PITTSBURGH — At times, it’s hard to keep up with Luis von Ahn. The Duolingo CEO and co-founder speaks quickly, ideas flowing out of a mind that helped the Guatemalan native win the MacArthur Fellowship, also known as the “genius grant,” in 2006.
But inside a conference room at Duolingo’s headquarters in Pittsburgh, our back-and-forth chatter is replaced by brief silence. Asked for thoughts on improving public education around the world, given his experience building a language learning startup for the past seven years, von Ahn pauses for a moment.
“I have a lot of conflicting things to say about education,” he says, taking a deep breath before explaining how the system in some countries is “completely broken.”
Von Ahn is already an accomplished entrepreneur. As a rising star at Carnegie Mellon University’s acclaimed computer science school, he sold two startups to Google, including reCAPTCHA, a computer-generated bot prevention tool you’ve likely used online that simultaneously helps digitize books. Some refer to him as the father of crowdsourcing.
But Duolingo, which counts 200 million users and has been valued at $700 million, is personal for von Ahn. The inequalities created by traditional education, especially in his hometown of Guatemala City, light a fire inside the 39-year-old, who launched the CMU spinout in 2011 with Severin Hacker.
“Particularly in poor countries, education is just really broken,” he explained. “People who have money can buy themselves the best education in the world and people who don’t barely learn how to read and write, and because of that they never make money. It’s this thing that’s self-perpetuating — those with money remain with money because they’re well-educated, and those with no money remain with no money because they’re not educated.
“That’s why we decided that Duolingo would be completely free,” continued von Ahn, whose idea has caught the interest of everyone from Barack Obama to Bill Gates. “This is always a stated goal of Duolingo: somebody who doesn’t have a bank account should be able to use all of the learning content of Duolingo. That’s our stated goal and we remain true to it to this day.”
As part of our month-long GeekWire HQ2 project, I sat down with von Ahn before he took the stage during our panel discussion at Duolingo last week in Pittsburgh with 250 people from the local tech community. von Ahn is unassuming at first glance, but watching him mingle with fellow techies, it’s clear that he’s a tech celebrity of sorts in this city.
The CEO is also not afraid to speak his mind or crack a joke. He drew big reaction throughout the panel talk — applause for his comments on diversity in tech and laughs when he credited Primanti’s sandwiches for keeping him in the Steel City.
Asked about Pittsburgh’s reputation for innovation, von Ahn reminded me that the Big Mac was invented here. His playfulness started at an early age; during our interview, he retold childhood stories about creating machines to help do his homework faster or hacking the local radio contest with friends.
“I always got good grades, but I was also usually in the principal’s office for ridiculous stuff,” von Ahn said.
Duolingo is one Pittsburgh’s hottest startups. It now offers 75 language courses for 30 distinct languages, and its lessons still remain free. In the past few years Duolingo has monetized via ads, in-app purchases, monthly subscriptions, and the Duolingo English Test.
Some highlights from our conversation:
- Von Ahn said Duolingo is preparing for an IPO by 2020. “I’m a little scared about that just because I think there’s a lot of crap that I don’t particularly want to do that I’m going to have to do,” he said. The company has raised $108.3 million from Drive Capital; CapitalG; Union Square Ventures; Kleiner Perkins Caufield & Byers; New Enterprise Associates; Tim Ferriss; and Ashton Kutcher.
- Gates is a big fan of von Ahn, who made a trip to Seattle last month to meet with the Microsoft co-founder and talk about Duolingo. Gates tried to recruit von Ahn for a job at Microsoft Research years ago. “He understood that I wanted to own what I did,” von Ahn recalled. “I said, ‘I’m sure at Microsoft I can do really cool stuff but I want to own what I do.’ He said, ‘I can’t argue with that.'”
- It’s becoming easier to recruit employees to Pittsburgh, von Ahn said. About 85 percent of Duolingo’s 115 employees are from outside of Pittsburgh, which has transformed from an industrial powerhouse to modern innovation factory. “The clincher is usually when you take them to go look at a house,” he said of employees who relocate for the job (median home price in Pittsburgh: $130,000).
- Von Ahn never sold his first big invention, CAPTCHA (the predecessor to reCAPTCHA), to any company. It was simply given to Yahoo. “I was an academic trying to get a PhD,” he said. “It just hadn’t occurred to me. We were very happy with the fact that Yahoo was using it.”
