I'm Jessica Harris. This is from scratch. My guest is Louise von On, an inventor and computer science professor at Carnegie Mellon. Louise is one of the creators of Capture, a device that allows websites to determine whether you're human or a computer. Trying to abuse your system captures our wavy, distorted words on screens that people need to type out when you're submitting a form toe a website to confirm that you're not a bott. His company, Recapture proved users were humans while digitizing books. The company was sold to Google in 2009. In 2011 Louise founded Duo Lingo the most popular language learning programme on the Web, attracting over 150 million users in its 1st 4 years. Users worldwide learn a language for free while also translating the Web into every major language. Investors include Google, any a Kleiner Perkins, Ashton Kutcher and Tim Ferriss. Louise graduated from Duke in 2000 and got a PhD from Carnegie Mellon's in 2005. He won a MacArthur Genius grant in 2006 and is from Guatemala. Welcome.

Thank you for having me

and all this, by the way, since 1978 the year the year you were born. So I want to talk about capture first and about your research. Because your research at Duke and Carnegie Mellon has informed your career in your companies. Can you tell us what it means again?

Yeah, so captures is those distorted, squiggly characters. They have to type all over the Internet whenever you're buying tickets on Ticketmaster, whenever you're getting, Ah, a new account. For example, an email account from from Gmail. Or whenever you're opening an account on Facebook, you usually have to take these distorted characters and and capture stands for completely automated public Turing test to tell computers and humans apart. Basically, what it is is it's a test to determine whether you're actually a human. So, for example, in the case of Ticketmaster, those capitals are there. The distorted characters are there to make sure that somebody didn't write a program to buy kind of all the tickets for concert. I am kind of two at a time.

You develop the term capture with a professor of yours, manual Bluhm, who was a professor at Carnegie Mellon. How do you come onto this concept of security versus any other form

in about the year 2000 I was. I was listening to a talk at Carnegie Mellon from a guy named Rudy Member who at the time was the chief scientist of Yahoo. He had a talk about 10 problems that they couldn't solve at Yahoo. One of them was they had a problem with people who were writing programs to obtain millions of email accounts from Yahoo. The people who were doing this for spammers. And so they were basically supporting spammers on they didn't want to do that. Over the next few months with my PhD advisor Manuel, we came up with this idea off off, coming up with a test. You know, at the time nobody had thought, Well, we should test whether they're human or not. Just the problem was, how do we stop these people from getting free email accounts on? We came up with the idea that, well, one way to stop the Mr make sure that it's a human actually getting them because humans can't get millions of accounts. The humans can only get a couple 100 accounts because afterwards they get bored. So that that was the idea.

I've always liked the name capture because it's sort of like Caccia. You look through

the idea. It's like your capture or gotcha or something.

Isn't it Interesting path dependency like you happen to be sitting at this lecture with this Yahoo guy and had you not encountered him or or had your relationship with manual this might never have been. And it's always hard to think of Counterfactual is, and you might have gone on to create something else that's innovative, but funny how things happen through happen, stance like this.

Yeah, there are all kinds of things. I mean, I also I was torn about whether I should go get my PhD, a Carnegie Mellon or whether I should go. I was considering at the time going to Berkeley or Stanford, and it wasn't clear. So, yeah, there were all kinds of happenstance

there. Why did you choose Carnegie Mellon over Stanford

or Berkeley because off because I wanted to work with with my advisor Manuel. I mean, he was This guy wanted turning a word, which is kind of the equivalent of the Nobel price for computer science. And he was, you know, we hit it off really well as soon as I met him. So I chose Carnegie Mellon over all these schools because I could

work with him. And by the way, we mentioned Turing. You tell me exactly.

Also turning Island Turing is widely considered the guy who you know, the father of computer science. There's a movie about him, the imitation game. I mean, he was in the, you know, it was a mathematician in the middle of the century, British mathematician in the 19 forties, 1919 fifties. The British government contacted him to break the enigma machine, but in, you know, among other things, that he did, he essentially started the field of computer science and he had all kinds of ideas. And one of his ideas was,

you know, he thought, Well, at some point, computers are gonna be probably about as human as intelligent. Assume it's on. One of the questions that he addressed was, when will we know that a computer is as intelligent as a human? And he came up with this idea off what's called the Turing Test? Imagine that there's a human judge that is talking over texting with a human and a computer, and if the human judge cannot tell which one's which. Then we will say that that computer has passed the Turing test everybody in the field, computer science and us about the turning tests. I mean, this is very similar to our Turing test, where you're trying to distinguish if you're talking to a human or computer.

