How I Built This with Guy Raz - reCAPTCHA and Duolingo: Luis von Ahn

Episode Date: May 25, 2020

In 2000, Luis von Ahn was starting his PhD in computer science when he attended a talk and happened to learn about one of Yahoo's biggest problems: automated bots were signing up for millions of free Yahoo email accounts, and generating tons of spam. Luis' idea to solve this problem became CAPTCHA, the squiggly letters we type into a website to prove we're human. He gave away that idea for free, but years later, that same idea had evolved into a new way to monetize language learning on the web, and became Duolingo. Today, the popular app is valued at $1.5 billion, and is seeing a big spike in growth while people are confined to their homes. See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

This podcast is brought to you by Squarespace. I talk to entrepreneurs all the time who are looking for a way to upgrade their digital footprint. Well, whether you're just starting out or you're scaling your business, Squarespace is the easiest way to build a great website that stands out. It's an all-in-one website platform that gives you everything you need to claim your domain, showcase your products, and get paid. Anyone can use Squarespace's cutting-edge design tools to build an online presence that truly reflects what makes your business special.

There are templates, intuitive drag-and-drop editing, and even an AI-enhanced website builder. Then, Squarespace's built-in analytics tools help you make smarter business decisions. Review website traffic, learn where to focus engagement, and track revenue all in one place. Looking to grow your business, Squarespace even offers fast, easy business financing

through Squarespace capital. Go to Squarespace.com slash built for a free trial. And when you're ready to launch, use offer code built to save 10% off your first purchase of a website or domain. Loans issued by Celtic Bank and serviced by Stripe, all loans subject to credit approval. This show is in partnership with Airbnb. This past summer, I took my family to Vienna, and it was incredible. We spent our days wandering the old streets, stopping for coffee and pastries, visiting museums, and just soaking up the history of one of the most beautiful cities in the world. And one of the things that made the trip so special was the home we booked on Airbnb. It had tall windows, beautiful old details, and plenty of space for all of us. And being in that home on Airbnb, right in the middle of Vienna, walking distance from so much of the city,

made it feel less like a visit and more like we were actually living there. Plus, taking a trip is the perfect time to host your space on Airbnb. your place with all of its personal touches and its amazing location could make someone else's vacation even better. Your home might be worth more than you think. Find out how much at Airbnb.ca You know, I was feeling good. I thought, okay, well, we can continue raising money for a while. And at some point, we raised money from Google. Capital G is one of their investment arms. and the partner from that firm sat me down and she said, nobody else will invest in this company if you don't figure out how to make money.

You know, you're not going to find a bigger fool. We're the biggest fool. And that's when we started kind of freaking out and trying to figure out how we're going to make money. But we had no idea how we were going to make money. From NPR, it's how I built this, a show about innovators, entrepreneurs, idealists, and the stories behind the movements they built. I'm Guy Raz, and on the show today, how a perfect storm of skill and luck drove Luis von

On to create Captcha, then Recapture, and then Duolingo, a foreign language app valued at $1.5 billion. Think about the small moments or decisions in your life that actually had a huge impact on how your life turned out. Maybe it was a conversation you struck up with the person next to you on an airplane. Maybe it was a party you were reluctant. went to only to meet the person you'd eventually marry. Or maybe it was a decision to stay on vacation an extra day that sparked a new idea. For Kevin Sistram, it was a random remark from his girlfriend that made him decide to use filters on Instagram. For Blake Mikoski, it was a chance meeting with a group of young Argentinians who took him to the countryside where he saw kids with no shoes that one day inspired him to create Tom's.

And for Luis von On, it was a free lecture at Carnegie Mellon University in 2000. We'll get deeper into the story in a few minutes, but that single lecture would lead him to invent two ingenious new tools. The first was CAPTCHA. Yes, CAPTCHA's, those annoying, twisted and blurred letters you have to type into a website to prove you're human. And the second one was Duolingo, now the biggest language learning app in the world, which is now getting even more popular because people are looking for new things to do now that they're stuck at home. Both CAPTCHA and Duolingo were designed to harness the power of crowdsourcing to solve problems. And I'm going to blow your mind here.

If you have ever typed in a CAPTCHA or used Duolingo, there's a good chance you've taken part in a massive online collaboration that you probably weren't even aware of. And it's amazing how Luis came up with all this. but let's start at the beginning. Luis was born in Guatemala in the late 1970s. Both his parents were doctors, and though he was surrounded by poverty and violence in Guatemala City, Luis grew up in comparative privilege.

And as a kid, he spent a lot of time hanging out at the family business. My mother's family actually had a candy factory. Everybody is always amazed at the fact that I grew up with a candy factory, and they think that it was like Willy Wonka or something. I was not all that much into the candy itself. I was into the machines because basically the candy is made by these gigantic machines that pump out, I don't know how many thousands of pieces of candy per hour. And basically all my weekends I spent playing at the candy factory

and I would take the machines apart and put them back together. There would be some extra pieces after I put them back together usually, and that would be a problem. What kind of student were you? Was school pretty easy for you? Yeah, I was pretty nerdy. Basically, I was really good at math. Math was just easy to me. What I would do in the summers is basically get either next years or a couple years later math books and basically do all the exercises.

Wow. It kind of came easy, but the way I really got good at it is by doing hundreds and hundreds of exercises. That's what you would do in the summertime? Yeah, I kind of was bored. I mean, I was an only child. I didn't have that much to do. This is, remember, this is also pre-internet, pre-everything. So what was I going to do?

And man, that's what I did. I was putting playing cards in the spokes of my bicycle and buying Jolly Ranchers at 7-Eleven. Woo! Should have done math books. So you were, did you just love math? I mean, it sounds like kids don't think about their future. They're not like, I'm going to study math so I can be in tech one day. Like, you must have really enjoyed it.

I did. I enjoyed it. It was like a puzzle for me. By the way, this is not the only thing I did. I mean, I also played a lot of video games. Pirated video games in my Commodore 64. Like floppy disks on floppy disks? Floppy disks, yeah, that's right.

I wanted a Nintendo when I was eight. My mother would not get me Nintendo. She instead got me a computer, a Commodore 64, and I couldn't figure out how to use it, but eventually I read like the manual and stuff, and I figured out how to use it more, and then I figured out I could pirate other people's video games, and so I became a little hub in my little neighborhood.

But these were not other kids, these were adults or basically young adults who had a computer, and they would come to my house, and I would take their games and give them my games in exchange. So then I collected a pretty large number of video games. But I should mention, right, that, I mean, because your childhood sounds pretty nice, but like as a kid, I guess, or even as a teenager, there was a civil war in Guatemala, right?

I mean, we know that today there's a lot of violence there. And obviously there's violence in the U.S. and other countries too, but Guatemala has been particularly hard hit. I mean, did it feel dangerous when you were a kid? Yes, it did. There was a civil war pretty much since I was born in 79 to 1996. There was a civil war going on the whole time. It always felt dangerous.

