Anouthinh Pangthong discovered coding in prison. He never graduated from high school after joining a gang at age 12. By age 15, he received a conviction that would put him in prison for decades. “Education was the last thing on my mind,” he said in an interview. “I wasn’t a student.”
Behind bars, he grew fascinated with the potential of computer software. But the few coding courses available in prison were often shut down without warning. Instructors taught COBOL, an outdated programming language from the 1960s. Without internet access, Pangthong wrote down functions on scraps of paper in his cell, and returned to class the next day to see if they worked. It wasn’t until age 38, more than two decades after entering prison, that Pangthong was released from San Quentin maximum-security prison.
It was his first chance to attend a coding class outside of confinement. The bootcamp General Assembly accepted Pangthong in its 2019 class. He enrolled in a user experience design course and said he thrived on learning the fundamentals of technology and design. He interned at Adobe before finding a steady job after graduation at the criminal justice reform non-profit, Restore Justice. “I really felt I had an obligation to show what a formerly incarcerated person looks like,” Pangthon said in an interview.
Who are bootcamps for?
Coding bootcamps love promoting stories like Pangthon’s; General Assembly connected us to him. Rags-to-riches and redemption stories reinforce the mythology the technology industry tells about itself: Anyone with the smarts and grit can succeed rising through the ranks to work at Facebook or Google. And success stories do happen.
What is seldom mentioned is that these stories are exceptions.
Today, more than 133,000 students have graduated from coding bootcamps with names like AppAcademy, General Assembly, Hack Reactor, Metis, and Lambda School, according to CareerKarma, a firm that connects students with coding programs. These short, intensive training schools are designed to turn capable students—many of whom already have a college degree—into entry-level programmers, product managers, data scientists, and more.
Most share a similar model. Students apply, but unlike college, GPA matters less than aptitude, experience, and a demonstrated commitment to learn. Once accepted, students can receive as much as 100 hours of pre-work before arriving. The classes—online, in-person or some hybrid—deliver basic concepts and coding exercises. The most popular subjects last year were web development, user experience design (UX), digital marketing, software engineering, and data science. Students are usually grouped into teams organized around projects. Over several intensive weeks, they are expected to build a portfolio, demonstrate technical skills and teamwork, and then go find a job in the tech sector where developers and designers can, theoretically, command six-figure starting salaries.
It’s been a winning formula for the schools. The number of students graduating from coding bootcamps in 2019 hit a record 34,000 and generated nearly half a billion dollars in tuition, estimates CareerKarma. Venture capitalists have poured more money into educational bootcamp startups, according to the private equity research firm PitchBook, hoping to spread the model to sales, marketing, manufacturing, and other sectors despite a rash of school closures in the past few years.
The attraction of bootcamps is simple: Learn coding essentials in a few months at a fraction of the cost of a four-year computer science degree. With out-of-state tuition for a Bachelor’s topping $171,752 and bootcamps charging between $15,000 to $20,000 for a few months, the decision appears easy.
As the industrialized world grapples with rising inequality and stagnant wages, these schools promise they can train the next generation of software engineers for a fraction of the time and money of traditional universities. For thousands of people stuck in dead-end jobs, it looks like their best shot at a better life. But for every story like Pangthong’s, there are students who never even get near the classroom, let alone a stellar job. While bootcamps catapult some people from behind the Starbucks counter to Facebook, it’s more realistic to think about them as finishing schools for college graduates to enter the technology industry.
Pangthong recalls standing at the orientation for his first General Assembly class last year and realizing he was the only one without a diploma. “I stated my name, where I studied—San Quentin state prison,” he said. “Most of my cohort said the schools they attended: Berkeley, UCSF, UCLA. I just felt like I wasn’t supposed to be in this room among all these people.” He praised his class and teachers for rallying behind him, but “I wasn’t really prepared for GA,” he said. “Everything was new.”
A far more common student profile at bootcamps is someone like Denise Shephard, a mid-career computer scientist from Mills College who studied full-stack web development at Galvanize before joining the non-profit EducationSuperHighway as a vice president. Or Steven Gordon, a graduate of General Assembly class who now works at Slack as a backend engineer and holds three degrees from Yale and Harvard (a BA, JD, and MBA). A LinkedIn review of alumni profiles from several bootcamp class cohorts, as well as the school’s own audited review of class demographics, confirmed this trend.
This disconnect between the promise and reality of bootcamps has sparked furious debate online, as have some very public accusations surrounding Lambda School this month. While bootcamps undoubtedly work for some people, they may be setting others up for failure by disclosing misleading job placement statistics, and recounting stories of wildly successful graduates without setting expectations of what they can realistically deliver for most of their students.