- Von Ahn’s advice for entrepreneurs: stop coming up with excuses for not getting started. “It won’t be easy, but you’ll be a hell of a lot more likely to get funding if you show up with something that’s actually done, as opposed to a blue sky idea,” he said.
- Von Ahn wants a new way for university professors and researchers to be incentivized. “It could work a lot better if more people had a goal to have impact as opposed to writing some papers,” he said.
- Von Ahn loves TV shows. His latest interest is Halt and Catch Fire. “It’s like a combo of ‘Mad Men’ and ‘Silicon Valley’ at the same time,” he noted.
Read on for more from our interview with von Ahn.
GeekWire: Tell me a bit about growing up in Guatemala. Your parents owned a candy factory and your childhood seemed to influence you in a lot of ways.
Duolingo CEO Luis von Ahn: “I was very fortunate that in the late 80s when I was 8 years old and I wanted a Nintendo, my mom bought me a computer. That was probably the single best decision she ever made. She got me a Commodore 64, which is not super easy to use and for an 8-year-old in particular. You had to type commands and I had no idea what to do. I read the instruction manual and that didn’t tell me all that much.
Eventually I figured out how to play games and at some point I realized there was a very easy way to get around the copy protections. I started copying games from other people who had Commodore 64s. All these adults knew me as a kid that would be able to copy games for them. I got into pirating games, I guess, but the statute of limitations for that is probably gone. I got good at that and that’s kind of how I got good at computers.
My parents did have a candy factory and I worked there every summer. I didn’t really do much. One year I was supposed to set up their computer network. It took me like a day to set it up, but then I kept on claiming I was setting it up for the next three months.
When I was a little younger I would sneak in and break all the machines because I was trying to understand how they worked. I didn’t know how to put them back together; I would put them back together and pieces were missing. People were pretty pissed off at me. That’s basically how I spent my childhood.”
GeekWire: Didn’t you make a homework machine?
von Ahn: “Yeah, I got into trouble for that. You know those penmanship classes where you have to make a million O’s? I basically made a machine that made O’s. It was much faster. It did not look impressive, it was just a bunch of pens that were stuck with each other and a little motor that broke all the time. But it did go faster.”
GeekWire: You were a troublemaker.
von Ahn: “I always got good grades but I was also usually in the principal’s office for ridiculous stuff. One time we hacked this radio contest. Every week the main radio station of the city had the top 10 songs of the week and if you could guess them you had the chance to be a DJ the next week.
They had just started accepting faxes and with my computer I could send a ton of faxes. I sent them faxes to try to guess what the winning combination was and I did win. Then I was the DJ the next time around and I had all my friends listening. By code I would tell them what the winning list was and we won for six months in a row. It was a big scandal and the principal of my school, because this was a fancy school, was like, ‘You are not representing the values of our school.’ I got in trouble like that.”
GeekWire: After you did your undergrad studies at Duke you arrived at CMU, where you helped invent CAPTCHA AND reCAPTCHA. Do you have any estimation of how many users have used it to date?
von Ahn: “Billions. It’s done 200 million times a day.”
GeekWire: “How did you come up with the idea for reCAPTCHA?”
von Ahn: “I did a little back of the envelope calculation about how many times CAPTCHAs were done every day around the world. The original number that I came up with was probably 150 million or 200 million. I realized that that was a lot of time being spent — 10 seconds each time you do one, you multiply by 200 million, that’s half a million hours a day. I started thinking if we could use this human effort for good and that’s where reCAPTCHA came about, where the idea is that you’re helping to digitize books as you’re typing CAPTCHAs.
reCAPTCHA was a Carnegie Mellon research project. I was a professor and I wasn’t thinking of starting a company. At some point Facebook was one of the websites that decided to use our service. Then I was giving a talk somewhere and the CTO of the New York Times was there and he said, ‘We have a huge archive of old New York Times, can you help us digitize them?’ We said yes, he asked how much we charged, and we had no idea and said something like $50,000 per year of content. He said sure.
This was still inside the university and at some point they told us we needed to leave. We started this company and we’re digitizing the New York Times and it was a $5 million contract. In the middle of it, Google heard about this and they were digitizing all the world’s books. They decided they would just buy reCAPTCHA outright and that’s when we sold the company to them.”
