When he was 12 years old, Luis von Ahn came up with a plan to make gyms free.
People exercising on machines can generate electricity, he figured, and that energy is valuable. So why not eliminate gym fees, hook all the machines to a power grid, and sell the wattage produced to a major electric company? Everyone could go free of charge, the world would have a new source of power, and people would be healthier to boot.
“It turns out it’s not a very good idea,” von Ahn, now 34, chuckles. “People aren’t very good at generating electricity. It’s much better to charge a membership fee.”
While that idea didn’t pan out, the computer science professor at Carnegie Mellon University has been dreaming up innovative business models ever since. And he’s done it well. Over the past eight years, von Ahn has created and sold two projects to Google. His new venture, free language-learning app Duolingo, is a perpetual favorite in the Android and iOS app stores and has already accrued more than 12 million users. In 2006, he was awarded the prestigious MacArthur Fellowship, or so-called “genius” grant.
If there is true genius to be found in von Ahn’s work, it lies in the theory that underscores all of his projects: the idea that by using technology and a little bit of fun, you can harness tiny bits of time and energy from people all around the world and make them collectively useful. In what might be the cleverest application of crowdfunding principles yet, von Ahn is turning our mindless Internet activities into something productive.
Von Ahn’s entrepreneurial ventures began in earnest in 2004 with an idea he had for a new kind of online game. The program would randomly pair each player with another user on the web, and show them a series of images. Both players were instructed simply to “type whatever the other guy is typing.” The more overlap you produced, the better your score was. So, for example, if a picture of a dog appeared, both users would probably type “dog” along with other words like “animal,” “pet,” “puppy,” or “cute.”
It’s the kind of time-killer that most of us love: a perfect medley of fun images, competitive quizzing, and mindlessness. But for von Ahn, it would have a second use. “When people play the game they help determine the contents of images by providing meaningful labels for them,” he and his co-author wrote in a 2004 paper. “If the game is played as much as popular online games, we estimate that most images on the Web can be labeled in a few months.”
Take a moment to consider that proposition. A tremendous number of unlabeled images are floating around on the web, which impairs everything from the accuracy of image searching to the blocking of inappropriate content. Tech companies have created an entire job category for people who review content and flag it for various graphic violations. Von Ahn was proposing that much of this could be outsourced to your everyday person, if only it were made a little fun.
The program launched in 2005 as The ESP Game. Within four months, it had lured 13,000 bored web cruisers into producing 1.3 million labels for roughly 300,000 images, Wired reported in 2007. Von Ahn’s demo of the game at Google caught the eye of both Sergey Brin and Larry Page, and just months later it had been acquired and relaunched as the Google Image Labeler.
Von Ahn’s next venture, reCAPTCHA, also managed to utilize the work of unsuspecting web users. In the early years of his Ph.D. study, von Ahn had helped his advisor, CMU computer science professor Manuel Blum, develop a handy identity verification device known as a CAPTCHA. Think of those distorted words you’re asked to translate after attempting to log into your email too many times to verify that you’re human. Those are CAPTCHAs. Initially invented to help keep spambots out of chat rooms, these tests are effective because computers have a difficult time reading distorted text, while people are rather good at it.
Von Ahn watched the work on CAPTCHA and decided it had potential beyond distinguishing humans from robots — the extra 10 seconds people were taking to access their email and other accounts could be put to use. In 2006, von Ahn launched reCAPTCHA. Unlike its predecessor, reCAPTCHA challenged users with two distorted words to decode, and looks something like this:
The brilliant twist is that this test isn’t just verifying your humanity; it’s also putting you to work on decoding a word that a computer can’t. The first word in a reCAPTCHA is an automated test generated by the system, but the second usually comes from an old book or newspaper article that a computer scanner is trying (and failing) to digitize. If the person answering the reCAPTCHA gets the first word correct (which the computer knows the answer to), then the system assumes the second word has been translated accurately as well.
In 2009, Google acquired reCAPTCHA for an undisclosed amount (von Ahn says the sum was somewhere between $10 million and $100 million) and put the program to work on a tremendous scale, digitizing material for Google Books and the New York Times archives. In 2012, it was translating about 150 million distorted words a day.
“The CAPTCHA was really my idea,” says Blum. “Getting humans involved and getting them to help do this stuff was Luis’s idea. He was the one that pointed out, ‘Look how many hours have gone into building the Panama Canal or the Pyramids — and with all the people that are on the web now, you can get a lot more hours.'”
The latest incarnation of this theory is Duolingo, von Ahn’s popular language-learning game. The free service offers lessons in Spanish, French, Italian, German, and Portuguese, and uses a computer-game structure with level-ups, un-lockable bonus skills, and a virtual currency to intrigue users. It’s classic gamification.
“When you talk to people using Duolingo, they usually say ‘I’m playing Duolingo,'” von Ahn notes. “If you ask people the main reason they’re using Duolingo, it’s not because they’re learning something but because it’s fun.”
The app’s 12.5 million active users spend, on average, 30 minutes a day with Duolingo, but it’s also designed for people to pull out for two or three minutes as a time-killer while waiting in line at the grocery. Von Ahn says his research shows that spending 34 hours on Duolingo teaches the equivalent of one semester of a college language course. Eighty percent of traffic to the app comes from mobile.
A quarter of Duolingo’s users are from the U.S., but another 35% are from Latin America and Brazil, and 30% are from Europe. This is important to von Ahn, who grew up in an upper-middle class family in Guatemala City before heading to the U.S. for college, and saw firsthand how his fellow citizens struggled to climb the socioeconomic ladder.
“Guatemala is a very poor country,” von Ahn says. “Everybody in Guatemala seems to want to learn English, but no one can afford to.”
To that end, von Ahn has sworn to keep Duolingo entirely free for users. And using the same logic that built The ESP Game and reCAPTCHA, he’s come up with a clever alternative for monetizing the product. When users sign onto Duolingo, one of the options they have for practicing their language is “immersion.” In this section, users get a chance to apply what they’ve learned by trying their hand at translating real documents on the web.
Where do those documents come from? CNN and Buzzfeed, for starters. The major media companies have contracted Duolingo as a translation service for their materials. Even with novice users, the translations are fairly accurate because several people on Duolingo work on each document and then up/down vote other translations before the final version is sent back to the media outlets. For the users, it’s another language-learning tool; for Duolingo, it’s a way of generating hundreds of thousands of dollars.
As with all von Ahn’s projects, the trick in Duolingo comes down to shrewdly harnessing the time people happily spend on one project to do something useful in another. One invention after another, he is satisfying our desire for mindless fun while tricking us into making society as a whole more efficient. The brilliance of the theory, ultimately, is that it’s so simple and yet extremely effective.
“It’s just taking something that people do anyways,” von Ahn says, “and trying to extract value out of it.”