But the big difference is the judge in this case had to be a computer, not a human. So it had to be. So is this paradoxical thing that computer needs to be able to determine whether it's talking to him in our computer? But a computer should be able to pass this test

now in the same research as you were trying to develop this idea of capture, you also came up with the idea of gua piss, which are games with a purpose. Now can you describe that? And how are they related?

Yeah, so? So I helped develop this notion of a capture on Ben. It started being used by all kinds of websites pretty quickly. Yahoo started using it. They had an actual big problem. Eso pretty quickly they started using it on all kinds off website, started using it and then that became my research, where coming up with things that computers could not do but humans could do in particular. I had just gone to a talk about somebody who was at the time working on a computer program to try to solve crossword puzzles. I thought, Okay, computers can't do that. Then I got on a plane and I noticed that everybody in my row was trying to solve a crossword puzzle. And I thought, Well, these people are doing something that computers cannot do and they're doing it willingly.

So can we get people to do the things that computers can't willingly on? That's where the idea of games of the purpose came. Where the idea is, these are games that people playing like crossword puzzles. But as they're playing them, they're trying to solve a problem that computers cannot solve. And, you know, this was before the word crowdsourcing even existed, it said. But this is like the first instance she ation off idea of kind of crowd sourcing. Instead of getting computers to solve problems, can we get humans over the Internet to just solve them for us

now? This is all around 5 4006 and 2006. You get the MacArthur Genius Award. How does that happen? It was that out of the blue. Or did you know you were nominated? What is that

process had no idea I was nominated. I suppose somebody dominates you, but you're never aware of it. And in the meantime, the MacArthur Foundation is doing a whole lot of research on you And what, you know, whether you should get one of these or not. And at some point, you just get a phone call saying, Hey, I just got the MacArthur word, but they're a little sneaky. The before that they do. They do contact you in ways that you don't realize it in. For example, in my case, there was there was somebody who was supposed to be a PhD student in psychology who was very interested in my research.

Andi. Andi. Actually, that that was that was them on. One of the things they needed to make sure of is that when I got the phone call, I was there in in my office, and this person I said, I'm gonna you know, I'm gonna come. Teoh, I'm gonna come on that day. And so I was in my office waiting for this person, and I just got a phone call. Eso So they're they're a bit sneaky like

that. Who nominated you? Ultimately,

You know, you never find out

you were less than 30 years old. You were, what, 28 years old when you won the award. So in this, you know, five year period, you get the MacArthur Genius Award. You develop capture, you're working with companies like Yahoo. How did you go about commercializing capture with Yahoo,

for example? So that the first instance creation of that was not commercialized at all? We just give it to you. Hope for free. We're super happy for them to use it. We were academics, and we were just happy that that somebody was using what we were doing. Um, later, I did start a company related to capture us, but it was this kind of that was in 2006 where I started this other company called Recapture, and that became a commercial venture. But that was that was

much later. And prior to recapture, your first company was called PSP, which was matching images where to users from in two different geographic locations air using their computers to identify images. And they're playing a game. They have to match their description of the image. The founders of Google, Sergey Brin and Larry Page got wind of this. You gave a speech at Google where they met you, where they founders met you.

Yeah. So that's what happened. I mean, I mean, a few people from from Google heard about my work. Um and so I went to Google. This this is 3 4000 for us. A much smaller company. Eso I gave a speech and the two founders were actually in the audience and then right after they just, you know, very simply just said, Hey, uh, where we're gonna make use of this. Look, can we buy it?

And And it became part of their image label.

