This story is part of a series on how we learn—from augmented reality to music-training devices.
Early this summer, the tech entrepreneur Austen Allred was on Reddit, as he often is, when he noticed something suspicious. The coding forums he frequents, where people usually talk about JavaScript bugs and command line functions, were being suffocated by rants and raves on a particular topic: a coding bootcamp called Lambda School, which Allred happens to run.
To Allred, the posts looked suspiciously robotic. Perhaps a competitor was posting spam to get any mention of Lambda banned from the forums. Maybe it was a bootcamp critic on a tear. Allred had heard that people were spreading rumors he was behind the accounts, trying to gin up more exposure, good or bad, for his startup. “People want us to fail,” he says, casting a glance from under the brim of his signature grey and red trucker hat, emblazoned with the eponymous Greek letter. “People want to think that Lambda is a scam, because they want to believe the results we’re producing are impossible.”
Lambda School is an online coding program that’s free until you finish and get a job. The central conceit is an income-share agreement (ISA): students pay nothing while attending the school, and then pay a portion of their earnings once they’re employed. The concept, first proposed by the economist Milton Friedman in the 1950s as a “human capital contract,” has been heralded by some as a market-based solution to student debt. Everyone is on the same page about the goal: finding a good-paying job.
That idea has proven especially alluring to a certain subset of Silicon Valley for whom the phrase “aligning incentives” sends hearts aflutter. Plus, advocates say the model allows for wider access in a space dominated by private student loans. Lambda School has no campus; the video lectures are available wherever students have wifi—and thus infinitely scalable, so long as there are coding jobs to fill (for the moment, there are plenty). In January, Lambda received $30 million from investors including Google Ventures, Y Combinator, and Ashton Kutcher.
As Allred sees it, anyone can be a coder, provided they’re willing to put in the work. The most important quality, he says, is “grit.” That upsets some people, he suspects. “We eliminate the excuses people rely on to say why they’re not successful, or why they’re not rich, or why they went down a different path.”
That message has wide appeal. Allred’s social media following has become so large that even he at times finds it bewildering. On Twitter, @Austen offers a viral mixture of pathos—retweeted tales of students delivered from the brink of poverty to plum coding jobs—and sober charts about student debt and the pace of innovation. He replies to anonymous critiques of the school, bot-written or not, point by point. (So much so that Y Combinator cofounder Paul Graham, a frequent booster on social media, tweeted at him to maybe stand down once in a while.)
By now, the tagline “free until you get a job” has burrowed deep into not only the subconscious of Silicon Valley investors, but basically anyone who has Googled “how to learn programming” in the past three years from their living room in Tallahassee or coffee shop in Kalamazoo. But it’s also, largely, an untested model.
Gregory Barber covers cryptocurrency, blockchain, and artificial intelligence for WIRED.
Critics point out that ISAs are designed not just to pay back tuition, but to maximize financial returns—to the school and its venture capital backers, as well as other investors who might buy an interest in individual ISAs as a new type of asset. Lambda says enrollment, now 2,700 students, is growing at 10 percent a month; it foresees soon bringing the ISA model to other subjects, like nursing. But as the pool of “human capital” grows, what else changes when students become investments?
Allred, who is 29, started what would become Lambda School in 2017. He was living in Utah, where he had grown up in a Mormon family, and had returned after stepping away from his first Silicon Valley startup, a crowdsourcing site where users report and fact-check the news. (At the time, a San Francisco magazine feature chronicled the young entrepreneur’s time living out of his 2002 Honda Civic.) Back at home, he co-wrote a “growth hacking guide,” titled Secret Sauce, with tips for drumming up excitement about your business on Twitter and Instagram.
Lambda School began as a short coding course (the name refers to a concept within the original subject, functional programming). Allred didn’t have a background in teaching, but he was interested in “creating opportunity,” he says. Earlier in life, he’d dropped out of Brigham Young after deciding he wasn’t getting bang for his buck, and had later watched his friends struggle with student debt. So Lambda, Allred decided, would be free. And he saw how the crowded coding school industry, popular with career-switchers after the 2008 recession, was increasingly struggling to sign up students with upfront fees.
Allred was not the first to alight upon the benefits of an income-sharing model. ISAs have been tested at other coding schools and even a few traditional colleges—most notably Purdue, which began a pilot in 2016. They remain a tiny fraction of the $1.5 trillion student debt market, perhaps worth $200 million, and there’s little data yet to analyze how they pan out for students. Nonetheless, the idea is growing—largely due to Lambda School, which observers say represents more than half of the ISA market.
Like most coding schools, Lambda’s curriculum is unaccredited, and self-reports its performance metrics. (State regulations vary for online schools; Lambda was fined $75,000 in March for failing to register with California’s regulator—the result, Allred says, of poor legal advice. Its application is now pending.) The idea is to welcome beginners and career-switchers alike. At nine months, the program is longer than the norm and includes a section on basic computer science concepts. Everyone follows the same course progression, regardless of initial coding ability, but the video structure allows students to repeat coursework if needed.