Are coding schools really a chance for a better life for most people or, as many critics charge, are they scams?
Origin story
More than 70,000 college students graduated with a computer science degree last year, reports the National Center for Education Statistics, but it is not enough. The number of open computing jobs has far outstripped students graduating with CS degrees since the 1980s. In 2011, with salaries skyrocketing and tech companies desperate to hire, people began to ask: What if you could skip college, and just train people to code?
That year, a co-working space called General Assembly opened its doors to people teaching coding classes. Within a short time, the concept exploded. Students eager to learn marketable skills for a job as a designer or software engineer flooded into classrooms. Today, more than 100 coding bootcamps operate in the US (even as many others foundered) with the premise that anyone with the dedication could learn to code to address a deepening tech talent shortage. Last year, 33,959 people graduated from a coding bootcamp.
The attraction of some of these bootcamps is that you only pay if you land a job. While students can take out loans or pay cash for tuition, more than two dozen schools are using a novel approach (for vocational training) called income share agreements (ISA). It builds on a concept proposed by libertarian economist Milton Friedman in the 1950s known as human capital contracts. Instead of charging tuition, students pay a portion of their salary if they land a job with a minimum salary. If not, they owe nothing. Though terms are not yet standardized, a CareerKarma survey of 27 US boot camps offering ISAs found the average contract required students to pay 13.8% of their monthly income after they began earning at least $42,476. Total payments maxed out at $32,754 or within 38 months.
While ISAs make up just a tiny sliver of the more than $200 billion (pdf) in education loans and financial disbursement made each year, investors are starting to buy portfolios of such agreements. The ISA startup Blair says investors can expect 10% annual returns based on historical data from Australia, Germany, and elsewhere in Europe where ISAs account for a significant part of student lending.
The goal is not to displace subsidized government student loans (which are still a good deal, especially when combined with income-adjust repayment plans). It’s to provide easy access and an insurance policy for those otherwise unable to pay tuition, says George Jian, an investor at University Ventures Funds. He believes ISAs could compete against the $10 billion in private student loans issue each year—about 10% of the total—which can carry interest rates as high as 13%.
Done right, advocates say ISAs free students to attend school without risking going deep into debt, since students can pay back loans over time with higher salaries. But Robert Kelchen, an education researcher at Seton Hall University in New Jersey, warns “not a lot of rigorous research evaluating ISA has been done because it’s so new.” While he estimates only 1,000 or so US college students have signed up for ISAs, bootcamps have embraced the model. Edly, which helps package income-sharing agreements for accredited investors to back, predicts schools will offer at least $500 million worth of ISAs agreements in 2020.
Of course, the model’s promise rests on landing its students a job. And that’s where it seems to be breaking down for some.
Get a job
Mark Bickford thinks bootcamps are promising more than they can deliver. A lawyer who attended the Lighthouse Labs coding bootcamp in Vancouver, Canada in 2017, he says the program never placed 93% of grads in jobs within 120 days, as it advertised. “It’s not even close to accurate,” he said estimating the percentage in his cohort as closer to 15%. “I think there is a huge group of people out there, including myself, who were duped/disappointed and chose to keep quiet out of embarrassment, feeling like you are part of the problem, and just moving on with life. These schools are largely a scam.”
While Bickford called the school’s course material “adequate,” he said he received little help from career services, and few interview leads. A “job seeking” agreement mandated constant self-directed study to upgrade coding skills, coding meetup attendance, and short-notice interview availability to even qualify as part of the “93%” who find a job. “It’s basically the same as if you were self-taught and just tried to network yourself into a job,” he says. “Students that received jobs either had prior coding experience or tapped personal contacts to find work. It’s not clear that Lighthouse helped any student from my class find lasting work.”
Lighthouse disputes this. Its operations director, Nicole Husain, pointed me to a student outcomes report (pdf) published in 2018 showing 93% of its 992 of its job seekers (out of 1,504 graduates) had found jobs within six months of graduating. It lists a range of reasons for excluding the school’s remaining graduates from the pool, including some that have enrolled in school, remain in corporate positions they never left, or have adopted freelance careers. “We don’t want to take credit for that,” she said. If these graduates are included, the job placement rate would be 64%. Although the report itself was not audited by a third-party, the numbers were validated by a review of alumni identifying as Lighthouse Lab graduates: 95% held jobs with either “developer” or “engineer” in their job titles.
It’s all in the numbers
Bickford’s assessment of the purpose of bootcamps—to secure a job—isn’t wrong. For most graduates, a better job and higher income are the reasons to spend months of intensive work to learn a new skill. That’s driving enormous pressure to keep employment numbers as high as possible.