GeekWire: What happened to CAPTCHA?
von Ahn: “The original one, CAPTCHA, we just gave to Yahoo. It occurred to us later that we probably should have done something about that but we didn’t. That was just given out.”
GeekWire: Wow. Were you not thinking that this could actually be a huge company and you could make a ton of money?
von Ahn: “I was an academic trying to get a PhD. It just hadn’t occurred to me. We were very happy with the fact that Yahoo was using it.”
GeekWire: Do you look back now and wish you sold it?
von Ahn: “No, because particularly with reCAPTCHA, I did very well with that. I’m fine.”
GeekWire: Have your perceptions changed in regard to taking ideas from academia to commercialization, especially now that you’re building Duolingo, which spun out of CMU?
von Ahn: “It has changed a lot. At the very beginning it was doing cool stuff for the sake of doing cool stuff. At some point it became doing cool stuff for the sake of doing cool stuff, but it would be cool if I got paid. Then after having a couple of exits it became, forget doing cool stuff, most importantly, can we do something that has impact?”
GeekWire: Do you think more researchers and PhDs at universities should think like that? One criticism of CMU or other universities is that researchers make cool stuff, but for who? And when you want to spin something out, it’s a convoluted and complicated process.
von Ahn: “This is a huge problem with universities. I can talk about this for years. The problem is the tenure system. In order to get tenure at a university, you need to have these publications. There’s a whole game that you need to play and people get really good at playing this game. That’s what the whole academic institution is really good at, coming up with an idea and the end goal is a paper that gets cited a lot.
A paper is usually not quite the same as a working system that has impact. That requires more effort and 99 percent of the people stop right after the paper because that’s what you need. Everybody gets trained to just write these papers. They’re writing hundreds of papers, but impact is very little in most cases.
Now, this is a generalization. There are some people that have had really crazy impact outside of academia and somehow overall the system kind of works, because you can always point to the fact that Google came out of academia — somehow, it works. But I think it could work a lot better if more people had a goal to have impact as opposed to writing some papers.”
GeekWire: How would you fix that?
von Ahn: “I don’t know if there’s an easy way to fix it. All kinds of people have proposed all kinds of things. Some people have proposed that you cannot write more than three papers per year, so you better make sure they’re good and you better spend all your time doing that. Other people have proposed that you get judged not on your papers at all but just on impact on the world. I don’t know if this will ever change.”
GeekWire: You worked at Google for a year and a half after they bought reCAPTCHA. What was like that?
von Ahn: “I loved it. But I want to own what I do so I’m not going to work for a company for a very long time that I don’t own. Google was an amazing company. I have all respect for Google.”
GeekWire: You also had a chance to work for Microsoft, earlier in your career.
von Ahn: “Before reCAPTCHA, I came up with this game that helped tag images on the web so we could improve image search. Google bought that game. I had offers from all the top Ph.D programs to become a professor. Microsoft Research wanted to hire me and that’s when Bill Gates called saying, ‘Hey, come work for Microsoft Research.’
He understood that I wanted to own what I did. I said, ‘I’m sure at Microsoft I can do really cool stuff but I want to own what I do.’ He said, ‘I can’t argue with that.'”
GeekWire: Tell me how you came up with the idea for Duolingo.
von Ahn: “I wanted to do something related to education because I think it has a lot of impact. Particularly in poor countries, education is just really broken. What happens is people who have money can buy themselves the best education in the world and people who don’t barely learn how to read and write, and because of that they never make money. It’s this thing that’s self-perpetuating — those with money remain with money because they’re well educated, and those with no money remain with no money because they’re not educated.
You see that here in the U.S. as well but it’s just not as striking as it is in a country like Guatemala. In Guatemala, there is essentially no social mobility. It’s extremely rare to see somebody that was from a low socioeconomic background that made it to the top. In the U.S. it’s rare, but not unheard of, it does happen. There’s this American Dream.
That was the reason why I wanted to do something that would teach people something useful and be free so that anyone can learn. In 2011, all the software there was to learn a language was Rosetta Stone. The list price was $1,000, which was insane because the majority of people that are learning a language are doing so to get out of poverty. It’s crazy that they have to pay $1,000 for software to try to get out of poverty.
