching involving artificial intelligence technologies like machine learning.

Some of the discussions at the event, held in San Jose, seemed like science fiction such as storing digital data in DNA. Other topics like the problems of diversity in tech were more practical amid recent discrimination and harassment allegations in Silicon Valley at companies like Uber.

Here's a roundup of some of the most fascinating items from Facebook's Scale conference.

1. Storage drives of the future may be built with DNA

It turns out there may be another use for DNA besides serving as blueprints for humans.

University of Washington computer science professor and Microsoft researcher Luis Ceze discussed how experts are looking into how DNA could be used to store huge amounts of digital data, more so than any current technology like spinning discs or flash drives.

Ceze explained the progress his university and Microsoft (msft, +1.04%) have made in storing things like music videos, the Universal Declaration of Human Rights, and literature from the Project Gutenberg archive initiative into manufactured DNA. Some of DNA’s advantages over conventional storage are that it won’t become obsolete, like the old floppy discs of the past, he said.

However, researchers have a lot more work to do. As of now, it takes about a week to transfer 10 megabits of data from DNA, which Ceze said is the equivalent of the speed of old modems from the 1970’s.

2. It takes a lot of computing power to make Netflix run

Millions of people visit Netflix (nflx, -0.14%) daily to binge watch House of Cards and Daredevil. But it takes a lot of computing power to make sure that those videos don’t stutter or crash.

Netflix senior software engineer Frank San Miguel said that on a “good day,” the video streaming service is powered by roughly 300,000 CPUs running in Amazon’s data centers. Those thousands of servers are not only responsible for delivering video to people, but they also power the 100 apps Netflix built that handle tasks like audio processing, subtitles, and other post-production services.

3. Google’s big efforts to translate languages

Google’s interest in using artificial intelligence to quickly translate languages became clearer. Barak Turovsky, a Google (goog, +1.02%) executive overseeing machine learning and language translation, said that 95% of people using Google to translate text are not based in the U.S. In fact, residents in Brazil, India, Thailand, and Indonesia are the biggest users.

One reason for its popularity overseas is that half of all Internet content is written in English, Turovsky said, a consequence of the English-speaking world being able to afford the computers to access the Internet. But the rise of smartphones and increasingly improved wireless infrastructure in developing countries has led to non-English speakers getting online—and needing to use Google’s translation service to read online content.

“There’s a huge population of people who speak Indian languages, but they can’t find content created in their language,” Turovsky said.

4. Nest and Slack executives talk hiring and diversity

Although Facebook's conference focused on cutting-edge technology, several executives from companies like workplace messaging service Slack and Google’s Nest home automation business talked about the challenges of hiring workers. You can’t build technology without a savvy workforce, after all.

Julia Grace, Slack’s head of infrastructure engineering, discussed the problems companies with diversity problems (ie. most tech companies) face when recruiting. For instance, if someone walks into a job interview and only sees people who look the same in terms of gender and ethnicity or some other characteristic, they may question their chances of succeeding at the company.

“What happens if you never see anyone who looks like you?” Grace asked the audience of techies. “You, from day one, will ask ‘do I belong here? Can I be successful here?’”

Nest senior director of e-commerce and security Julie Pearl explained that because Nest builds technology for everyday consumers, it must also ensure that those web-connected thermostats and other gadgets are built to accommodate people from diverse backgrounds. For this, Nest needs more workers with different cultural backgrounds to build its products, she explained.

“We talk about certain types of diversity with hiring,” said Pearl. “Now I’m really starting to think about how does that come out in the products that we are building.”