When I was at 15 or so, my aunt was kidnapped for ransom. I mean, she was gone for seven or eight days. People's cars would be stolen. Every couple of months, somebody's car would be stolen in my family. Going out past 7.30 p.m. was rare. You needed to go out in a large group if you were going to go up past 7.30 p.m. And, you know, I did him. My house had walls and barbed wire.

And, yeah, it felt dangerous. I mean, this is one of the reasons I came to the U.S. actually. I mean, you know, when I was, after my aunt was kidnapped, I thought to myself, I don't want to live here. Yeah. And I guess you did end up leaving Guatemala for college because you went to Duke in North Carolina. And you described yourself as a, like a math nerd in school. And is that what you intended to do, like to do something? something in math? That's what I wanted to become. A math professor. I was pretty certain I wanted to

become a math professor. You know, at the time, I thought the best thing that I can do is really learn a lot of math, and I really loved it, and I thought it was futile to learn how to deal with other people. And it is interesting because my job these days is 100% just dealing with other people and people's problems. Wait, I'm just trying to understand this. So by becoming a math professor, you thought, hey, I wouldn't have to deal with people. I would just deal with people. I would just deal with like facts and data and numbers. Yes, yes. And, you know, I'll do math research all day long.

And every now and then I'll teach a class, but whatever, that's like a tax. That's what I thought. So, all right. So you get your degree and you're following this path to go into academia and you go into a PhD program at Carnegie Mellon. Correct. And I guess you go into computer science, right? Yes.

I changed from math to computer science because I visited a math grad school. And what people were saying, the professor was saying, Oh, I'm working on this open problem that nobody's been able to solve for the last 300 years. And I thought, hmm, I don't think I'm smart enough. I mean, if you haven't done it and nobody's done it in 300 years, that's kind of not for me. Whereas when you visit in computer science, I mean, this is crazy thing. People are like, oh, yeah, I solved an open problem yesterday. Wow.

It's a much younger field. Yeah. And so I thought that was much more exciting for me, at least. So you are, you start, this is year 2000. You start your PhD program at Carnegie Mellon. And I guess, like, really soon after you start. you go to some talk by someone from Yahoo! Who comes to campus to talk about Yahoo! Yahoo!

Yahoo is a big deal in 2000. What's the story? Yeah, that was serendipity again. Most of the things that I've happened in my life are serendipitous. I was a first-year PhD student. I had been at Carnegie Mellon for maybe two months. You know, the first thing you've got to do when you become a PhD student is find an advisor. I had found myself an advisor.

And we went to a talk together. This guy from Yahoo, who was the chief scientist of Yahoo, at the time. And like you said, Yahoo at the time was the biggest tech company out there. And he came to give a talk at Carnegie Mellon. And the talk was basically 10 problems that they didn't know how to solve at Yahoo. And I was an enterprising PhD student and I thought I'm going to try to solve these problems. And what was the problem that he said they had? Yeah, so the problem for this particular one was there were people who were writing programs to obtain millions of free email accounts. So Yahoo at the time gave out email accounts for free.

And some people thought it would be good to send spam from Yahoo accounts. The problem is each Yahoo account only allowed you to send like 500 messages a day. So if you want to send 10 million spam messages per day, you just have to get a bunch of Yahoo accounts. And from each one of them, you send 500 messages. And this is the era, and I remember this, and we all remember this, where you would get hundreds of messages, spam messages about certain bodily enhancements. You would get messages about, you would get messages about, the hormone growth things.

I mean, baldness. I mean, it was... Canadian pharmacy selling you by Agra. Right. Yeah, yeah, yeah. Just spam. It was a huge problem. Yep.

And he was like, we can't figure out how to stop these computers or programmers from making, you know, just creating all these email accounts. Yep. And so that was the problem.

And just to be clear, these spammers were not physically setting up each individual email account. No, they had written programs that would just set up millions of email accounts a day or however many per day. So I thought about it for weeks and months and I talked about it with my PhD advisor, Manuel.

And together we came up with the solution, which was the thinking here is, look, computer programs can get a million email accounts per day because, well, computer programs can do things very fast. Yeah. And it also doesn't get bored. Right. Humans, human can't get very many email accounts. So how about this? How about if we make sure that whatever is getting an email account is, actually a human and not a computer program.

Then we started thinking, okay, well, how can we distinguish between a human and a computer? Yeah. And that's where this idea came about. It's this idea of a CAPTCHA, where it's basically these distorted characters that you have to type whenever you're buying tickets on ticket master or getting, you know, an email account or stuff like that. The idea is that a computer can generate one of these, you know, basically take some letters, put them in an image, distort them.

and then it can give them out, and it turns out that humans can read these. Well, at the time, humans could read these very well. They still can't, but computers could not. By the way, I can't read them. I'm the human that can't read them. I can't read catches. I don't know how much of a human you are.

Now they're like point to every stop sign. Point to every bridge. Yeah, it's changed. But basically, at the time, the idea was humans can read these distorted characters much better than computers. So let's give basically, every time that somebody's trying to get an email account, let's give them a test to see if they're a human or not.

And you just start doing this on your own? Like, did you tell Yahoo? Did you tell him? Or were you just like, this is fun? I just want to figure this out. Like, just working independently? Yeah, I mean, with my PhD advisor. It became our research.

It became our research subject. And, yeah, we did not tell Yahoo. Because it sounds like here's the kid getting the mathematics workbooks in the summer. It's like, you just seemed like a fun thing to solve. Yeah, it was fun. But that's like the core of the PhD program, I think. You know, you're trying to find problems that others haven't solved. I got you.

And you're, this seems like a problem that other people hadn't solved. So we were working on it. And you called it CAPTCHA. Did you name it CAPTCHA? And what does that mean? It's an acronym. Sounds like CAPTCHA to like capture. Yeah.

Yes. It's an acronym that sounds like capture, which, you know, the idea was to capture the, you know, the bot or the computer program. It stands for completely automated public touring test to tell computers and humans apart. So, you know, it was just an acronym. But now, I think most people have heard of the word CAPTCHA, but do not know what it stands for. And you come up with this thing, CAPTCHA. And what do you do? Do you email this guy Yahoo and say, hey, I've got your solution for you?

Basically, I mean, we emailed them and we said, hey, you know, we think we have a solution. Here's some code. You know, and we explain how it works. And what was amazing, and now that I know how large companies operate and how slow things move, what was amazing is within about a week of when we sent them this, it was already functional. Wow. You know, alive. It was live. So I guess the problem that they had was so big that when they saw this, they were like,

okay, let's do it. Wow. So they saw this. They're like, oh, my God, this is great. And so you must have just, I mean, you were like, what, 25, 26? 21, 21, 21. My God. So you, right, all of a sudden you're an overnight millionaire, multi-millionaire.