That’s why we decided that Duolingo would be completely free. This is always a stated goal of Duolingo: somebody who doesn’t have a bank account should be able to use all of the learning content of Duolingo. That’s our stated goal and we remain true to it to this day.”
GeekWire: In the past few years Duolingo has monetized via ads, in-app purchases, monthly subscriptions, and the Duolingo English Test. How has the company’s revenue model evolved over time?
von Ahn: “We had ideas of revenue models. But back in 2011, there were companies like Tumblr that had no idea how they were going to make money, but they were really big and everybody was investing in them. It was okay to just show up and say, ‘We don’t know how we’re going to make money but we’re going to make this big, trust us. Surely once you’re big you can figure out the money.’ I think that was pretty okay back then; I think it’s much less okay now.
For the first three years of the company, we did not worry too much about revenue. That was fine because we kept on growing. We were the number one education app and we remain the number one education app. I think investors were quite happy with us.
At some point, when we reached a valuation of almost $500 million, we started feeling bad. We were like, ‘all right, we probably should figure something out to monetize.’ That’s when we really started thinking about how to monetize and how to do it well. That’s also when we hired our head of business and it’s been pretty successful. We monetize through ads and through subscriptions.”
GeekWire: You have also made money via crowdsourcing, helping sites like CNN and BuzzFeed translate their articles.
von Ahn: “We don’t do that anymore. It was too hard. Two things happened. First, most of our employees started shifting toward making translations more accurate because that’s where the money was coming from. We realized we were shifting from an education company to a translation company. That’s not what either Severin or I wanted to do.
Then we also saw that translation is a shitty business. We decided not to do that and we thought we could make a lot more money with things that are related to actual education. That was a good decision, at least for the company.”
GeekWire: Do you advise other entrepreneurs to just go build something and see what happens and hopefully you get enough users and then you can start charging? How should someone think about that?
von Ahn: “It’s really hard to give that type of advice. It really depends on the context. In some cases there are certain things that I will tell you, you better have a freaking business model for that. If you’re making an online game, yeah, you probably should launch it with some in-app purchase monetization scheme. It’s hard to say, but I think for what we did it was actually the right call to not worry too much about monetization. We came in with a product that was superior to Rosetta Stone and free. By now, they’re pretty much dead. It’s amazing that they’re still around. Their stock has plummeted, whereas we’re worth six times as much as they are.”
GeekWire: What did you guys do differently?
von Ahn: “Rosetta Stone at some point had 1,000 employees, and something like 800 were in a marketing role. This was not a language learning company. This was a marketing company that was marketing some stuff. They were not innovating. They were an excellent marketing company, but they were just not a great product company. The better product doesn’t always win, but in this case, I think the better product won.
One other thing we did really well is we rode the mobile wave perfectly. We came in at exactly the right time. This was not because we were so smart; this was by luck. We launched our iOS app on a Thursday and by Tuesday we were the number one education app. We remained the number one education app from then on. It just happened.
Now, it’s very hard to compete with us because we have so much more data. We can do much more intelligent things; the rich get richer phenomenon. If you go look at the charts, we’re number one, and more people download us because we’re more featured. It’s very hard to compete with that.”
GeekWire: What’s your secret sauce?
von Ahn: “We’re a product organization. We have 115 employees, and probably 90 are working on the product. That means we’re just making the thing better. We’re not spending money trying to sell it, we’re not spending money trying to dress it up or anything. We’re making the thing we do better. I think that has gotten us pretty far.
Staying true to our mission has helped us in two ways. One is being able to attract really good employees. Most of our employees have competing offers from very well known companies, like Facebook or Google. They choose us usually because of the mission. Also our users, a lot of them are big advocates for us because they see our mission and think Duolingo, they’re the good guys. Nobody thinks Rosetta Stone is the good guys. It’s not like they’re bad people, they just didn’t position themselves as being the good guys because their mission was ultimately make money. For some people that works, but that’s not an inspiring mission.”
GeekWire: How do you maintain your mission while growing the business?
von Ahn: “When Severin and I started, we believed in our mission pretty strongly. The first few employees that we got were people who only came to work here entirely for our mission. We couldn’t even pay them very good salaries. They turned down jobs that were much higher paid. Severin and I believed in our mission and those people were zealots. We have hired a number of people who were zealots about the mission.
