Of all the arts, dancing is the one I appreciate least. When I read the Arts section of the New York Times, I unfailingly skip all stories on dance without fear of feeling uninformed. Chalk it up to my bottom-quartile dance skills, or to the fact that I once flunked out of a ballroom dance class for failing to dance in a clockwise direction. But my supercilious attitude to dancing is the primary (but by no means sole) reason I find Footloose (i.e., small town passes anti-dancing ordinance, rebellious teenager from Chicago stages uprising) totally hilarious.

In the annals of dance, only one has been truly forbidden, and not by ordinance. Just a few years after Kevin Bacon dance-liberated Bomont, the Lambada – a sensual dance from Northern Brazil – swept the globe as “the forbidden dance.” Dance-lovers everywhere couldn’t resist its forbidden fruit and danced the Lambada through the end of Reagan-Bush.

Kids, nothing important happened in the late 80s and early 90s that wasn’t foreshadowed by Footloose.

In the annals of coding bootcamps, only one is on its way to forbidden status. Lambda School – not named for the forbidden dance, but rather the Greek letter and CS term – is having a tough go of it. Last week, Inside Higher Education reported Lambda is being sued by former students for falsely advertising an 80% job placement rate. In April, Lambda laid off 65 employees. Last year, it was the subject of a withering New York Magazine attack questioning Lambda’s value to students and breaking the news that it had partnered with a company owned by a banker known as the grandfather of collateralized debt obligations (CDOs). Lambda also had the displeasure of tangling with both the California Bureau of Private Postsecondary Education (BPPE) and Department of Financial Protection and Innovation (DFPI): an alphabet soup of regulatory headaches.

On the surface, it’s hard to see what’s not to like about Lambda School. After a year in which thousands of accredited higher education institutions demanded that students borrow tens of thousands of dollars to pay for online courses, Lambda’s no tuition / income share agreement (ISA) model for online courses appears positively virtuous. Plus, rather than ad hoc remote courses, Lambda’s coding courses are expressly designed for online learning.

Dig deeper, though, and it’s clear why Lambda has become a target. The company raised $122M from dozens of Silicon Valley investors like Google Ventures and Ashton Kutcher at a valuation in excess of $200M. And Lambda’s voluble founder, Millennial poster boy Austen Allred, is a college dropout and Y Combinator grad previously best known for sharing every waking thought – mostly about stuff happening in Silicon Valley – with his 167,000 Twitter followers. Tweets from Lambda executives like “if you don’t think Lambda is at least a $100B company you don’t understand the American economy” haven’t helped.

Will Lambda become Lambada School: The Forbidden Coding Bootcamp? It’s an important question not just for Lambda, its thousands of students, and Demi Moore’s ex, but also Lambda imitators popping up around the world and policy makers seeking new models to close the skills gap. It would be awfully convenient if all we needed to do was launch lots of Lambdas and tweet more.

Austen and Lambda have gotten many things right. First, Lambda adroitly recognized that closing the skills gap requires a vertically integrated model including admissions, learning, and a “job finder,” and that “school is only a tiny part of the engine.” Second, Lambda determined that the best measure of student success is an applicant’s performance on pre-coursework. Other postsecondary programs and employers would do well to adjust their application processes accordingly.

Lambda is also doing a good job trying to reduce friction by eliminating tuition from the equation. Why don’t millions of Americans run out and get the digital skills employers desperately need? The first barrier is cost of training and concomitant financial risk. So ISAs can smooth the path to upskilling for millions. Lambda has further tried to reduce education friction by fronting living expenses in return for a longer ISA.

Critiques of Lambda’s ISAs are misplaced. Lambda has been unfairly attacked for ISAs that could cost graduates $30k. Leaving aside that most online master’s programs are priced at least at this level, Lambda’s share only reaches $30k at salaries over $100k – a trade most prospective students would gladly make. (Grads making $50k will share $17k over several years.) And don’t get overexcited about Lambda selling ISAs to investors, including the grandfather of CDOs. Bootcamps typically remain on the hook via a risk-sharing formula. So Lambda won’t make any money unless its ISAs pay off. Moreover, in the very unlikely event risk isn’t shared for one cohort, the only way Lambda will be able to sell ISAs for its next cohort is if the current cohort produces a satisfactory return for investors. So regardless of who ultimately holds the ISA, Lambda must remain focused on placement and remuneration. Which explains why Austen tweets constantly about employment outcomes of Lambda graduates – annoying for sure, but a big improvement over your garden variety postsecondary leader with no earthly idea.

