What have the Pyramids of Egypt, the space race and Wikipedia all have in common? Apart from being great achievements for humanity they were all accomplished through small contributions by a massive amount of people. But what else can be achieved through massive-scale collaboration?
Luis von Ahn has already begun to answer this question. He is the guy responsible for re-CAPTCHA, the service that stops spam on webforms by forcing the user to enter a distorted sequence of characters. Re-CAPTCHA is different to the original CAPTCHA because it presents two words instead of random characters.
Whilst you have probably filled in a Re-CAPTCHA, what you might not have known is that by filling in the two words, you are helping to digitize the worlds collection of print books – something that computers struggle to do automatically.
How? The trick is that one word in the re-CAPTCHA is the security word that the computer knows, and the other is a digital image from a print book. You get the security word right, you are probably going to get the other word right as well. And if ten other people say that the word is what you have said it is – then we have one accurately digitized word.
Because re-CAPTCHA is so popular on websites, it is managing to digitize 100 million words a day – the equivalant of 2.5 million books a year.
Fascinating stuff! But what is more exciting is his next project – translating the web. Computer translation is not going to be perfect for at least another 10 years, and hiring professionals to translate the the non-Spanish Wikipedia pages into Spanish (only 20% of the English Wikipedia is in Spanish) would apparently cost $50 million – and that’s at almost slave labour wages.
Luis von Ahn has tackled this by connecting this problem with the 1.2 billion people that are learning another language. His website Duolingo offers people the opportunity to learn a language for free (language lessons are notoriously expensive) in exchange for their time in translating the web.
So, if ten people learning a language all translate a sentence the same way then it is going to be correct. And the people taking part are learning by doing! Although it is still only in testing phases – it is apparently a powerful language teacher and a really accurate way to translate content. If the site gets a million active users – it will be able to translate Wikipedia into Spanish in 80 hours.
To hear more check out the TED talk