×

Sebastian Thrun, the CEO of Udacity, finds that long bike rides help him think through the problems of being a first-time entrepreneur.

Udacity's Sebastian Thrun, Godfather Of Free Online Education, Changes Course

He captivated the world with visions of self-driving cars and Google Glass and has signed up 1.6 million students for online classes. So why is he pivoting away from MOOCs? "We don't educate people as others wished, or as I wished," Thrun says.

There's a story going around college campuses--whispered about over coffee in faculty lounges, held up with great fanfare in business-school sections, and debated nervously by chain-smoking teaching assistants.

It begins with a celebrated Stanford University academic who decides that he isn't doing enough to educate his students. The Professor is a star, regularly packing 200 students into lecture halls, and yet he begins to feel empty. What are 200 students in an age when billions of people around the world are connected to the Internet?

So one day in 2011, he sits down in his living room with an inexpensive digital camera and starts teaching, using a stack of napkins instead of a chalkboard. "Welcome to the first unit of Online Introduction to Artificial Intelligence," he begins, his face poorly lit and slightly out of focus. "I'll be teaching you the very basics today." Over the next three months, the Professor offers the same lectures, homework assignments, and exams to the masses as he does to the Stanford students who are paying $52,000 a year for the privilege. A computer handles the grading, and students are steered to web discussion forums if they need extra help.

Some 160,000 people sign up: young men dodging mortar attacks in Afghanistan, single mothers struggling to support their children in the United States, students in more than 190 countries. The youngest kid in the class is 10; the oldest is 70. Most struggle with the material, but a good number thrive. When the Professor ranks the scores from the final exam, he sees something shocking: None of the top 400 students goes to Stanford. They all took the class on the Internet. The experiment starts to look like something more.

Higher education is an enormous business in the United States--we spend approximately $400 billion annually on universities, a figure greater than the revenues of Amazon, Apple, Facebook, Google, Microsoft, and Twitter combined--and the Professor has no trouble rounding up a group of Silicon Valley's most prestigious investors to support his new project. The Professor's peers follow suit: Two fellow Stanford faculty members launch a competing service the following spring, with tens of millions of dollars from an equally impressive group of backers, and Harvard and MIT team up to offer their own platform for online courses. By early 2013, nearly every major institution of higher learning--from the University of Colorado to the University of Copenhagen, Wesleyan to West Virginia University--will be offering a course through one of these platforms.

Suddenly, something that had been unthinkable--that the Internet might put a free, Ivy League–caliber education within reach of the world's poor--seems tantalizingly close. "Imagine," an investor in the Professor's company says, "you can hand a kid in Africa a tablet and give him Harvard on a piece of glass!" The wonky term for the Professor's work, massive open online course, goes into such wide use that a New York Times headline declares 2012 the "Year of the MOOC." "Nothing has more potential to lift more people out of poverty," its star columnist Thomas Friedman enthuses, terming the new category "a budding revolution in global online higher education."

It is a good story, as well manicured as a college quad during homecoming weekend. But there's a problem: The man who started this revolution no longer believes the hype.

"I'd aspired to give people a profound education--to teach them something substantial," Professor Sebastian Thrun tells me when I visit his company, Udacity, in its Mountain View, California, headquarters this past October. "But the data was at odds with this idea."

As Thrun was being praised by Friedman, and pretty much everyone else, for having attracted a stunning number of students--1.6 million to date--he was obsessing over a data point that was rarely mentioned in the breathless accounts about the power of new forms of free online education: the shockingly low number of students who actually finish the classes, which is fewer than 10%. Not all of those people received a passing grade, either, meaning that for every 100 pupils who enrolled in a free course, something like five actually learned the topic. If this was an education revolution, it was a disturbingly uneven one.

"We were on the front pages of newspapers and magazines, and at the same time, I was realizing, we don't educate people as others wished, or as I wished. We have a lousy product," Thrun tells me. "It was a painful moment." Turns out he doesn't even like the term MOOC.