Er yes, it be became their Google image label, which was the thing that was used to to improve search

engine E. S B was a one man band that that you sold to Google and then your next company was recapture that you start in 2006 which basically was proving that you're human again. But while digitizing books, how did you come up with

that idea? So this was similar to the idea off of the PSP game or the Googling much leveller where? I mean, we're trying to get people to solve problems for us that computers cannot. And in this case, the idea was the following bite by 2006 captures where he was all over the Internet, every single kind of major website use captures. I did a little back of the envelope calculation about how many captures were tight, but before on the world. And the number that I came up with was about 200 million times a day, 200 million times a day. Somebody was typing and capture first. I was very proud of myself. I thought, look at the impact that I had. But then I started feeling bad because not only everybody kind of hated those things. I mean,

the capitals that kind of universally hated. But also each time you type one of these e waste about 10 seconds of your time on if you multiply 10 seconds by 200 million, you get the humanity was wasting about 500,000 hours every day typing the so I started thinking, OK, can we get them to solve Ah on important problem? Because during those 10 seconds, they're doing something that by definition, computers cannot do. So can we get them to do something something useful? And I came up with the idea of helping to digitize books. The ideas here you want to put the book on the Internet. The first EPC scan every page of the book scanning literally what it is, is taking a digital photograph of every page. The next step in the process is that the computer needs to look at these pictures of pages and extract all the words. So the site for all of the words in there the problem is that for older books, the computer cannot recognize many of the words for the same reason that computers cannot read.

These distorted the captures A. For older books. Sometimes the ink is a little faded, and the pages have turned yellow and or the picture maybe a little blurry and you just cannot read. And at the time, about 30% of the words could not be read by by the computer, so I thought, Well, I look at one of these pictures and I can read 100% of the words. So can we get people to do it? So that was the idea of the idea was to take all of the words that the computer could not recognize in a book digitization process and then extract them as a picture and then send them over the Internet to somebody who's typing a capture. And whenever they type it, we're gonna use what they typed to help digitize the book on. That was the idea of this company Onda started in 2006.

So incorporating the games with a purpose into this incorporating crowdsourcing into this all kind of, you know, jelling into this one stew of a company recapture when you started the company, how did you monetize? How did you make money from this and are with digitizing

the books? So the idea is we provided a free capture service, the websites. But the way we would make money is we would charge for book digitization. For example, we started helping to digitize all the New York Times archives where they had scanned them on, and they had this huge problem that yet with scan. But computers couldn't recognize about 30% of the words.

Did you have venture

capital? No, no, we didn't need it. I mean, we had. I mean, on the day we started the company we were making we're making a few $1,000,000 a year already, and we didn't need to spend that much.

Google acquired Recapture in 2009 you and 2011 started Duo lingo, which is the most popular language app on the Web. And basically, you want to provide free language learning to anyone in the world. How did you come up with that idea?

Well, it started. I mean, so I in 2009 Google bought Recapture on Guy was I went to Google for a little while after they acquired it on and I I started really thinking, What do I want to do after this? And it was an interesting time because I had more than you know, more than enough solved my financial problems forever on for me and for my kids, and it was fine, so I kind of didn't I would have to worry about money anymore.

Do we know how much they bought it for

is that it's private, but, uh but I was, you know, I was in a situation where I was very fortunate situation. Um, but I started thinking, OK, what is it that I want to do for the next? You know, 10 years ever since age 10 I wanted to be a professor, So education was always my passion. My my views on education were always related to the fact that I'm from Guatemala. It's ah which is a very poor country, and a lot of people talk about education or something that brings equality to different social classes. But what I saw in Guatemala and what you see in most developing countries is that it's the opposite. Usually,

people who have a lot of money can buy themselves the best education in the world. And because they're so well educated, they keep on having a lot of money. Whereas people who I don't have very much money in, especially in developing countries, barely learn how to read and write. So I wanted to do something related to education, but that would give equal access to everybody. So I decided that to work on one aspect of education, which was, which is very large, happens to be very large in most every country except the U. S. Which is learning foreign language. There's 1.2 billion people in the world learning a foreign language. The majority of these people,

100 million of them satisfied three properties. First, they're learning English. Second, the reason they're learning English is to get a job on third. They're of low socioeconomic conditions. So most people learning a foreign language are basically trying to learn English in order to get out of poverty. That's that's it. But the ironic thing is, at the time, most of the ways that work to learn a foreign language were really expensive, particularly through software. So at the time, and it was kind of a height over a set of stone, I was about between $500,000.

So if you're poor, it's hard to get the $1000 to pay

this part right. I mean, you're trying to get learn English to get out of poverty, but it seems you need 1000 bucks, which makes no sense. Uh, so we decided to launch something to teach languages But the whole goal was no matter what, this is always gonna be free. Now the question is, how are we gonna how we're gonna keep it free? And at the time I thought, you know, this is one of those If you're a hammer, everything looks like a nail. I thought, let's do the same thing with it with recapture. Where with Recapture.