If tomorrow I were to say, “All right guys, we did a little calculation and we’re going to charge people to use Duolingo.” I think half the company would quit. I just don’t think they’ll stick around.
Fortunately our board members and investors also have been very patient with us. Whenever we take on a new investment, very early on I tell them that we’re never going to charge for learning content. I tell them, ‘I don’t want it to be the case that two years from now in a board meeting you’re telling me you could make money if you charge for the app.’ They came in warned, they’ve been warned, and I think that has worked out pretty well.”
GeekWire: Has your experience with Duolingo made you think about education more broadly? Should the U.S. public school system do something different?
von Ahn: “We have been involved with education systems all throughout the world. Only 20 percent of our users are in the U.S.; 80 percent of our users are all over the world. I have a lot of conflicting things to say about education.
We have a partnership with the government of Colombia, where every school that is connected to the Internet is supposed to use Duolingo to teach English. That sounds more impressive than it actually is because only a small fraction of their schools are connected to the Internet. But there are a number of schools that are using Duolingo to teach English. As part of that, we had to work with their English teachers. We were having a lot of trouble and at some point we realized that English teachers in Colombia do not know English. We decided to test them with our own English test. The average score that they got was 20 percent. It’s hard to do that bad. These people knew no English.
The system is completely broken, and they know it. It depends on the country. The U.S. works relatively well; as always, you can improve it.
But yeah, there is a lot to say about education. Teachers are very underpaid. What is there to say? There’s a lot.”
GeekWire: What would you fix?
von Ahn: “I don’t know how to fix it, but I’ll tell you this. With Duolingo in the U.S., we teach languages to more people than the whole U.S. public school system. I don’t know this number, but my best guess is they probably spend $5 billion a year teaching languages. We teach to more people and in the U.S. our costs are probably $7 million. We’re 1,000 times cheaper.
I’m not saying we should replace all teachers; I think teachers are really useful. But you can get a lot of efficiencies if you start doing things like Duolingo because we are able to give students individualized attention. Even if it’s from an app, it is individualized, as opposed to a teacher who has 50 students and cannot give individualized attention a lot of times.”
GeekWire: This gets into a larger debate about technology and AI and replacing jobs. Duolingo’s apps use a lot of machine learning and AI, almost acting like a personal tutor.
von Ahn: “We’re not trying to replace teachers. That’s important to say. I’m very happy to replace shitty teachers. Shitty teachers are useless. English teachers in Colombia that don’t know English, they shouldn’t be teaching English.
Good teachers, I don’t want to replace them. They’re really good. We know this because we’ve done studies internally with public schools. Learning outcomes are better if you do Duolingo plus a teacher, than if you just do Duolingo or than if you just do the teacher. The combination is better. It’s because teachers can do all kinds of things that Duolingo can’t do. They can put things into context, they can answer very specific questions that the app just may not know how to answer.
Another thing that people don’t understand about the value of teachers, is that a human is really good at motivating. Just a frowny face from somebody that you respect can get you to work for weeks. No matter how much gamification we put into the app, we cannot get that amount of motivation. We put all kinds of little gamifications, we try to make it addictive. We’ve been pretty successful at it but, boy, good teachers are amazing motivators. That’s going to be hard to replace.
We have a whole school team and we try to work with schools. As far as we know in the U.S. we’re used by about 20 percent of language classrooms. We don’t want to replace the teachers. We think it is better to have the teacher there. But if somebody doesn’t have access to a teacher, we’d be happy for them to use Duolingo.”
GeekWire: Let’s talk about Pittsburgh. Why start the company here? Why have you stuck around Pittsburgh?
von Ahn: “I came to Pittsburgh because of Carnegie Mellon, to be a PhD student here. I’ve been able to build a pretty successful career here. I’ve never had a desire to leave.
When we started Duolingo, we were at Carnegie Mellon. There was a time when we were raising our third round of funding from Kleiner Perkins where there was talk about opening an office in San Francisco. They weren’t forcing us; our investors have been pretty patient with us. But since Kleiner Perkins was such a well known investor in Silicon Valley, we thought maybe we could make use of that and have an office in San Francisco. We thought about it a lot, but in the end it’s never been needed. We also sit there and think, what is the business case for doing it? It’s never been clear why we should do that.