When Thrun says this, I nearly fall out of my chair. He is arguably the most famous scientist in the world--and perhaps only Elon Musk bests him in successfully persuading regular people to embrace wild ideas. Thrun has been a public figure since 2005, when a modified Volkswagen Touareg of his design won a Department of Defense–sponsored competition that pitted cars without drivers through a 128-mile, pedestrian-free course in the Mojave Desert. That such a competition almost seems ho-hum eight years later is itself a tribute to Thrun's genius. He joined Google in 2007, where he led the program to develop its self-driving car, and then founded Google X, the ultra-secretive research lab behind Google Glass and other research projects so far-out that Google calls them "moon shots."

But building a company is different from building a research lab. It requires compromises, humility, and, crucially, taking in more money than you spend. And it's why Thrun might be giving up the moon--free education for all! Harvard on a piece of glass!--in favor of something far more pedestrian. It will be, Thrun admits, "the biggest shift in the history of the company," a pivot that involves charging money for classes and abandoning academic disciplines in favor of more vocational-focused learning. In short, Thrun must prove that Udacity is something more than a good story.


Sebastian Thrun is in a hurry.

"Let's just get dressed here," he says, leading me into an empty suite two floors below the Udacity offices. He tosses a pair of bike cleats and a Lycra cycling kit onto the ground, kicks off his sneakers, starts taking off his pants, and then motions for me to do the same. "I don't mind," he says. There's no locker room at the Udacity office, so he's led me downstairs, into a part of the building that is still under construction--never mind the floor-to-ceiling windows. After a few awkward seconds, I move into an adjacent room, throw on my gear, and follow Thrun east toward the Los Altos Hills.

Thrun, who is 46 years old and originally from Germany, is a committed athlete who possesses that outdoorsy vigor (and lack of physical modesty) often found in middle-aged European men. He has run half a dozen marathons; he snowboards; he kite-surfs; and he is an avid road cyclist. "I haven't been biking as much as I'd normally like to," Thrun confesses before we set out, explaining that he's done "only two" centuries, or 100-mile bike rides, this year.

I'd been warned that keeping up with Thrun tends to be a challenge in any setting, but I hadn't entirely appreciated it until Thrun clipped into his custom-made road bike and scooted up Arastradero Road, leaving me panting a few lengths behind. "Sebastian is like the smartest guy you've ever met, but on speed," says the entrepreneur Steve Blank, a friend of Thrun's and a Udacity investor. "And he hates to lose."

When I catch up to him, trying not to seem out of breath, he acknowledges that he normally doesn't ride with anyone, for this very reason. "I feel like everyone has this competitive instinct," he says. "And I want to be able to go at my own pace. I have trouble with all of these little decisions of running a company. Being alone--that helps."

The youngest of three children in a lower-middle-class family in Hildesheim, a town of 100,000 just outside Hannover, Thrun was a geeky kid, spending much of his free time in libraries or in front of a NorthStar Horizon home computer, on which he tried to write software programs to solve puzzles and play solitaire. As a lonely undergraduate at an obscure provincial college, Thrun thrust himself into trying to understand people better, dabbling in psychology, economics, and medicine. Eventually, he found his way to what was at the time a relatively obscure field: artificial intelligence, or the study of making machines that make their own decisions. "Nobody phrases it this way, but I think that artificial intelligence is almost a humanities discipline," Thrun says. "It's really an attempt to understand human intelligence and human cognition."

Thrun seems to owe much of his academic success to this early insight. As his peers wrestled with theoretical quandaries and high mathematics, Thrun's work had a romantic, populist flair. He designed and built robots around human problems, and gave them accessible names. Rhino, part of his thesis project at the University of Bonn, gave guided tours of the local museum. During a stint at Carnegie Mellon University, Thrun developed Pearl, a Jetsons-like "nursebot" with a human-looking face, to assist in elder-care facilities. His greatest achievement, though, was Stanley, the autonomous car that won Stanford a $2 million Defense Department prize and won Thrun the notice of Google cofounder Larry Page.