But bootcamps still operate in a largely unregulated, opaque environment, leading to tangles with authorities. In 2014, California regulators began sending cease-and-desist letters to bootcamps which had failed to register with the state. Some did, but many schools have had to negotiate with state regulators to stay open. Recently, California regulators ordered the Holberton bootcamp to shut down for failing to meet “institutional minimum operating standards” and for fraudulent practices. Students went public describing its “world-class” curriculum as a series of Google links. (It’s still operating.) More recently, coding bootcamp Lambda failed to register with California’s education regulator, incurring a fine of $75,000. (Lambda says it will pay the fine and register.)
This freewheeling environment has encouraged some schools to make outright misleading (if not technically false) statements. Early on, the most successful schools claimed that more than 90% of graduates found jobs. In 2014, for example, Hack Reactor (acquired by Galvanize) claimed it “maintains a 99% employment rate and a median graduate salary of $110,000.” Many got hooked on these statistics, but they have steadily dropped as more students enter bootcamps and it’s not quite as easy to place so many junior developers all over the country. But schools have struggled to let go of those numbers, resorting to, at best, some creative accounting.
Sheree Speakman runs the Council on Integrity in Results Reporting (CIRR), an industry-backed non-profit that publicly discloses student outcome reports submitted by schools. Speakman, who has an accounting background, says she doesn’t suspect outright fraud in the reporting numbers, even among those not disclosing their numbers to CIRR. “A surprising number of them are honest,” she said by phone, even if they spin the numbers. “When they don’t have to count 100% of their students, their figures change: ‘I shrunk the denominator and declared ourselves victorious.’”
Speakman plans to mandate bootcamps reporting to CIRR count all students when disclosing placement rates, and secure a third-party audit. But Speakman believes in the model. “Coding bootcamps have set a new standard for accelerated learning,” she said, noting their bootcamps job placement numbers still compare well to outcomes she has seen in sectors such as healthcare. “The accelerated work programs for coding are your best shot at a different and successful life.”
But the bootcamp model, and ISAs, are untested for large numbers of students. “I don’t think it’s been rigorously evaluated,” says David Autor, an economist at the Massachusetts Institute of Technology. “It’s certainly not for the randomly selected adult.”
Lambda School, a high-profile bootcamp, will test this hypothesis. Founded by tech entrepreneur Austen Allred in 2017, Lambda is one of Silicon Valley’s newest coding schools. The online bootcamp—motto: “free until you get a job”—has raised $48 million from investors including Google Ventures, Y Combinator, and Ashton Kutcher, according to PitchBook. It offers four tracks—data science, full-stack web development, iOS coding, and UX design—all paid for through ISAs.
Allred says Lambda’s success so far (it has 2,500 students) is not due to its curriculum, most of which is already free online in some form across the web. Instead, he says, Lambda built an online classroom environment that is far more efficient than physical classrooms to teach large numbers of students, as well as a pipeline to help get students hired. Most of Lambda’s technology is off-the-shelf: Zoom for video conferencing, Slack for chats, YouTube for video hosting.
Its classes are designed around a thesis first proposed by educational researcher Benjamin Bloom in 1984: you can turn average students into superior ones by giving them personalized tutoring and mastery learning (students repeat material they haven’t fully grasped). Allred calls it “our core operational thesis.” Offline, it’s far too expensive to deliver such personalized attention to everyone. But Lambda has tried to do this by having participants and teachers work together online, and ensure students don’t progress until they’ve mastered a topic. They use technology to automatically track individual students’ progress and then correct any deficiencies.
Lambda has hired more than 85 people for its learning and instruction team, 30 contract instructors, as well hundreds of students (paid about $13 per hour) to be teaching assistants known as “team leads” for eight people from incoming classes. Recruiters then try to place these students at companies around the country.
“The function of a school like ours is to figure out where there are gaps in the labor market and fill them,” says Allred in an interview at their office in San Francisco. “The economy is really, really bad at distributing labor where it is most valuable.” Lambda is hoping to solve that by teaching thousands of students and the workers to ensure they all have jobs in communities around the country. As Lambda attempts to grow by a factor of five this year, it’s the biggest challenge facing the startup valued at $150 million by investors. “It’s not that the jobs aren’t there,” says Allred. “It’s the ability to get people in the right job.” Lambda is now on a hiring spree to find job placement specialists around the US, but Allred was confident that “what we have now is scalable.”