What we were doing is we're providing this free capture service. But in the background, we were making money from people, you know, helping us digitize books. So can we do the same thing with languages? Can we provide a free language learning website? But in the background, we would be making money with people doing some something useful for us. I started this with my PhD student at the time a guy named Severin Hacker, which is funny. His last name is Hacker. We thought, OK, the thing that people can do while they're learning a language is to help us translate stuff. Can we ask? People are learning along which get them to translate useful stuff. And that was the idea with dueling Go on, get worked.

So again, you used the capture idea and the Guapo. The games with a purpose and crowdsourcing idea to come up with the monetization tool for dueling Go. Having companies pay you to translate content

on their websites? That's exactly right on dso west. Pretty soon after we launched, we CNN became our client. So they what they were, what they were doing is they would write older news in English, CNN, and they would send us the news in English. And then what we did is we We had this language learning website dueling go which we had just launched, and we would give this news to some of our users. And then it was users that we're learning English. So, for example, Spanish speaking users learning English, they would get it and they would help translate the news into their native language into Spanish. And they were doing it, of course,

to help learn English. And then we would send that news story back to CNN. But translated, we've actually moved away from this. This is not what we're doing any more. But this was what we launched dueling go with.

What do you doing? Since then

we realized that over time we were spending more and more time on the translation quality than actually teaching, because that's where the money was coming from. At some 0.7 and I were having dinner one time and we thought, Well, we've turned into a translations company on Do we really want to be a translations company? And first of all, this was not what we set out to do. We didn't want to be a translations company. And secondly, translations business is not a good business to be in its, ah, race to the bottom. This was a tough decision because it was a business model that seemed to be working. But we decided we're gonna stop this. And so we actually cut the contract with CNN and Buzzfeed. We moved away from that.

You did that because you want to primarily to be a language learning company, nor the translation company would basically becomes a commodity. Ah, and it's it's Those decisions are the hardest. When things they're still going Well, Did you have any pressure from your venture capital partners to do that?

Yes and no. I wouldn't say strong pressure. I mean, the language learning market is a much larger market than the translations market and a growing one as opposed to a you know, you have this other travelers translation. You also have this other problem that there's this looming thing that at some point Google translates going to get really amazing.

You had a pivot moment at this dinner with Mr Hacker with Severin. What have you decided to become?

We decided to become ah, language learning company. We saw. We thought, Oh, how are we gonna make money? Particularly given the fact that one of our tenants is we can't charge. We have three business models and you know, it is one of those things. When somebody says they have three business models, that means they don't have a business model. But we have three that are Each of them is working. Ah. Okay, so the 1st 1 and the one that we've been spending most of our time on, actually a couple of them came from from our own users. We started getting a very common email.

Very similar email from a lot of our users on it was the following. It would say, Thank you so much for teaching me English. I was not able to afford learning English before this. But now I have a problem. And it is that I need a certificate that shows that I am no English. People were asking, you know, can you give me a certificate? We started looking into this whole certification business on what we found was pretty insane. So first of all, $10 billion a year spent by people certifying that they know English. That's a huge amount on there. So kinds of reasons. People who applied to come to college in the US, the way they certified as they have to take a standardized test called the Tofel.

But there's other reasons. If you want to get a work visa in the U. K. Or Canada or Australia anywhere in the Commonwealth, you have to take a standardized test approves that. You know English. If you work for many multinational corporations in non English speaking countries, you have to take a stand. It s s so it costs about $250. You usually have to take them in. Ah, physical testing center. We have to make an appointment a few weeks in advance, about four weeks in advance. Then you take the test, then You have to wait another four weeks for your results.

So the whole process takes about eight weeks. Cost about $250 you have to go somewhere. That's under pretty annoying to us. But then it's even worse because most people who are doing this I'm developing countries. It's way worse. 250 bucks, That's Ah, that's a month salary.

So what is your solution?

We came up with a test that you can take from a nap. You don't have to go to a physical testing center. And also, our test is way cheaper. Supposed to under $50. We charge $50.

You've raised venture capital for dueling with the first time you have done so, Do you have any interesting stories on that process?