Thrun and his team originally planned to spin their research out into their own company that would create detailed images of the world's roads, using car-mounted cameras like the ones used to steer Stanley. Page offered to hire them instead. The collaboration helped lay the groundwork for Google Street View, and eventually for the fleet of self-driving Google-branded Priuses that these days navigate rush-hour traffic on Bay Area freeways without incident. Page and cofounder Sergey Brin went on to ask Thrun to launch Google X.

His trip in March of 2011 to the TED Conference in Long Beach, California, where he delivered a talk about his work, led to an unexpected change in his plans. Thrun movingly recounted how a high school friend had been killed in a car accident, the result of the kind of human error that self-driving cars would eliminate. Although he was well received, Thrun was upstaged by a young former hedge-fund analyst named Sal Khan, who spoke of using cheaply produced, wildly popular web videos to tutor millions of high school students on the Internet. Thrun's competitive streak kicked in. "I was a fully tenured Stanford professor . . . and here's this guy who teaches millions," he would later recount. "It was embarrassing." Though Thrun insists the timing was coincidental, just a few weeks later, he informed Stanford that he would be giving up tenure and joining Google full time as a VP. (He did continue teaching and is still a faculty member.)

Initially, Udacity was just another modest research project on Thrun's docket; he didn't even bother warning the higher-ups in the computer science department until after he had announced that first AI class. After two weeks, more than 56,000 students had signed up. "The conversation took a radically different turn," says Blank of his friend's interaction with Stanford after the response far outpaced anyone's expectations. The university was initially cool to the idea but ultimately embraced it, allowing two other computer science courses to be offered in the same manner. (Blank's popular entrepreneurship class at Stanford would eventually be offered on Udacity as well.) Thrun contributed $300,000 of his own money in seed funding, installed one of his old Stanford graduate students, David Stavens, as CEO of the new company, and set about recording crude course videos about Markov models and the like.

"It was this catalytic moment," Thrun says. "I was educating more AI students than there were AI students in all the rest of the world combined." By the end of the semester, he'd raised another $5 million and was standing in front of the Digital Life Design conference in Munich, promising a world in which education was nearly free, available to poor people in the developing world, and better than anything that had come before it. "I can't teach at Stanford again," he said definitively. "I feel like there's a red pill and a blue pill. And you can take the blue pill and go back to your classroom and lecture your students. But I've taken the red pill. I've seen Wonderland."


It's hard to imagine a story that more thoroughly flatters the current sensibilities of Silicon Valley than the one into which Thrun stumbled. Not only is reinventing the university a worthy goal--tuition prices at both public and private colleges have soared in recent years, and the debt burden borne by American students is more than $1 trillion--but it's hard to imagine an industry more ripe for disruption than one in which the professionals literally still don medieval robes. "Education hasn't changed for 1,000 years," says Peter Levine, a partner with Andreessen Horowitz and a Udacity board member, summing up the Valley's conventional wisdom on the topic. "Udacity just seemed like a fundamentally new way to change how communities of people are educated."

The dream that new technologies might radically disrupt education is much older than Udacity, or even the Internet itself. As rail networks made the speedy delivery of letters a reality for many Americans in the late 19th century, correspondence classes started popping up in the United States. The widespread proliferation of home radio sets in the 1920s led such institutions as New York University and Harvard to launch so-called Colleges of the Air, which, according to an article in The Chronicle of Higher Education, prompted a 1924 journalist to contemplate a world in which the new medium would be "the chief arm of education" and suggest that "the child of the future [would be] stuffed with facts as he sits at home or even as he walks about the streets with his portable receiving-set in his pocket." Udacity wasn't even the first attempt to deliver an elite education via the Internet: In 2001, MIT launched the OpenCourseWare project to digitize notes, homework assignments, and, in some cases, full video lectures for all of the university's courses.