Yet this month, Lambda School endured a tornado of bad news. A scathing letter written by students in its user experience course castigated the school for “unprofessionalism, incompetence, and disorganization” and asked to be released from their ISA requiring them to pay 17% of their salary once they began earning $50,000 or more (up to $30,000), according to The Verge. Student work posted to GitHub, a public code repository, was panned as shoddy without instructor feedback by one expert. Start dates had been inexplicably delayed, a costly expense for students. And internal documents revealed Lambda had given investors a lower six-month job placement rate than the 86% advertised on its website, and was selling some ISAs to investors as a way to fund operations, rather than holding on to them.
All that news belied more than the usual startup chaos. Former UX students took to Twitter to disparage the school. Taylor-Nicole Watson, a designer in New York, said program coursework was chaotic and constantly changing. “I’d click the Zoom link, realize nothing of value was being taught and would ultimately have to teach myself,” she wrote. “That’s how about five months of Lambda went…The idea that I wasted so much time and energy on something like this genuinely hurts to think about.”
The firestorm has prompted Lambda’s founder Allred to admit its coursework for the user experience track was inadequate, suspend admission into that program (the coding tracks are still open), as well as release those students from their ISA. He later admitted that some early Lambda cohorts had job placement rates as low as 50%, but claimed outcomes in the first half of 2019 had improved to 70% among the graduates it was tracking. “The most important thing is that our students are happy,” he told me. “Some weren’t happy. We made it right.”
That didn’t mollify everyone. But several Lambda coding students did come out on Twitter and in person to defend the school’s record. Yasirah Krueng, who graduated in May of last year, said she joined Lambda with measured expectations despite having a bachelor’s in mathematics from a university in Indonesia. After a decade out of the workforce, and three kids under the age of ten, she wanted a way into the tech industry after immigrating to the US. “[Lambda] showed me a way to the goal,” she said by phone. “It’s not zero to hero. Coding is not something you can teach in just three or six months.”
Krueng said despite Lambda’s flaws (instructions and coursework were not always stellar) it gave her a chance to produce a body of work and become a software engineer. “Some people think you can leave knowing how to code and graduate, then have a job at Google,” she said “Is that really possible? I don’t think so.”
Lambda is now planning a massive expansion with as many as 10,000 students enrolled in 2020. On Feb 19, it announced a $100 million partnership with Edly allowing it to borrow against the value of the ISAs and finance operations while students complete their courses, and after repaying it, splitting future revenue from the ISA with the loan issuers.
Ultimately, Allred says, the vision is for an integrated school that can efficiently screen, train, and graduate thousands upon thousands of students into well-paid professions with sought-after skills. He wants to expand the model, and the student body. “I don’t think there’s actually a sustainable business model taking the top 1% of the students and just not doing anything,” he said. “Because it turns out everybody wants the top 1% of the students.” While coding was an obvious first target, nursing and other industries with labor shortages and high salaries are next.
Lambda is not alone in casting a wider net. The coding bootcamp Microverse, which just raised $3.2 million, is looking to train tech workers in developing countries. SV Academy offers a three month (part-time) training program in sales and support roles. Others like Pathstream are working with companies to offer certifications—digital marketing, customer relations, data analytics— at community colleges around the country. Cluster is working to train people for advanced manufacturing roles.
Yet this breakneck expansion—or scaling, as Silicon Valley likes to refer to it—may prove too much for schools struggling to meet existing obligations to all of their students.
Jake Schwartz, the founder of General Assembly, said Lambda’s pursuit of Silicon Valley-like growth in a business not amenable to software strategies would backfire. He also doubted whether venture capital was the right way to finance the massive growth Lambda envisioned. Schwartz said he acknowledged this reality by reorienting General Assembly as a talent pipeline for companies. In 2018, Adecco, one of the world’s largest staffing firms, acquired General Assembly for $413 million. “Everybody called us an education company,” he has said. “We were really an employment company.”
Commenting on a story (paywall) in The Information about Lambda’s travails, Schwartz did not mince words. “I’ve literally seen this entire movie,” he wrote. “The secret is, enrolling a bunch of students for ‘free’ is the easy part. Getting them through and hired is really hard, and really hard to scale. It took years for us to figure out processes and systems, they are trying to do it in months. I would guess this is going to get uglier, and more harmful to a lot of hopeful young students.”
The promise of coding bootcamps, and the model more broadly, is an affordable education that leads to better jobs for whom the traditional advice to go “back to school” is not an option. Millions of people stand to benefit. But we’re now in the messy experimentation phase and students may be in for a bumpy ride. But high standards and smart regulation can protect students while fostering new approaches. The digital economy demands lifetime learners, not just college graduates.
“So far, technology has been most successful at increasing the learning opportunities for people who are really good at learning,” said Autor at MIT. “We have not found the kind of secret sauce how to use all this technology to make education fast and cost-effective. But many people believe they are good opportunities. We just haven’t found them all yet.”