The Boy Kings

text box. Other users could vote the translation up or down. This type of crowdsourced interface is all over the Web now as a way of managing the Internet’s increasingly heavy flow of content, but it was a newer interface then, exciting to engineers in its limitless possibilities for mechanization of things formerly considered subjective.
    Voting on highly subjective content, such as the right way to phrase a complicated concept like poking or the wall, can producemore conflict than agreement. There was often no definitively correct answer but, instead, many different interpretations of a given word. For example, the Spanish translators wanted to know if wall meant the side of a building or something more like a bulletin board (the answer was the latter, though then there were an array of different words for a bulletin board for translators to vote on and choose from). Usually, the voting results produced passable translations, but when there was a translation impasse, I noticed that some engineers placed an almost religious faith in the voting process, and seemed to feel threatened by the idea that the algorithmically decided results might not be perfect. “The voting will fix it,” they said, like a mantra, as the translations rolled in and vied for victory on the page.
    The engineers were beside themselves with glee when the French version of Facebook was translated literally overnight by local users; however, having spent a year in a French school as a child while my mother was on academic business in France, even I knew that the translations, while they were certainly done marvelously quickly, were not polished and correct enough to launch. I argued to my team that some kind of human review of the final product was needed. I just wanted to know, for sure, that the translations made sense and were at least a proximate version of the quality on the English site. Some people on my seven-person team, composed of engineers, Hassin, and a business development guy from Spain by way of Stanford, grew frustrated with my stubborn defense of human cognition over algorithm, but I didn’t much mind. Being the odd defender of the value of the human was something that I was used to and was, after all, sort of my job. In more ways than one, I was likethe humanist troll to the company’s obsession with technologizing everything.
    Hassin, a linguist rather than an engineer by trade, agreed that some human input would be worthwhile and so I worked with professional translators to review the site in our first non-English languages, French and Spanish. Once launched, a user would view the site in French or Spanish by toggling a button on the Facebook homepage that would switch the language of the interface (user-submitted content like comments and status updates would remain in whatever language the user wrote in). I spent days with the professional translators while they read through pages of translations and made corrections as needed. They were working by the hour, clocking out at six o’clock, and thought it strange that I seemed perennially online the entire week, answering chats, reading Facebook, talking with them, answering questions, and responding to emails at all hours. When they left the office at the end of the day, they were done until the next morning. That, in turn, seemed strange to me. I couldn’t remember when the last time was that I wasn’t within spitting distance of my computer and smart phone. As much as I had once made fun of the Facebook boys for staring at their phones more often than they looked up, I had become one of them.
    We launched the Spanish version of Facebook in February 2008 and followed with French one month later. Both Facebook interfaces launched to good press and widespread adoption, as the site was programmed to appear immediately in French or Spanish when a user signed up or logged in from a country in which one of those languages was primarily spoken. We movedon from those languages to getting the site translated into German, Japanese, and Italian. French and Spanish came first because they addressed the largest number of potential users but, after that, we translated in order of the wealthiest countries. Internationalization, like everything else, follows the money.
    This was where I got lucky, and where my job began to save me from my dry, tech-saturated Palo Alto existence in a new way. Because we were striving for authenticity as well as technical accuracy in translation, it