And yet, all of these efforts have been hampered by the same basic problem: Very few people seem to finish courses when they're not sitting in a lecture hall. Udacity employs state-of-the-art technology and sophisticated pedagogical strategies to keep their users engaged, peppering students with quizzes and gamifying their education with progress meters and badges. But a recent study found that only 7% of students in this type of class actually make it to the end. (This is even worse than for-profit colleges such as the University of Phoenix, which graduates 17% of its full-time online students, according to the Department of Education.) Although Thrun initially positioned his company as "free to the world and accessible everywhere," and aimed at "people in Africa, India, and China," the reality is that the vast majority of people who sign up for this type of class already have bachelor's degrees, according to Andrew Kelly, the director of the Center on Higher Education Reform at the American Enterprise Institute. "The sort of simplistic suggestion that MOOCs are going to disrupt the entire education system is very premature," he says.

Thrun had assumed that low completion rates in his early classes would be temporary, and during Udacity's early days he continued to spend most of his time at Google, recording his Udacity classes in the middle of the night. His investors had been urging him to expand his role for months, and in May 2012, Thrun informed Page and Brin that he'd have to step down from Google X to focus on Udacity. For the first time in his life, he was now CEO of a company. "There was no one who understood the nuances of what he was trying to accomplish as well as Sebastian did," says Levine, who led a $15 million investment in Udacity, on behalf of Andreessen Horowitz, in October 2012. (Thrun still serves as a part-time consultant to Google X, spending one day a week working there.) "If it hadn't been for Sebastian," says Levine, "we wouldn't have done this investment."

Thrun initially approached the problem of low completion rates as one that he could solve single-handedly. "I was looking at the data, and I decided I would make a really good class," he recalls. Statistics 101, taught by the master himself and recorded that summer, is interactive and full of accessible analogies. Most important, it is designed so that students who are not particularly adept at math or programming can make it through. Thrun told me that he tried to smile whenever he was recording a voice-over, so that even though he couldn't be seen, his enthusiasm for the subject would be imputed to his online students. "From a pedagogical perspective, it was the best I could have done," he says. "It was a good class."

Only it wasn't: For all of his efforts, Statistics 101 students were not any more engaged than any of Udacity's other students. "Nothing we had done had changed the drop-off curve," Thrun acknowledges.

He then set about a number of other initiatives to address this thorny problem, including hiring "mentors," many of them former academics looking for a change, to moderate class forums and offer help via live chats. But he also pursued the more obvious way to incentivize students to finish their courses: He offered college credit. In late 2012, Thrun proposed a collaboration to California Governor Jerry Brown, who had been struggling to cope with rising tuition costs, poor student performance, and overcrowding in state universities. At a press conference the following January, Brown and Thrun announced that Udacity would open enrollment in three subjects--remedial math, college algebra, and elementary statistics--and they would count toward credit at San Jose State University, a 30,000-student public college. Courses were offered for just $150 each, and students were drawn from a lower-income high school and the underperforming ranks of SJSU's student body. "A lot of these failures are avoidable," Thrun said at the press conference. "I would love to set these students up for success, not for failure."

Viewed within this frame, the results were disastrous. Among those pupils who took remedial math during the pilot program, just 25% passed. And when the online class was compared with the in-person variety, the numbers were even more discouraging. A student taking college algebra in person was 52% more likely to pass than one taking a Udacity class, making the $150 price tag--roughly one-third the normal in-state tuition--seem like something less than a bargain. The one bright spot: Completion rates shot through the roof; 86% of students made it all the way through the classes, better than eight times Udacity's old rate. (The program is supposed to resume this January; for more on the pilot, see "Mission Impossible.")