No, no, no, no, no, no money exchanged hands. Wait, you gave Kapshet to them? You just said, hey, here you go. Yep. I mean, listen. You gave it to a multi-billion-dollar company, just gave them CAPTCHA? I didn't know how things worked. And, you know, the truth of the matter is this is probably a good thing in retrospect.

I mean, who knows what would have happened. But, you know, I don't have regrets. But yes, that was, that's what started happening. And then basically every other website started, you know, they saw that it was in the front page of Yale. So every other website started copying or making their own version. And within a couple of years, pretty much everywhere on the web, you know, when you had to get an email account when you had to enter a comment on a blog, where you had to buy tickets on Ticketmaster, you had to enter one of these.

You were the guy. You were the guy who made it really hard for dopes like me to get it right. I had to constantly... It was annoying to people. And every time that I would be at a party or something, and they would ask me what I did, and then they would get me to explain my research. I would explain that I had done that. And I would say, what the hell?

They would kick you out of the party. Screw you. Screw you. But some of the... the research you did. I mean, you did actually write some software that you sold for, I think, for some money this time. I mean, just like, could just briefly explain how that happened and what you were working on? Yeah, I mean, this is, it's funny. I mean, at the time, this was not called crowdsourcing, but my PhD research was basically on crowdsourcing. I had worked on

capture where the idea was that humans can do some things computers cannot. Yeah. So my PhD research was basically that finding things that computers could not yet do, and then finding ways to get people to do them. And in particular, the main thing that I did in that time was a game, and people loved to play it. But as they were playing it, they were actually helping computers figure out what's inside an image. How did the game work? The way the game work was this. You went to a website and you got randomly paired with somebody else who had shown up to the website. Got it. And you were both shown the same image. Any image, a random image, like an elephant.

An image from the internet, yes. And you were told, type whatever the other person's type. Just start typing words. And if one of your words matches one of their words, you both get points. Got it. What was interesting about this game is that the words that people typed, the only thing they had in common was the image. So the words that people typed were basically related to the image. Because if you're told to do this, what are you going to type? If it's a picture of an elephant, you're going to type elephant because you think the other person is also typing elephant.

So those words were really good labels for the image. We were really good tags for the image, basically. So I put it online, I programmed that millions of people played it, and this was a game that was helping label all images on the web. And at some point, actually, Google, in fact, bought it. They just bought the technology, and then it got implemented. They changed the name from the ESP game, which I call it the ESP game, because it's like extrasensory perception. The idea was like tried to think what the other person was thinking to the very amazing name of the Google Image Labeler, which is much less interesting. And so they bought this from you, and was that like life-changing money?

I mean, were you all of a sudden super rich? No, it was not super rich, but it was good. Got it. Okay, so you went on to earn your Ph.D. And you get a job at Carnegie Mellon. Correct. You're a young professor. You're on tenure track.

And by the way, early on in your academic career, I should mention this, you were granted a very, like one of the most prestigious awards, the MacArthur Grant, known as a genius grant, basically they pick a few people a year and give them like half a million dollars of unrestricted money to work on whatever they want. I mean, it must have been mind-blowing. Yeah, that was, you know, I've gotten a number of awards in my life,

but I mean, that was a particular, that was probably the most impactful one in terms of just how I felt. And also, they do this crazy thing where they don't, you have no idea that you're being considered or anything, and they call you randomly. Yeah. They figure out your phone number one that you pick up, and then they just tell you, have you ever heard of the MacArthur Fellowship?

And, you know, you say yes. And then they just tell you. And you're like, wow, what the hell? First, you think it's a prank. I thought it was a prank. Yeah. And I guess, like, around this time, too, I don't know if this is apocryphal or not, but I read that Bill Gates personally called you to try to convince you to leave your job at Carnegie Mellon and to go work for Microsoft. Is that true?

That is true. You pick up your cell phone. It's like, please hold for Mr. Gates? Or was like, he wanted to lie? It was pretty much like that. It was pretty much like that. I mean, it was somebody, you know, basically said that. Please hold for Mr. Gates.

And you were like, okay. And he gets on the phone. And he's like, Luis, it's Bill Gates here. That's exactly right. I had been an intern. I had been like a year and a half before that. I had been an intern at Microsoft. And so I had good ties with Microsoft.

And they really wanted to hire me for Microsoft. research. And so that is, you know, Bill Gates spent, I don't know, 45 minutes to an hour trying to convince me to go there. Wow. I was very flattered. But, you know, at the time, I just wanted to do my own thing, be a professor. How do you say no to Bill Gates? You were like, oh, wow, thank Mr. Gates. It's really pretty so nice. I'm so honored, but I just, you know, no. I didn't actually say no. I said I would think about it. And then, and then I said no to, like, a recruiter. But, yeah, it was not easy. I mean, he's a major hero of mine. I mean, he's just an amazing human being. But I just wanted to be a professor at the time. I thought, I want to do

my own thing. I want to be a professor. And that's, that's what I wanted to do. Yeah. Was money important to you? I mean, you know, was it important to you to? It mattered. And interestingly, I guess it mattered more than it does now. It was nice to know that, you know, I could buy a nicer car or get a nicer apartment or something. So it's nice, but it's never been my driving force, I would say. Okay. So you decide I'm not, Bill Gates, thank you, no thanks. I'm not going to Microsoft. I'm going to stay doing this work. So you are now, were you lecturing? Were you teaching classes?

I was teaching. I was teaching a huge class. It was called Great Theoretical Ideas in Computer Science. It's just a fancy name. It's basically a discrete math class. The people who don't do well there usually change majors. And you're like, not like John Nash, but I'm thinking about that movie. You're in front of like a big blackboard or a whiteboard and doing equations and problems in front of 250 kids. Yeah, slides had already been around.

So I use PowerPoint. I use PowerPoint. It's not that old. I got you. Yeah. You're right. You're right. Okay.

So you're teaching and you had done CAPTCHA and lots of people are saying, hey, it's annoying. And you're like, yep, I hear you. And people kept saying that to you, like, kept hearing that? Yep. Yep. And that's when I started thinking, each time you do it, you waste about 10 seconds of your time. And I just started, this thought just came to me of, wow, how much time am I wasting of humanity? And, you know, I did a little back of the envelope calculation and realized about 200 million

times a day somebody types a capture and that's when you get 200 million times a day times 10 seconds so humanity as a whole is wasting about 500,000 hours a day typing these captias and that's when I started thinking huh related to you know all my research that I have done is there something we can get humans to do while they're typing a capture that is also useful and we got them to do something useful yeah and it occurred to me that yeah we could get them to help digitize books how did sorry pause how did you come to that idea. So, like, when you type in a random letters from a CAPTCHA, you were thinking, wait, how do we have a secondary use for this? And maybe it's books, digitizing books. How did that even

occur to you? I was thinking for a long time, can we get people to do something, but at the same time kind of extract valuable computational effort from it until it finally occurred to me, oh, there's all these projects out there trying to digitize all the world's books. Google had announced a really At the time, it seemed kind of insane and ambitious project that we're going to take all of the world's books. Everything that's ever been written is 100 million books ever written, and we're going to put them online. This is what we're going to do. And now in this process, the way the process works is you take a book, a physical book, then you scan every page, and then the computer needs to be able to decipher all of the words in those photographs.