When you do the math, the ISA terms are steeper than your typical $10,000 coding camp. Once you make at least $50,000 a year working in technology, you pay 17 percent of your income. (That’s a minimum of $708 per month.) The total is capped at $30,000, or after 24 months of payments. However, even if you still don’t have a tech job five years after graduation, your payment obligations end.
Many students graduate happy with the arrangement. Chris Atoki, a 23-year-old from New Jersey, was a few months into a community college program in electrical engineering when he realized he didn’t have a taste for hardware. He ended up in sales at Mattress Firm. After seeing an ad for Lambda School on Facebook, he joined one of the earliest classes, in 2017, and began watching lectures from the mattress store over an iPhone wi-fi hotspot. Learning to code was challenging, he says, but ultimately as transformative as advertised. He decided to stay on as a teaching assistant after graduation—an opportunity to refresh himself on the coursework before entering the job market—and later joined a company in Philadelphia as a developer.
Atoki says his new income, roughly $95,000, puts him on track to pay the full $30,000—far more than he would have likely paid to attend a competitor. He deposits his income in a bank account monitored by a company called Leif, which administers Lambda’s ISAs. Leif withdraws roughly $1,345 each month. That’s more than typical rent in most US cities, but Atoki says he doesn’t mind, since he’s making about three times as much as he did selling mattresses.
Still, fewer than half of his roughly 50 classmates graduated with him. “Lambda School wasn’t taking just anybody, but they were lenient in their policy to accept people,” he says. Some decided the course was too difficult, or balked after taking more time to consider the terms of the ISA.
It’s midday in July, and all across the country, students are logged into their lectures and group projects. Lambda School’s headquarters, on a high floor in San Francisco’s Financial District, are quiet. Allred, a cherubic figure with a brushstroke of strawberry blonde hair, has just finished a brainstorm on the topic of automating Lambda’s admissions; a whiteboard behind him holds a sketch of potential pipelines. Between sips of Coke Zero, he tells me retention has improved since the early days—85 percent of students now graduate. (Lambda says that figure is based on a “representative sample of recent cohorts.”) That’s in part, he says, because his staff had gotten better at picking students who will succeed.
The admissions process involves a few weeks of pre-coursework, followed by a phone interview to assess how likely an applicant is to stick with the school. The hope is to make the process more efficient by taking the human out of the equation. Allred’s team had explored buying credit scores and administering IQ tests, but ultimately found an applicant’s attention to the pre-coursework to be the best measure of “grit.” The school now gets 2,000 applications per week, he tells me. (The company clarified that the number refers to people who start the pre-coursework; “several hundred” complete the process each month.)
Since the $30 million venture capital infusion in January, the school has expanded rapidly, with a goal of 4,000 students by the end of the year. Last October, there were 700. New waves of students begin every four to eight weeks, allowing Lambda to quickly try out new instruction methods. The week of my visit, the school had launched a program designed for students in West Africa. A cybersecurity course is in the pipeline for later this year. Meanwhile, a group of 50 students is participating in a pilot that fronts them living expenses, in exchange for a longer ISA contract.
The grand vision, Allred says, is to build an optimized engine—admissions, learning platform, job finder—that will serve as the basis for an education empire. “School is only a tiny part of the engine,” he says. In time, it will become an alternative to the current college model, offering a wide variety of subjects. To get there, Lambda will first focus on careers in which there’s a shortage of qualified labor, he says. Nursing is a “concrete” next target. Lambda would likely buy an existing nursing school, including its licenses, and integrate it into the ISA model.
For now, though, Lambda is “burning through millions of dollars” a month, Allred says. “The core equation is cost per student, and then the success per student.” He was preoccupied with the latter part, which involves getting students hired so the school can earn money from ISAs. Hiring, he says, is the main limitation on admitting more students. The company has invested heavily in a “sales team” that pitches graduates to potential employers and works to get them prepared for interviews. The challenge, Allred says, is overcoming the stigma of for-profit coding schools, especially among Silicon Valley firms—which is tricky when you’ve been around less than three years. The Blue Crosses of the world are easier to crack than the Facebooks; Google, which is prominently featured on Lambda’s homepage as a destination for graduates, has hired one student, the school says.
“It’s much easier for us to get somebody hired in rural Ohio than it is in San Francisco,” Allred says. That makes sense for many Lambda students, he adds, who want to get hired where they live.
In the meantime, Lambda shifts the equation: For about half of the ISAs, the company sells the rights to a portion of its returns to investors; in return, it gets cash up front. The arrangement isn’t ideal for Lambda, Allred acknowledges. The school has to sell the rights at deep discounts given the risk that returns will never materialize. But, given the school’s first ISA was only paid off in full this May, the arrangement “lets the business operate without going bankrupt,” he says.