The first time around, it was not as difficult as in the first. The first round of funding the first and second round of fundings. We were raised them essentially on the fact that I had done something before way had an idea what a whatever,

Um, and that was Google and Union Square.

Union Square Ventures was the 1st 1 and they weigh, loved working with them I mean, they're they're here in New York. The one of the interesting things that were not in Silicon Valley. So Duelling was actually located in Pittsburgh because of Carnegie Mellon University. One of the interesting things is when we raised funds, we found that the Silicon Valley investors will want to invest, but they all in their term sheets. It said, you gotta move to Silicon Valley. They all said that. Whereas we found that the New York investors didn't say that, they said, Oh, yeah, well, traveled to you.

Um, so we we went with a New York investor of the first round. After a while, after doing that have been going, you know, you can get money probably one or two rounds based on your previous stuff, but it once the once you've gotten a couple of rounds, you have to get money based on your success. Eso in the you know, when we got, for example, when we raise money from Kleiner Perkins that was based on dual Ingle success in traction. And you know, at that time we started being able to get money from West Coast investors without making us move because they're like All right, fine. You're You seem to be doing okay in Pittsburgh

a year from Guatemala. Your mother and father are doctors. What kind of doctors air They,

Um So my mother's a pediatrician. My father is Ah, they do basically bone

type stuff. Your last name is Von on your your originally German, Uh, your When did your family come from

Germany? My father's parents came in the early early 20th century.

And what was the impetus for their coming?

There were there was quite a number of people from Germany and from Europe coming to Latin America because of essentially for business opportunity. I mean, in Guatemalan, particular what was going on is the educational levels were pretty low aware us. If you came with the German education, you probably could do pretty well. So, you know, they started a farm and they started. They knew about technology. Enough to essentially succeed with little Dimon. No competition.

In addition to your parents being practicing doctors, they owned a candy factory. Tell me about that.

Yeah. So my mom is my mom's side of the family. So my my grand parents from my mother's side of the family decided to start a candy fact. They are both from Spain. It happened to grow quite a bit My grandfather died at at an early age. So it was then my grandmother who ran this candy factory and became the most popular candy factory in Guatemala. They made, like, marshmallows and chocolate and all kinds of Candies. Um, at some point, they they had, by the way, 12 Children. Yeah, and my mother was one of them. And at some point,

they passed on the baton to their Children. And so my mother was one of the basically 12 owners of this, and yeah, that was me growing up. I eat Essentially. Didn't like the candy come. Well, I got bored of it. After a while. I ate so much that I got bored of it. But I loved the machines. So in Sunday's, I would go in there and essentially break machines. I was not loved by the U of engineers.

You'd break them because you wanted to see what they were like inside. Or you are you accidentally

broke? No, no, no. I want I took him apart. And I would do stuff about taking on a partner of course I would take them apart. And then when I try to put them back together, I will be left with two or three pieces that I didn't know where where they went on. Uh, that was the problem.

What was the name of the candy company?

I was called Tropical Candy. It's

still there? Yeah. So, Marshmallow chocolate.

What else is going? Me? Bears. You know, almost any public

candy from an early age. You seem to have this curiosity about how machines worked. Yet you're focuses on computer science. Did you think you might want to be more of a physical like engineer? Uh,

Fiona Early on, I wanted to be. I mean, I really wanted to understand how the machines worked, and I sat there at home making little, you know, gizmos. That probably didn't do much, But I was very excited by that. It all changed when I was eight. I want a Nintendo on. My mother bought me a computer instead of a Nintendo. I was I was pretty pissed off. All my friends had a Nintendo, and I didn't get me a Nintendo. This was, you know,

in the eighties computers were not super easy to use. Um, she also didn't happen to buy me a super easy to his computer on dso. What I wanted to do was play games. So what I did is I had to figure out how toe how to use the computer to play games. I started essentially pirating games from from other people, so I became quite a game pirate in Guatemala. That's how that's how I got good with computers.

Eso This is, at the age of you know, 89 10 11.