Now, the problem is that computers cannot record, or at the time could not recognize many of the words. About 30% of the words computers could not recognize. And the reason for that, by the way, which is this amazing thing, is the same reason why computers could not read CAPTCHAs. Basically, this looked kind of like distorted letters. And distorted letters are just inherently difficult for computers to recognize. Like that's the point of CAPTCHA. That's right.

And so it occurred to me, oh, wow, what if we just take the words out of a book that the computer could not read, and then we send them to people while they're typing CAPTCHAs on the Internet. So basically, next time you type a CAPTCHA, the words that you see are actually coming from a book. and whatever you enter is in fact used to help digitize a book. Do you remember when you came upon that idea? Was it like in a flash or was it just like, did it occur to like? I do remember. I was driving from Washington, D.C. back to Pittsburgh because I had been to some

National Science Foundation panel or something like that. Then I spent the next few weeks just kind of going over it and going over it and I thought, okay, this will work. And, you know, I started building the system, but then I recruited a guy. a guy who was, his name is Ben Moore, he was maybe a freshman in computer science at Carnegie Mellon. He was one of my students. He was in my class, in my large class. He was the best student.

And I thought, you, you can help me build this. And so together we built a system that would take a scan of a book and extract all the words that the computer could not recognize and then send them as a catch. Now, we decided we're going to go to big websites and we're going to tell them, hey, look, we have. have this capture service. It's free. It's better than yours. Yours is relatively crappy. This one's pretty good and we'll give it to you for free. But all we have to be able to do is see what your users can type, our typing during this so that we get to digitize the books. And did websites start to use this tool? You know, some small websites started using it. The first largest, it wasn't even large, the largest website for a while that was using us was called Online Booty Call.

Nice. And they were to get an account in online booty call, you had to type a CAPTCHA. And it was our CAPTCHA. And every time every person that was getting an account in online booty call was starting to help digitize stuff. God bless them. God bless them indeed. And at some point, a website that was relatively new at the time decided they needed to put a capture on their registration flow.

And they called us up. And it was a website called Facebook.com. And they were relatively new, and they were like, hey, listen, you know, we saw, we came across it. It seems like you have a good capture. Can we, you know, can we put it up? This in 2007. Probably 2006. Yeah.

You know, we said, okay, cool. And they did it. And that website was growing very fast. And so pretty soon we just had a ton of people using our thing. And then at some point I was giving a talk somewhere in Dallas. about this system and how we could digitize books, et cetera. And the guy who was the CTO of the New York Times was in the talk. And he said to me, hey, you know what, we have this archive of 130 years of content from the New York Times

that we have not been able to digitize particularly because computers cannot recognize the words. How about you digitize it for us? And I'm just out of curiosity. Eventually, I'm assuming you thought this could make money. We thought, we thought, but we really didn't. We kind of thought we could make money, but we didn't know how. And what happened is this guy, you know, he then sent me an email. He said, no, I was serious.

Let's do it. How much would you charge me to do this? Charge them to do what? To help digitize their content. The old New York Times. The entirety of the New York Times archive. And so, you know, we kind of scramble and we had no idea how much to charge them. And then we just thought to ourselves, okay, how much would this person have to pay people to digitize?

Because the only other solution is people. Yeah. So we came up with something and we said, okay, divide that by four. Yeah. And that's what we'll charge you. And the number we came up with was for every year of content, we will charge you $42,000. For every year of content in the Times, okay? Yes.

If you take one year, all editions of that year, we will charge you $42,000 to digitize that whole thing. That's the number we came up with. And then they said yes. And how many years of that's like 120, 30 years of? Yes. This is several million dollars of a contract there. But they said, okay, we'll pay you per year. Let's do one year, then another year, then let's see what happens.

And so they sent us the first year. We did it. And it turns out because Facebook was already using our capture, we actually were digitizing relatively fast. Like how quickly would it take to digitize a year? A year was being done in about a week. What? A week of people typing in captions could digitize a whole year of the New York Times? Yes.

So that's what was going on. I'm just, here's a question, right? Which is, say I was getting on Facebook in 2000. This is like 2008, I think. when the Times approaches you, right? Yep. Yep. So I'm on Facebook in 2008, and I sign up for it, and I've got to write a CAPTCHA, and the word that I've got to decipher is anxiety.

Mm-hmm. One of my favorite words. And I type that in, but how do you know, how does CAPTCHA know that I did that accurately? That's a very good question. So what happens is for you, we actually give you two words. We give you, one is the word anxiety. It's a new word that we just got out of whatever the New York Times. And then the other word is a word that we already have digitized.

We already know what the answer is for the other word. And then we don't tell you which one's which. We just tell you, please type both. Yep. And if you type the correct word for the one for which we already know the answer, we assume you're a human, and then we also get some confidence that you type the other word correctly. Got it. And if we give this new word to like 10 different people and they all type the same, you know, the same word,

they all type anxiety, then we agree. Then you know it's right. Okay. Then we know it's right. So your accuracy was pretty high. Very high, yeah. Wow. So within a week or so, you'd get this first year of the New York Times digitized, and then presumably they're like, okay, let's keep going. That's right. And so what started happening is they started sending us checks for $42,000 coming in pretty quickly.

Nice. And by the way, we had no company. We had nothing. And at some point, Carnegie Mellon got wind of this. And they're like, wait a second. You're kind of running a company here, but not really, because there's not even a company. You've got to get out of here. Yeah. They had a standard deal that says, okay, if you're going to start a company based on. on something that you came up in your research, we'll take 5% of this company. And so that's what happened. So we just said, okay, how about you keep 5%.

We'll start a company. And so we did. We started a company. I got a lawyer. We formed a company, and pretty soon we were making, you know, $42,000 every few days. And did you call it Recapture? It was Recapture, yes. And did you leave Carnegie Mellon or did you stay on as a better?

No, I stayed as a professor. This was really a side hustle. I stayed as a professor, Ben, who was also working on it, stayed as an undergrad. By then he was like a sophomore. He just stayed as an undergrad. I stayed as a professor. It was the side hustle.

But it was making quite a bit of money for us not doing all that much in there. It's incredible. I mean, you basically have a side hustle where all of the work is being done for you by people who aren't, don't even know and wouldn't really care because it takes them two seconds and they're getting a free Facebook account. Meantime, they're digitizing the Times and you're getting $42,000 checks every couple of days from the Times. Yep. That was the, that's what was going on.

It's like one of the most genius businesses ever. Like, can we, can you, we talk after this interview is over to figure out a thing like this? This is amazing. You don't, you don't have any employees. It's just incredible. Yeah, it was pretty good. But then, you know, that went on for some time and then actually, Google, in fact, bought it for their own book digitization process. Wait, hold on.

