Google Translate has reached a milestone of 200 million users on translate.google.com. Ordinarily, we wouldn’t report a basic usage statistic like that, but this one deserves to be celebrated. Google’s scale could someday make it possible for all human beings to understand each other. Here’s how.
Translating the Web into every language on Earth, on the fly, is one of the most important missions in tech. It’s the key to a global society. We still have a long way to go. Online translation has been possible for years, but not in all languages, and results are often hilarious and unhelpful.
We might complain about that, but first we should appreciate how hard translation is to do.
Franz Och, a former DARPA researcher working on Google Translate, offered a brief history of the project on the occasion of the 200-million-user landmark. Google Translate began in 2001 as a machine-translation service with nine languages. It was state-of-the-art at the time, but the translation quality was still poor.
In 2003, a team at Google decided to get serious about translation, which is when Och got involved. The team quickly figured out that a data-driven approach – using vast amounts of language culled from the Web – could succeed at scale. But its first system was too slow. It took 40 hours and 1,000 computers to translate 1,000 sentences.
The team then turned to focus on speed. Google announced the new approach in 2006, beginning with Chinese and Arabic. In the six years since, Google Translate has incorporated 64 languages, many of which have little presence on the Web. But now Google Translate can help speakers of those languages reach the rest of the world, and hopefully that will encourage them to come online.
Google sees so much of the Web that it’s in the best position to perfect Web-scale translation. “In a given day we translate roughly as much text as you’d find in 1 million books,” Och says. “What all the professional human translators in the world produce in a year, our system translates in roughly a single day.”
Language Is an Interface
It’s easy to forget this, but language is a fundamental part of a computer interface no different from graphical windows, avatars or follow buttons. Translation is not just important for abstract, academic reasons. It’s necessary in order to get the Web to work. If people can translate Web content into their own language, they gain access to parts of the Web they couldn’t reach before.
It’s easy to understand the value for Google in offering a free translation service. The more translation data Google can get and understand, the better its translations will be. With better translations, the Web is more accessible for more people. The more people with access to more Web content, the more money Google makes.
But it’s crass to suggest that Google is building Translate with ulterior motives. Google goes a long way to make Translate a great experience. Its native apps enable voice-to-text transcription, and it even contains a voice synthesizer to read translations back to you. Android users can even do speech-to-speech conversions out loud with other people!
We don’t need those features to get around on the Web, but Google built them anyway.
“We imagine a future where anyone in the world can consume and share any information, no matter what language it’s in, and no matter where it pops up,” Och says. This is an important case in which Google’s business goals are right in line with the best interests of users. With all the worries about privacy arising from Google’s new social layer, it’s easy to lose sight of this, but sometimes Google’s goal is just to help the world communicate.