Yeah, the parting of games

was more like 12. Yeah. Now, at the age of 12 also, you had this idea Teoh make gyms free or start this business for free gyms. Tell me about

that. Well, at the time, I thought I was the first person to ever come up with this idea. I was not. A lot of people have come up

with this idea. Let me guess. Harnessing the electricity that you produced by exercising

thousands of not millions of people have had this idea right? It is. But at the time, you know, it is like a 10 year old kid. I was like, I have this amazing idea I will become a gazillionaire from this idea. I also thought it really worked. You know, this is by age 13 or 14 I would think I thought the way we're gonna make money is by selling the electricity to the electric company. Unfortunately, you know, when I got older, you start doing being able to calculate how much electricity you can sell, it turns out, is not a lot. It's on Ben. You start also realizing about the gyms that most of the money that Jim's make are actually by people who sign up but never show up. So this is not

a good idea. Eso You seem to be kind of a self starting, industrious little kid. Did you have that view of yourself? And was that kind of the perception among your family members? Or how would they describe you? Yeah,

they would describe me. Probably a zoo, nerdy and with a little too much energy. Thats probably I mean, I was always doing basically getting in trouble, not not getting in trouble, because, you know, I was going out to later anything but getting in trouble by breaking things. I mean, one time I was trying to. I wanted to make a helicopter with an electric motor. I thought the way to make it look up there is basically right having a fan pointing it down. And if you if it spins fast enough, it's a helicopter. Turns out that is not is not the same kind of late. And so I had I bought this huge battery for it wasn't doing it.

Then at some point I realized that if I connected it to the electricity, the this would help. This would help make it go faster, like connected it to electricity. And then I made you know, not only my house, but a couple of houses around in the lost power because of my brilliant idea. So I was always getting into trouble for doing

crap like, yeah, and when you say you had a lot of energy, it was like mind energy, not so much physical

physical energy to basically had too much energy. I'm pretty certain that had I been born today here in the U. S, I would probably be on some sort of a d h D drugs pretty certain of this. But back then, in Guatemala, there was no such thing. It's just he had a lot of energy, that's all what people said,

and you do exercise daily. Uh, you exercise on an elliptical machine hard core for 15 minutes. Every

day is my workout.

And how long has that been your regiment?

Your go to maybe about five or six years, and it's really just I'm always trying to break my record. And as fast as I didn't go

and Tim Ferriss is one of your investors, and he's all about efficiency and work and work outs. Did you have this idea from him where you were doing this even prior to your your your

relationship? I was doing this prior to to meeting him, but he has all kinds off. He has all kinds of ideas. He's really a pro with this.

What other things do you do that I might not know off know about what other quirky parts of Louise

I don't read books on, which is a rare thing for Ah, Professor. The last book I read was in 1998 main leads because I'm very slow at reading.

Do you go back to Guatemala?

I do about twice

a year and How does it feel being back?

Well, one of the things I've become kind of a celebrity there because there's not very many people who are in the tech world from Guatemala, So I've kind of become, Ah, a bit of a celebrity there. People recognize me on the street. I mean, not not all of the people, but it happens. Almost every time I go out, I go and I basically don't leave my house. I go and I visit my mom

because it makes you uncomfortable.

Yeah, that and it's Guatemala happens to be a dangerous country. You start being worried about being kidnapped or something like that. And I I'm also very happy with the fact that my mom lives in a gated community and barely leaves the house. And also, she's nobody. Nobody knows what she looks like or anything, so that I'm happy about that.

Are your parents divorced?

No. My dad passed away when I was 13 or 14.

Did that change your relationship with your mother?

Yeah. I mean, I mean, I was always very close with my mother. I mean, I So, uh, yeah, she's the one who essentially raised me

and Are you an only

child? I am an only

child. Isn't it nice that people appreciate you as a celebrity for doing something of the mind? Like for being a computer science geek? That's so nice that you're, you know, it's not like Matt Damon walking on the

street right now. I feel good. And I think, you know, in particular in Guatemala, Guatemala. So you know, it's a country with a lot of bad news. I mean, this is bad stuff happens in Guatemala all the time, so they're very eager to get good news. So when you when they see somebody, I mean, for example, we have 11 single Olympic medal off all time. And the guy who want it is a super celebrity, because it's like,

this is the guy who won the Olympic medal on. So anything that is that, you know, a Guatemalan that is successful. Ah, thus you know, people are very eager

for that. An Olympic medal and now a MacArthur genius award. Thanks very much for joining

us. Thank you for having

my guest has been Luis von Ahn. I'm Jessica Harris. This is from scratch.