They approached you, they saw this, and they saw that this was really, and they, and they said to you, hey, we can use this for our books digitization. Yes. And what they threw a number on the table? Yes. I mean, multiple numbers were thrown in and out. In the end, we decided this was a good home for it. So we sold it to Google.

And this time I did go to Google for two years. So I took a leave off of being a professor. Okay. So you were running this company, this small business. Google buys it from you, which is just incredible. you're still at this point really young. This is really life-changing. Yeah, especially since we didn't have investors or anything.

It's all the money gig went to us. And your startup costs were relatively low. Yeah. Wow. All right, so you get acquired by Google. You are still a professor at Carnegie Mellon in Pittsburgh, but you went to California to Silicon Valley to go work at Google? Google happens to have an office here in Pittsburgh.

Ah, okay. And so I was spending about half time here in Pittsburgh and a half time in California. And what were you doing at Google? Were you still just basically running ReCAPTCHA? Well, the first order of business was to integrate Recapture into the Google infrastructure. That took about a year. But I was still a professor.

And even though I was on leave, I still had PhD students. And towards the end of that first year, I started getting very, very interested in the project, which happened to turn into Duolingo. And presumably, Luis, I mean, given that you'd already turned down Microsoft like four years earlier, you were not going to become a Googler. You really, I mean, this is part of the deal. You had to go work with Google, make this happen. But it sounds to me like you're so restless.

You were not going to just work in a big organization for the rest of you. I wasn't going to do that. Yeah, it's not my thing. I mean, I kind of have to do my own thing. What do you think the reason for that is? I don't know. I mean, you know, probably because I'm an only child, I don't know. Are you just restless?

I'm restless and I'm also obsessive. I obsessed on one thing, and this is what happened with Google in the end. I mean, I was starting to get obsessed on this new project, Duolingo, and I just had to leave. And by the way, leaving cost me quite a bit of money, because some of the recapture payment required that I stay there for three years. Yeah, and I did not stay for three years. So you had to give up some of the money. But in the meantime, you're still at Carnegie Mellon teaching. And I guess you've got a graduate student there who would eventually become your co-founder.

at Duolingo, and I think his name is Severin Hacker. Is that right? Severin Hacker. Last name is Hacker. I mean, it's like a movie name. His name is Severn Hacker. And by the way, the way I met him, I was literally sitting in my office one day,

kind of minding my own business. And then this lanky guy shows up, super tall and super skinny. And he says, hi, I'm Severn Hacker. And I just said, what? They didn't even parse it. And then he just did hand gesture. Severin, he did a gesture that is like basically cutting his arm.

Yeah. As if he was severing it. Yeah. And then Hacker, he just did like as if he was like moving his fingers, like typing with the computer. And then I'm like, oh, Severn Hacker. Wow. And he said, yeah, I'm here and I would like to do research with you, kind of work with you.

And I said, yeah, sure. Your name's so amazing. I'll do it. When we come back in just a moment, how Luis and yes, Severin Hacker, take the recapture business model and try to use it again to teach people foreign languages for free. And how they discover that particular model is just not going to work. Stay with us. I'm Guy Raz, and you're listening to How I Built This from NPR. Hey, welcome back to How I Built This from NPR. I'm Guy Raz.

So it's around 2012, and Luis has been working at Google for two years, and he's got one more year to go to get fully paid out. but he's itching to get going on his next project. So he walks away from Google early. Interestingly, what motivates me started changing quite a bit after the recapture acquisition because I really thought, okay, I'm in a really fortunate position in my life. I've now, you know, I don't particularly need to worry about money.

Can I do something that helps people? And so I really started thinking about education. And Severin, fortunately, was also very much into education. So we started thinking, can we do something? something with education that gives it away for free. And the thinking really was, there's all these people that I grew up with in Guatemala, very poor. And a lot of people talk about education as something that brings equality to different social classes. But I always thought as the opposite, something that brings inequality, because what happened, at least, you know, in my case,

those who have money can buy themselves the best education in the world. And those who don't barely learn how to read and write. But you're thinking along the lines of, like, what is something that can be free, but that would kind of be self-generating. So for example, many business models offer a free service with advertising. That's Facebook's model. That's Google's model. They capture your data, which you get this service in return. But you were thinking, how can I create something that doesn't cost people anything, but that can be sustainable? That's what I was thinking. I was thinking, can we make it sustainable and can we make it free? And languages in particular changed. I mean, Duolingo is a way to learn languages. We ended up.

going to languages, particularly also because in both of our cases, we're not native English speakers. Severin, it was not either? No. Where's he from? He's from Switzerland. He's a Swiss-German native speaker. And so in both of our cases, learning English completely changed our lives. And I just knew, you know, when I was growing up, everybody in Guatemala wants to learn English.

Nobody could afford it. And it turns out that in most countries in the world, if you know English, you can double your income potential. So we thought, okay, can we figure out a way to teach people, language in particular, teach English in a way that's free but also self-sustaining? Like, how can we make it so that it kind of pays for itself? Okay, so you've got this language. You think language is where it's going to be at because that's how we can actually help people increase their income, for example, especially in other countries. And then the next question is, okay, what does that thing look like? Yep, that's what we were thinking.

and then eventually we came up with this idea, which is pretty similar to recapture. Turns out today, dualingo does not work this way. It is a good idea, but it is a relatively impractical idea. So what was it? So the idea was this. We're going to give the service entirely for free, but instead of having people pay us,

we're going to get them to help us translate stuff. So, look, there are all these companies that want to translate stuff. For example, CNN would like to translate all their news that they publish from English to Spanish. they're currently paying people to do that translation. We thought, okay, what if we get these people who are learning English on our platform, what if as a part of their way to practice, you know, after they learn some stuff, we tell them, hey, here's the CNN article, can you translate it?

It's in English, you're learning English. Can you translate it into Spanish? And then if we get multiple people to translate the same thing and they kind of collaborate with each other to make a translation, at the end we'll get a translation. and we thought, okay, we could sell that translation back to somebody like CNN and make money that way. And you would use the same technique as you used with recapture, like if 10 people got the same exact sentence, you knew it was right? It had to be a little smarter than that. It turns out if you give a sentence to 10 people, they'll all translate it differently.

Okay, yeah, right. But generally, yes. I mean, we would use, you know, kind of they had to be pretty close to each other. Then we also had another step where the exercise to the person wasn't translate this, but tell us whether they're. this translation is right or wrong. So some people would check the translations. But in the end, we built the system, actually, but worked pretty well that basically you could give it a text in English.

There were people who were learning English or who could be learning English, and then they would help translate it, and then the system would pop out a translation that was made by people, and the translation was very good. It's instead of CNN paying some translation service, they would just pay you. and you, again, like the New York Times sending you $42,000 objects, you would just have your students learning a language,

but they would also be translating these articles into their native language. That's exactly what the idea was. So, okay, so you have this really great idea. By the way, and I'm assuming you're funding the whole project, right? Because... Yeah, it was a Carnegie Mellon project, and it was being funded by a grant from the National Science Foundation and also from my MacArthur grant at Carnegie Mellon.