Still, that points to a more complicated “incentive alignment” than is often described. Deference to outside investors might encourage a school to issue lots of ISAs and keep instructional costs low, for example, betting high margins on a few successful students will subsidize the other agreements, says Julie Margetta Morgan, a fellow at the Roosevelt Institute who studies student debt. That’s especially true if it proves difficult to link students with jobs. But Allred insists the incentive structure that aligns the interests of individual students with the school’s isn’t broken. “We still hold the bulk of the risk,” he says. Allred doesn’t tell staff which student’s contracts have been financed.
Recently, those issues have been at the center of a wider debate about regulating ISAs, sparked in part by the Trump administration’s interest in making them part of federal student financing. In Congress, a bipartisan bill, supported by the ISA industry, would seek to regulate them as a unique category, distinct from student loans, setting guardrails on certain terms like caps on total payments and a minimum income threshold.
Some, especially on the left, argue that carving out a special category for ISAs could leave out consumer protections and carry additional dangers because the schools control both the financial arrangements and the marketing of their programs. In June, Senator Elizabeth Warren wrote a letter to Secretary of Education Betsy DeVos arguing that ISAs carry many of the perils of private student loans, “with the added danger of deceptive rhetoric and marketing that obscure their true nature.” The desire to maximize returns, she added, could lead to discriminatory practices in admissions and, ultimately, who is put up for hiring.
One concern is whether students recognize they’re getting into a form of student debt, says Margetta Morgan. People might hear “free” and sign on only to have trouble measuring their financial outlook. If students succeed in finding a job, like Atoki did, they’ll likely pay more than affluent students who paid for their education upfront or through loans. Some will ultimately have difficulty making payments. And while students may not be on the hook for the expense of their education if they fail to find a job—a key improvement over private student loans, Morgan notes—that still leaves out the time and money potentially lost from upending their lives for their education. She also notes that rosy terms might change if the economy tanks and companies stop hiring coders.
Backers of ISAs argue more options for students are a good thing, and that the industry puts students first. Leif CEO Jeffrey Groeber notes that, unlike the private loan industry, his company doesn’t allow credit scores to play a role in which ISAs are financed, eliminating one pervasive source of discrimination. At any particular school, all students receive the same ISA terms. The company also takes its role as a gatekeeper seriously, he says, only bringing in investors whose values it judges to be aligned with the schools it works with. (That’s a buffer, for example, against investors who might try to aggressively pursue a student in bankruptcy.)
ISAs aren’t a replacement for federal student loans, says Charles Trafton, president of Edly, which offers an ISA marketplace for investors but doesn’t currently work with Lambda. But he sees good in chipping away at the private student loan market, and opportunities for more investment. (His Edly cofounder, Christopher Ricciardi, was described by the Wall Street Journal as the “grandfather of CDOs,” the bundled financial products that collapsed during the 2008 financial crisis.) Trafton was initially skeptical of the growth pressures on online schools like Lambda, but says he’s been pleasantly surprised by the quality and the outcomes.
This summer, Lambda School began to show some growing pains. In late July, an anonymous Twitter thread highlighted the struggles of students at the school. Most had started in May, in a web development course with 226 students, the largest ever. Multiple students told WIRED that, in addition to the large class size, unexpected changes to the curriculum and a new policy for repeating coursework resulted in some students, especially newcomers to coding, being unable to keep up.
For some, the issues raised questions about whether the school was equipped to lift up students it admitted from all walks of life. “It felt like gaslighting,” says one student from the thread. The marketing had portrayed the school as at once elite, with references to Google and a strikingly low admissions rate, but also accepting of complete beginners who could build a new life from scratch. For all the talk of grit, the school didn’t feel like a level playing field, especially when some students faced personal challenges or lost their housing. Two students of color noted that, despite a diverse student body, the school seemed to have few nonwhite staff members.
According to Allred, the instructional changes were rooted in getting Lambda students into the hiring pipeline sooner and giving more structure to how students progress. It had been a challenge, he acknowledged, for the school to keep track of students moving through courses at different rates. (The school has even explored paying some students to leave.) Lambda offered the class a one-time offer: They could walk away from both the school and their income share agreements. Allred says a handful of students took the deal.
After the complaints surfaced, Allred took a break from tweeting success stories to acknowledge that Lambda wasn’t perfect. The school was trying to get better about handling feedback internally, he wrote, and had been working to improve diversity among the staff. Had the experience shown the limits of grit? Allred’s interpretation was that perhaps social media made it look too easy, and the school needed to get better at communicating how hard students were working to transform their careers and lives. Soon after, he tweeted a poll of students indicating 2.7 percent would be “not disappointed” if Lambda vanished. “They’re the people that I lose sleep over,” he says.
For now, though, he says he’s focused on getting his equation working. That means getting more people hired, faster. The only way to overcome the skeptics is to deliver on his Twitter promises, getting his graduates into those plush tech careers.
More Stories on How We Learn
-
This Galactic Primer Is the Future of AR Education
-
Become a Musician Using Apps and a Light-Up Piano
-
The Secret to Machine Learning? Human Teachers
-
Computer Scientists Really Need to Take Ethics Classes