But for Thrun, who had been wrestling over who Udacity's ideal students should be, the results were not a failure; they were clarifying. "We were initially torn between collaborating with universities and working outside the world of college," Thrun tells me. The San Jose State pilot offered the answer. "These were students from difficult neighborhoods, without good access to computers, and with all kinds of challenges in their lives," he says. "It's a group for which this medium is not a good fit."


BEEP-BEEP-BEEP.

A 43-year-old instructor named Chris Wilson sits hunched over a tablet computer in a soundproof recording studio--one of three in Udacity's offices--and hits a button that emits three quick tones that indicate the start of a new take.

The room is dark except for two bright drafting lamps pointed at the table. A digital camera mounted above his head records everything he writes, and a small headset microphone--the kind worn by megachurch pastors and TED talkers--records everything he says. Lounging on a beanbag chair just outside the studio is Udacity course developer Sean Bennett, who is staying close at hand in case Wilson needs help with a last-minute revision. All Udacity classes are scripted and storyboarded in advance by the same five-person in-house team, which means they generally look more uniform and polished than those offered by the competition. "A lot of the scripting process is thinking about what the students are going to be doing," Bennett says. "The words are mostly Chris's."

I watch as Wilson--a big man with wavy shoulder-length hair, wearing a baggy T-shirt and cargo shorts--struggles to communicate a web-development concept called fluid layout, which allows pages to render properly on differently sized screens. "Now, fluid layout means I should stop fixing all those width--eh. All right."

He tries again, and then stumbles a few words later. "The average for me is probably about three takes," he says.

If Wilson seems slightly unprofessional as an educator, that's because his only formal teaching credential is as an assistant scuba-diving instructor. Wilson works at Google as a developer advocate in the company's Chrome division. His class was conceived, and paid for, by Google as a way to attract developers to its platforms. Over the past year, Udacity has recruited a dozen or so companies, including Autodesk, Intuit, Cloudera, Nvidia, 23andMe, and Salesforce.com, which had sent a couple of reps to discuss a forthcoming course on how to best use its application programming interface, or API. The companies pay to produce the classes and pledge to accept the certificates awarded by Udacity for purposes of employment.

Udacity won't disclose how much it is making, but Levine of Andreessen Horowitz says he's pleased. "The attitude from the beginning, about how we'd make money, was, 'We'll figure it out,'" he says. "Well, we figured it out."

Thrun, ever a master of academic branding, terms this sponsored-course model the Open Education Alliance and says it is both the future of Udacity and, more generally, college education. "At the end of the day, the true value proposition of education is employment," Thrun says, sounding more CEO than professor. "If you focus on the single question of who knows best what students need in the workforce, it's the people already in the workforce. Why not give industry a voice?"

Thrun's friends and colleagues repeatedly told me that he has a great capacity for intellectual flexibility. "Most founder–CEOs have this belief that their vision of the universe will prevail and everyone else's vision will lose," says George Zachary, a partner with Charles River Ventures and Thrun's first investor. "Sebastian is the opposite. He's so far away from Steve Jobs on the CEO spectrum, it's amusing."

Still, I couldn't help but feel as if Thrun's revised vision for Udacity was quite a comedown from the educational Wonderland he had talked about when he launched the company. Learning, after all, is about more than some concrete set of vocational skills. It is about thinking critically and asking questions, about finding ways to see the world from different points of view rather than one's own. These, I point out, are not skills easily acquired by YouTube video.

Thrun seems to enjoy this objection. He tells me he wasn't arguing that Udacity's current courses would replace a traditional education--only that it would augment it. "We're not doing anything as rich and powerful as what a traditional liberal-arts education would offer you," he says. He adds that the university system will most likely evolve to shorter-form courses that focus more on professional development. "The medium will change," he says.