Got it. And funding at the time what it meant is basically, paying for Severin's salary. Yeah. And so at what point did you say to Severin, all right, let's spin this out and start a business and figure out who, you know. This was also an interesting thing. We, okay, we thought, okay, let's do this.

This is a good idea. We're convinced this is a good idea. Let's do this. But the first thing when you do is hire more people because the two of us just can't do it. So we started hiring people, but all inside Carnegie Mellon, we realized actually hiring people was pretty expensive inside a university. And then we thought, okay, well, what I need. need to do is apply for a bigger grant from a National Science Foundation.

And I started working on that. And these grant applications is like 30 pages long and you have to wait like a year to hear whether you get it or not. And it's like a couple million dollars maybe. And so I was working on that. And then suddenly I got connected to a venture capital firm, Union Square Ventures. And, you know, I talked to them about this and very quickly they get back to me and they're like, hey, we could fund this.

I didn't have to write 30 pages. It was much easier to get that money. So we thought, okay, well. They wanted in on this because, well, first of all, you had a track record. It was mainly the track record. In retrospect, you know, I now have a relatively good relationship, like close relationship with the Unisquare Fulks. I mean, they've now told me, look, we'd never thought this would work.

But you had a really good track record. So why not? They want to invest in you. Right. So when they approach you, they said, we want to fund this, we want to invest in this. And at this point, you had not dealt with outside investors in your career. We had not. And also, this, this was still not a company.

So at that time, we just thought, okay, well, let's make it a company and let's take some funding. Yeah. And how did you do that? Did you, was there, I mean, Severn was your student? So was there an uncomfortable conversation or was it pretty clear? You're like, listen, it was my idea. I'm the boss, so I'm going to have X percentage. You know, Severn is just, I think it's partly because he's Swiss or because he's like a computer.

He's like Mr. Spock. he had that conversation with me. I didn't have it. He sat down and he said, I want us to write a contract, but it's not a lawyer contract. I just want to write basically a word document that we write a few lines long

that just basically says what our understanding is. And we're both going to sign it. And we wrote down a contract that was just really simple. It's like we'll go in it half and half. And all decisions related to hiring, we're going to do it together, all decisions related to.

to, I don't even remember what else, but there's like three or four bullet points. And then it just said his name and my name and we signed it, and that was it. So, all right, so you create a company and you've got some interested investors. And I think you raised like three plus million dollars from a bunch of different investors, right? It was 3.3 million, which at the time, this was a normal Series A. So it's $3 million from Union Square Ventures and $300,000 from a combination of Tim Ferriss. and Ashton Coucher. Ashton Coucher.

He was like way ahead of the rest of Hollywood. Now everybody wants to be Ashton Coucher. Ashton Coocher was a very smart guy. Very smart guy. I mean, yeah. He was way ahead of the game. I mean, everybody now wants to be an Ashton Coucher. They're like, right?

It's like there's LeBron James and there's Serena Williams and there's, anyway. All right, so Ashton Cusor, with that money, did you get an office? Did you get a staff? How many, did we able to hire a bunch of people? Yeah, we hired a few people. We did get an office very close to the Carnegieville and Cambridgeville and And we started working on it.

And at the time, pretty much everybody told us, you're not going to be able to beat Rosetta Stone. Like, why would anybody use this? We just kept on saying, well, because it's free. This is going to be free. That was our main selling point. We didn't know if it was going to work. But we worked on it and we launched. We launched Duolingo.com in 2012.

And it was very fortunate that I had given a TED talk about this. As soon as that went online, we got a ton of people come to Duolingo.com. To try it out. To try it out. Yep. And it started growing pretty fast because it was a free way to learn a language. And at the end of your lessons, we would say, hey, here's this article. You want to help us translate it.

And we landed a contract with CNN and a contract with BuzzFeed, that we were going to translate all their news from English to Spanish. By the way, how did you convince that? I mean, was it? Was it relatively easy? We were just cheaper. It didn't cost us any money to get the translations. Our users were doing it. We just said, oh, how much do you pay for, at the time?

It was like they paid like 10 cents a word. We said, oh, we'll do it for five. Five cents a word. And it worked. You know, the translations worked and everything. There was one problem. The amount of money that we were going to make was actually relatively small. And it's because translations are, it's a pretty shitty business translation.

It's five cents a word, but then pretty soon, Suddenly, you know, the people from BuzzFeed found somebody else that would do it for four cents a word. Sure. And then we're like, fine, we'll do it for three. It's a race to the bottom. It's a race to the bottom. And then computers are getting better at it. And by the way, you, I'm assuming you didn't want to found a translation service.

You were founding an education company. That's right. And one thing we realized this, as we were executing on these contracts, we realized that's where the money's coming from. So we're spending more and more of our effort making sure the translations are accurate as opposed to teaching well. And it was, there was no ads. It was just a website initially. But you had a lot of people who signed up, right? Like a couple of million.

Yeah, exactly. We were growing. It was good. But then a couple of things happened. The first thing that happened is we realized this was not going to be a great business. The other thing that happened is we hired one engineer and a summer intern. They started on the same day on a summer. And we told them, hey, the two of you, you don't know each other.

we've heard these apps are the thing now instead of websites or they're starting to become the thing. Can you two make a companion app to Duolingo? To Duolingo.com. Let's try that. That's going to be your kind of summer project. Let's make a companion app to duolingo.com. They came back at some point.

They said, you know what? It doesn't have to be a companion app. We can actually do all the functionality of Duolingo. We can fully learn a lot of aspects of the language from this app. So we developed it, we launched it, And we were at exactly the right time because there were other apps to learn a language in the app store, but they were pretty crappy and also cost money. And so very quickly it became the most downloaded education app of all apps. Wow. And you guys were there like right as apps were like totally starting to explode.

Yeah, that's right. And how were you, just curious, how were, I mean, you did not have an, I mean, you're an educator, you're a professor, but how were you building a curriculum? in different languages. Yeah, this is great. I mean, we didn't. We had no idea. I mean, we went and read some books about how to teach languages. And first of all, most everybody does not agree on what the best way is to teach a language.

And then so we thought, okay, this is pretty hard. Let's just take what we can from the books and put it out there. So I made the first Spanish course myself because I'm a native Spanish speaker. Severin made the first German course himself. And this is going to be just like kind of the gamification. model where it was just level up, level up, level up, level up. That's the other thing that we realized in the process. When we started making the first versions of the courses, we had people try it out,

and people would say, oh, my God, this is really boring. The hardest thing about learning something by yourself is staying motivated. So let's make it fun. So fortunately, early on, we made something that was both fun and also taught you a language. Got it. All right. So you launched this thing. is it pretty clear to you, you know, by the end of that year, by 2013,

I mean, I think you got 15 million users by the end of that year. Yep. Was it pretty clear to you that it was not going to make money? It was not going to be sustainable just by doing contracts with news organizations? It was. It was. And particularly because we couldn't figure out how to make the translation stuff work on our app. Just people couldn't type that much in the app, et cetera.