By now, moving is impossible. People have kids, people have lives, significant others, friends. I don’t think we can move anymore and we don’t want to. I think we’ve done well.
The biggest problem we have here is hiring people for certain roles with previous startup experience. For example, if we are to look for a head of marketing, this person is not in Pittsburgh. There are some startups that have heads of marketing, but they’re much smaller than us and I don’t want to hire the head of marketing of a 10-person startup because it’s a different ballgame. The University of Pittsburgh Medical Center has a head of marketing, but I don’t want to hire them because they expect a marketing budget of a half a billion dollars and we don’t have that. Ideally I would like to hire the head of marketing of a startup that is a couple of years ahead of us. And there is nowhere to pick from here.
That’s the biggest issue we have in terms of Pittsburgh — for certain job roles, the person is just not here. Now you’re talking about moving somebody and wherever you are in the country, moving somebody is never easy, particularly these people in their mid-careers because they have families. We’ve had good luck moving people, but it just makes it harder.”
GeekWire: What’s your recruiting pitch to those people? Has it become easier over the past decade as Pittsburgh’s tech scene grows?
von Ahn: “Significantly. The pitch is come and see it. If they haven’t been here, they have an opinion of Pittsburgh that is much worse than it actually is. By the time they show up they’re like, ‘Wow, there’s five Ethiopian restaurants in a nearby area, I didn’t even realize that.’ The clincher usually is when you take them to go look at a house. When they see that the house is $400,000 and it has 3,500 square feet and half an acre of land and pretty nice, they think holy shit, that’s a third of the price of anywhere else.
It’s also the commute. You tell them, ‘you can live right next door. You can walk to the office and own your place. It will have a yard.’ For some people that’s very appealing. It has worked. People that are actually from Pittsburgh here at Duolingo, it’s probably only 15 percent. The other 85 percent have come from somewhere else. I think people like it.”
GeekWire: Is there something that scares you or keeps you up at night, whether it’s with Duolingo or just starting businesses or being an entrepreneur?
von Ahn: “I’m a little scared about AI. I’m mainly scared because I just don’t know what’s going to happen. I’m a little scared about xenophobia and the United States as a foreigner.
I know that over the next three years, we’re probably going to have to become a publicly-traded company. I’m a little scared about that just because I think there’s a lot of crap that I don’t particularly want to do that I’m going to have to do.
I’m sure that just over the last two weeks many companies that were about to start thinking about IPO just delayed it for six months. You just don’t know. We’re gearing up so that by 2020, we are IPO ready.”
GeekWire: What’s your advice for entrepreneurs?
von Ahn: “It’s really hard to give widespread advice because it’s so different. I would say the biggest failure point that I see are people who spend a lot of time finding reasons for why they can’t get started. ‘Nobody will fund my idea, that’s why I haven’t started.’ ‘I need to find a designer, that’s why I haven’t started.’ They just come up with a million reasons for why you haven’t started and I think the people that succeed the most just start. Just do it. It won’t be easy but you’ll be a hell of a lot more likely to get funding if you show up with something that’s actually done, as opposed to a blue sky idea. I would say just go forward and keep on going forward and don’t just sit there blaming the world. It’s very easy — the same thing happens to me. I come up with a million excuses for why I can’t work out today.”
GeekWire: You work out for 16 minutes every day, right?
von Ahn: “Yes, but it was very hard to get started. The first time I did that and then I delayed it for months because I was like, ‘I’m not going to start today.’ It was months until one day I said, ‘No, I’m going to do it.'”
GeekWire: A couple quick questions for you. You’re a big TV watcher — what do you have on these days?
von Ahn: “‘Halt and Catch Fire.’ It’s a pretty good show. It’s an AMC show like ‘Mad Men,’ a period drama, but the period is the mid-1980s with the PC revolution. It’s basically people trying to make PCs and nothing had been standardized yet. It’s like a combo of ‘Mad Men’ and ‘Silicon Valley’ at the same time.”
GeekWire: You’re a big gamer, too, right?
von Ahn: “I’m a softcore gamer, not the crazy hardcore gamer. I was playing the new Zelda game, but it was really time consuming so I gave it to the son of one of our employees. I got out of it.”
GeekWire: You’re too old for games now?
von Ahn: “They’re too much work.”