It might already be changing. This January, several hundred computer science students around the world will begin taking classes for an online master's degree program being jointly offered by Udacity and the Georgia Institute of Technology. Fees will be substantial--$6,600 for the equivalent of a three-semester course of study--but still less than one-third of what an in-state student would pay at Georgia Tech, and one-seventh of the tuition charged to an out-of-state one.

It's a bold program, partly because it is the first accredited degree to be offered by a provider of massive open online courses, but also because of how it's structured. Georgia Tech professors will teach the courses and handle admissions and accreditation, and students will get a Georgia Tech diploma when they're done, but Udacity will host the course material. Thrun expects the partnership to generate
$1.3 million by the end of its first year. The sum will be divided 60-40 between the university and Udacity, respectively, giving the startup its single largest revenue source to date.

Crucially, the program won't ultimately cost either Udacity or Georgia Tech anything. Expenses are being covered by AT&T, which put up $2 million in seed capital in the hope of getting access to a new pool of well-trained engineers. "There's a recruiting angle for us, but there's also a training angle," says Scott Smith, an SVP of human resources at the telco. Though Smith says the grant to Georgia Tech came with no strings attached, AT&T plans to send a large group of its employees through the program and is in talks with Udacity to sponsor additional courses as well. "That's the great thing about this model," Smith says. "Sebastian is reaching out to us and saying, 'Help us build this--and, oh, by the way, the payoff is you get instruction for your employees.'" Says Zachary, "The Georgia Tech deal isn't really a Georgia Tech deal. It's an AT&T deal."


I first became acquainted with Thrun's work nearly 10 years ago, in a very traditional university setting. I was getting my bachelor's degree in English--an experience that, I must say, taught me very little of obvious professional value but nonetheless seemed worth the outrageously high price--and had been required to take three science classes. In the final semester of my senior year, I took an introduction to mechanical engineering, where the professor showed us a video of the first DARPA Grand Challenge. I remember being moved by the quiet beauty of a driverless car winding up hills in an empty desert, and when I saw pictures of Stanley the following year, I felt a sense of awe, like a little boy getting a good look at a car for the first time.

I tell Thrun this, and he seems flattered. "They put it in the Smithsonian Air and Space Museum," he says proudly. "So now a lot of 8- and 9-year-olds know who I am."

Thrun's 5-year-old son, Jasper, is not yet old enough to be impressed by his father's work, but he's already starting his education. "In my son's kindergarten, they're telling us how to get him into Stanford," he says. "By their advice, I'm doing everything wrong, because I'm trying to make him happy rather than putting him through as many piano lessons as possible." He dreams that his son will take a less conventional view of education. "I hope he can hit the workforce relatively early and engage in lifelong education," Thrun says. "I wish to do away with the idea of spending one big chunk of time learning."

I ask Thrun if it isn't odd that someone like him--someone for whom the traditional education system has done so much--would wind up railing against it. "Innovation means change," he says. "I could restrict myself to helping a class of 20 insanely smart Stanford students who would be fine without me. But how could that impact not be dwarfed by teaching 160,000 students?"

All visionary entrepreneurs must, at some point, find their own sense of romance in the compromises they make to build a profitable business, and the size of the crowd is where Thrun finds his. He's moved by the idea of many, many students from many, many places learning something because of him--even if it's something as mundane as a Salesforce.com API. I have a hard time believing that he really wants his son to get Salesforce certified rather than Stanford educated, but in this one thing Thrun seems entirely earnest.

Two days after our bike ride, I return to the Udacity offices, where Thrun is rerecording a segment for his statistics class. He'd mistakenly used an incorrect notation in writing out a math problem, and he's returned to the studio to get it right, spending an hour or so alone in the dark room, talking into the microphone and scribbling on a tablet. "It's kind of like being onstage, where you have all these lights in your face and can't see the audience, but you still have to be able to excite them," he says. "So I think of the football stadium full of people that I'm facing. I get a kick out of that." Thrun's taken the red pill. There's no going back.

[Photos by David Black | Illustrations by Brown Bird Design]