So we just could not figure that out. And the majority of our users were in the app. So I made the difficult decision to cancel those contracts and forget about trying to make money and just raise more venture capital. And sure enough, we had some very good investors who decided to invest, and that's how we were sustaining ourselves. So you're basically sustaining yourself off investment cash. But did you feel like you had to change your business model? I mean, this, by the way, has been one of these biggest struggles with Duolingo. Our mission from the beginning has been to give free language education.

This is why we started the company. That is the thing. There's an easy way to make money here, which is just charge for it. And so when we talk to investors, the first thing I would tell them is, look, the one thing that will not change is, I'm not going to charge for people to learn a language here. And so, you know, people would invest. And I think a lot of times they would invest and they would think, okay, eventually they're going to charge. Yeah. Right.

But I told them, you know, I'm not going to charge. And so we raised funding. I mean, we raised, at some point we had raised close to about $100 million. Wow. Wait, hold on, stop, pause. 100, I mean, you're not manufacturing a product in a factory. Why? I'm just thinking it's ones and zeros in a computer.

What? You got to tell me why it's so expensive. Here's the thing. You want to hire really good engineers and really good designers and really good product managers. They are not cheap. I got you. You got to pay them. and that's that's the expense right these are the most expensive suffer engineers are the most expensive part of

but you're based in pittsburg you're not based in california correct and you would think that because we're based in pittsburg cheaper you yeah it's not true there's a reason from the beginning we decided we're going to hire only the very best people and the very best people have offers from companies in california and they say i have this offer that is for you know i'm just graduated from college I have this offer that is for $130,000. Are you going to match it or not? Yeah. And so, what do you do?

But we had no idea how we were going to make money. I mean, the first way we tried, which was these translations, we had given up on. And at some point, we raised money from Google, capital G is one of their investment arms. We raised money from them. I believe it was 45 million bucks that were raised from them. Wow. You know, I was feeling good. I thought, okay, well, we can continue raising money for a while.

Like, whatever. I don't need to figure out how to make money. We can just continue raising money for a while. And a partner from that firm sat me down and she said, you know, you're not going to find a bigger fool. Yeah. We're the biggest fool. And nobody else will invest in this company if you don't figure out how to make money.

If you have zero plans to make money, nobody else is going to invest in this company. And that's when we started kind of freaking out and try to figure out how we're going to make money. And that was a difficult time for the company. Yeah. But, I mean, it doesn't seem like it would have been that complicated because you could just run ads, right? Yeah. Was a part of you, was it part of you like, oh, ads? I don't want ads.

Yes. We were very precious about this. We cared a lot about the user experience in Duolingo. And we thought, first of all, we're never going to put ads in here. Secondly, we're never going to charge people. And basically, Lela from Google, was like, well, I don't know how you're going to make money. because these are the standard weights of making money, but you've got to figure it out. This is an important point because I think a lot of businesses start out this way with really strong principles that seem right.

Yep. But when you say that, you actually kind of box yourself in because sometimes there are reasons to do some of those things in order to accomplish the bigger goal, for example, of delivering a free product. You are completely right. So you had to basically do a 180. You had to have hard conversations with people and say, hey, you know, we kind of start thinking about maybe advertising on our site. We did. At the time we had, I don't know how many employees, 60, but it's funny because the people we had hired early on, they had offers from, you know, companies that had much bigger names than us, Facebook, et cetera. And the reason they had come work for us is because they really believed in our mission of kind of giving free language education. So we had ended up with basically 60 to 70 zealots who all just wanted.

Wanted to give free education. Yeah. Yeah, the mission. Like, we're here for the mission. To this day, it is still like that. So, which is great. But we had a year of turmoil where basically we're like, okay, well, we need to figure out how we're going to make this sustainable. Because we're not going to be able to raise much more funding, you know, if we never make money.

And so sorry, but there's got to be a way to make money. And if you don't have money, you don't have a business. We don't have a business. And we're not going to be able to reach the number of people we want to reach. At some point, enough people got convinced. that okay, we'll put an ad at the end of every lesson. I remember we showed it to our head designer who he almost vomited. I get it.

I mean, did people say to you, or did you get the feeling that, like, I don't know, people thought that you were selling out? Yeah, I think so. And now, you know, I don't think it's too many people because I think I myself went through a transformation that took me, it took me several months to go through this transformation. And, you know, we put the ad at the end of every lesson. Fortunately, that actually gave us quite a bit of money. I mean, as soon as we put that at, our run rate went from zero to like 10 million bucks a year.

Wow. One tiny ad. It didn't even cover the whole screen. What was the company? Oh, we used Google Ads. Okay. They came from everywhere, random places. Not late night booty call.com, though, I hope.

No. And soon after that, we got a bunch of people who would tell us, hey, I don't like ads. Can I pay you to turn off the ads? And we thought, okay, how about we launched the subscription service where the main thing is you can pay us to turn off the ads. It turned out, and this is just we really just didn't realize this. Subscriptions made us a ton more money than the ads. And what would it cost to subscribe? $9.99 per month.

Okay, so a little bit more than Spotify. But through that, you could do any language you wanted, any course with no ads. Right. And if you didn't want to pay, you just had to watch the ads, you know, see the ads at the end of a lesson. And so all of a sudden you had people who were willing to pay some money. But still, I mean, we, of course, I know we know the advertising model and podcasting. And it's still a small percentage of people. When you've got a freemium version, right, even Spotify, most people are still getting it for free and dealing with the ads.

Yes. So, well, at first it was very few. By now, I mean, it's been a couple of years since we added our subscription. By now, 3% of our users pay for the subscription. Which is a lot, because how many totally users do you have? A lot. Tens of millions. Yeah.

I mean, active users, right now we have about 40 million monthly active users, and about 3% of them are paying subscribers. All right. Now, here comes the tough part, which is Duolingo, of course, is free. So, you know, it's a free offering. but of course even with free offerings, there's going to be criticism. And, you know, there have been professors who specialize in this have said, you know, you can't really get that proficient with delingo. You know, maybe I can learn some words or phrases, but it's not, doesn't really actually teach me a language.

And before we dive into this, just in general, when you hear this criticism, does it make you defensive or does it make you think like, maybe we've got to make this a little better. The truth is, there's a lot of marketing out there that tells you that you can learn a language in nine days. This is not true. Actually, learning a language takes years. This is the first thing to understand. Secondly, there's various levels, of course. Duolingo teaches you pretty well from zero.

So starting from the beginning to about intermediate, depending on the language that you're learning. Yeah. If you want to use dualingo and start from zero and get to the point where you're the poet laureate of that country, that is unlikely to happen with duolingo by itself. That's just not going to happen. But the way we see it is, A, we keep getting better and better. B, if you look at any single method, I mean, take the, for example, the single method of U.S. high school education. It kind of doesn't work very well. How proficient are people who take Spanish throughout high school at the end of high school?

on average, not that proficient. And our goal with dual linguine, we're not there yet, but our goal is to get people from zero to a level of the language. It's a level called B2, which is where you can get a knowledge job in that language. You know, Google, for example, employs software engineers here in the United States

whose English level is B2. Okay. That is our goal. We're not there yet. So we interviewed James Park, who created Fitbit. And Fitbit, the genius of Fitbit was it was like a gamified exercise, right? People are like, ah, I did 10,000 steps today or, you know, I made it. It's hard to get humans to change habits. So how do you get, let's say, a significant percentage of people to finish a language program?

Yeah, this is what we spend a lot of our time on. And, you know, this is one of the things that when we get criticisms, one of the things that they don't take into account is a lot of the research about how to teach a language is done with people who are, in some sense, held hostage in the classroom. Yeah. The hardest thing about learning a language is staying motivated. So we spend a lot of effort trying to make Duolingo more motivating and more fun. And the way we do that is just by adding game elements.

I mean, using Duolingo, a lot of people refer to it as playing Duolingo. It's just, it feels a lot like playing a game. You get rewards, you get like level up kind of thing. That's exactly right. And other things we do, for example, this is another interesting thing we do. We want you to use Duolingo every day. I mean, the truth is, if you actually want to learn a language, you've got to turn it into a habit. One thing that is really powerful for us is we send you notifications.

So we have a pretty sophisticated artificial intelligence system. It's one of the most sophisticated things we have that figures out when to send you a notification and what to say in them. And for example, one of the most powerful things we've done is after about five days, if you haven't come back, we send you a notification that says, these notifications don't seem to be working. We're going to stop sending them for now. that turns out to be an extremely powerful way to get people to come back. Huh. Because they feel guilty that we're stopping to send the notification, so that gets them to come back.

Yeah. It's like you've got to hold people accountable. Yes. Which is very hard. This is why people have personal trainers, because otherwise they won't go to the gym. Yeah. But why would somebody pick Duolingo over any other language app? You mentioned Rosetta Stone, for example, right?

And I'm sure you've looked at it. And I mean, obviously, you're a competitor, but they probably do a decent job teaching languages. So why wouldn't you just, like, do that? Yeah. I mean, before we launched, Versa Estan was the biggest thing in the world. And everybody told us, you're not going to be able to beat them. Yeah. By now, I would say we probably have 50 times more uses than they do.

Our revenue is probably 5x their revenue. Oh, my God. Something to that effect. There's a big reason. It's the fact that we're free. And this is one of the things that our mission, which is really giving free language education, I think it's actually good for business.

Yeah. We don't spend very much in marketing. We have all these people, 97% of our users don't pay us. Yeah. So it's just, I think we just have this huge user base of people who are basically our marketing engine. But a lot of companies in education eventually turn into marketing companies. Yeah.

For the following reason, you realize, man, making improvements in how well you teach is hard. And it really is very hard. It takes years. So for every dollar you have, you quickly realize it's actually better to spend your dollar marketing your product rather than making your product better. I've been a hardliner on this. We spend all our effort just making the product better.

And the main reason is I think we're really in it kind of for the long term. And I think in the long term, the better product will win. And this is why we don't spend so much money on marketing. Luis, in December of 2019, your company was valued at $1.5 billion. dollars. Now we are all in a very different world. A lot of the businesses that have been on how I built this are really struggling. Huge companies, airlines, you know, products, they're in trouble. Some of them will not make it through this. Are you guys seeing a negative impact? Are you seeing people or because people are stuck at home may be different? I mean, Netflix is doing great. What's going on

with you guys right? Because you're all working remotely, presumably. We're all working remotely, yes. I'm here just by myself. I came here to record this. Yep. And we are closer to the Netflix. We're seeing increased demand. Like significantly increased? Yeah, significantly. Our revenue has gone up about 50%. So we used to make about $400,000 a day. We now make about $600,000 a day. Wow. But the ad market is drying up. Yeah, but we make very little money from our ads. Most of our money is from subscriptions. So because people are sort of stuck at home, you're seeing people say, up for Duolingo? Yes, we are seeing that.

And the interesting thing is we see it. You know, we have users everywhere in the world. Only 20% of our users are in the U.S. We've seen it in every single country. And this is this amazing thing. The first country we saw it, obviously, is in China. As soon as there was, you know, social distancing measures in China, our traffic in China doubled.

And I kind of never thought it would happen anywhere else. But soon after that, Korea was like that, Japan was like that, then Italy went like that, the rest of Europe went like that, then the U.S. And so we're seeing massively increased demand because of this. So what's the exit? I mean, you got investors. Obviously, they convinced you were convinced that you had to keep it sustainable and to grow it. And you're now valued at $1.5 billion.

And I know you're not public, so you don't probably disclose whether you're profitable or not, but it sounds like you're, you know, keeping the lights on pretty well. what's the end game? At some point, you know, it's sell or go public. The end game here is to go public. That's what we've been working towards. It's going to be very difficult to say these days when we're going to go public because, you know, with this whole COVID thing, the markets are really crazy. But, you know, we have the revenue now to be able to go public. We are cash flow positive. And one of the things that I think is really important is the way I see it, so much more still to go in terms of how much better we can teach. And also teaching all this

stuff, we recently released an app to teach reading to young kids. That's called Duolingo ABC. I think there's a huge amount of impact that we can do with that. So, I mean, to me, it's just keep going with this. That's the goal. Luis, given all of the things that have happened to you, I mean, your incredible success, how much of that do you attribute to your intelligence and how hard you worked and how much do you think it's because you were lucky? I was definitely lucky. I mean, launching an app at the time we launched the app was exactly the right time. And launching an app in late 2012 was a really good time to launch an app. So, you know, I've been incredibly lucky. I think that and I think just hard work on the same thing. I get obsessed with the

same thing. I mean, I've been working on Duolingo for now, eight years and I'm pretty obsessive about it. So I think it's both. That's Luis von On, co-founder of both Duolingo and Recapship. By the way, Luis says that having a tech company based in Pittsburgh actually helps to attract top talent even from places like Silicon Valley.

For starters, the median price of a home in Pittsburgh today is about $180,000. In San Jose, it's almost $1.1 million. In fact, a while back, Duolingo put up a recruiting billboard in San Francisco that simply read, Own a home, work in tech, move to Pittsburgh. And thanks so much for listening to the show this week.

You can subscribe wherever you get your podcasts. You can also write to us at hibt at npr.org. And if you want to send a tweet, it's at How I Built This or at Guy Raz. Our show is produced this week by Casey Herman with music composed by Rumtine Arablewe. Thanks also to Julia Carney, Candice Lim, Neva Grant, and Jeff Rogers. I'm Guy Raz, and you've been listening to How I Built This.