Notwithstanding, Lambda has flaws that are likely to prove forbidding. David Perell, another inveterate tweeter who calls himself “The Writing Guy” (which I guess makes me the “Guy Who Writes About The Writing Guy” – a solemn responsibility I may think deeply about at some point) is merely the latest in a 25-year-long line of self-proclaimed education experts to predict that the Internet changes everything about higher education e.g., online courses will have “Hollywood-level production budgets,” “teaching will become an extremely lucrative profession.” It’s Silicon Valley thinking like this that leads to theoretically scalable, high gross margin, online-only education models like Lambda, and that led the New York Times to call 2012 “The Year of the MOOC” (allowing me to call MOOCs “The Spice Girls of higher education” and fulfill a lifelong dream of quoting “zigazig ha!”).

I understand why Austen and his investors really, really, really wanna love online-only upskilling. But MOOC completion rates still hover around 5%. Asynchronous online training programs, like those offered by Microsoft and Google, aren’t much better. Online bachelor’s programs are almost always well under 50%. With few exceptions, only motivated students who have already learned how to learn can upskill successfully in a 100% online environment (and they’ll probably find their economic way regardless). The last time Lambda disclosed its completion rate (80%) was over three years ago. The sound of crickets as Lambda has scaled makes it hard to believe completion hasn’t fallen significantly.

In addition, survey after survey indicate employers view soft skills as essential and also hard to find, particularly among candidates with the requisite technical skills. How are Lambda students building essential soft skills like teamwork, communication, organization, creativity, adaptability, and punctuality in an online-only environment? Notwithstanding that some portion of the instructional day is synchronous (guided project and standup meeting), if Lambda graduates have the soft skills employers are seeking, it’s almost certainly the result of attracting students who already have these skills in the first place.

Online-only upskilling can work if it’s selective or deployed within an enterprise (selective by definition i.e., no employers are “open hiring” in the way that many education and training programs are open enrollment). Onground coding bootcamps operating online programs at a much smaller scale have reported achieving comparable outcomes. But Lambda’s plan was to grow to 10,000 students this year (on its way to becoming a $100B company). And that means less screening, not more. With an unselective online-only model seeking to scale rapidly, Lambda is likely to end up somewhere between (free) MOOCs and (costly) for-profit online universities, which – given its ISA model – sounds about right.

All these challenges pale in comparison with Lambda’s most serious problem: getting graduates good jobs. The lawsuit alleges that while Lambda was advertising an 80% placement rate, executives knew (and wrote a memo about how) the placement rate was closer to 50%. (Yet another reason not to be like “The Writing Guy”.) This makes more sense. By definition, new pathways like Lambda are unknown quantities for employers. They can grow rapidly (and sustainably) if they have a model for overcoming employers’ abundant and understandable fear of the unknown (i.e., high cost of a bad hire, high churn) – one that literally places students in jobs. But tweeting and praying that graduates find work is not a viable strategy for scaling a new pathway.

The challenge of placing graduates in good jobs is exacerbated by numerous and increasingly complex requirements. As I’ve noted, employers are throwing everything but the kitchen sink into job descriptions. Jobs with 30-50 skill requirements – many digital – are hard to get for graduates of Lambda’s full stack Web development program. The sad truth for Lambda is that America doesn’t have tens of thousands of open jobs for newly minted bootcamp grads fresh off a one-size-fits-all online curriculum. The skills gap is most acute for combinations of discrete technologies: not level 1 digital skills, but level 3 or 4 employer-specific tech stacks. And Lambda doesn’t take students to level 3 or 4. As such, an online bootcamp hoping to scale a uniform curriculum to tens of thousands of students may work at cross purposes to the goal putting graduates into good first jobs, or — at best — is only the first step in a training sequence. So a fair price is probably closer to Codeacademy’s $240 than Lambda’s $30k.

You’d think that Silicon Valley – font of all technology – would understand this. But the Bay Area has a bad habit of confusing and conflating everything (including – apparently – human capital development) with scalable, high gross margin software. And that means blissful ignorance, wishful thinking, or both.

Another problem is that it’s very hard for Lambda to have a good interface with employers. Few schools do. For colleges and universities, the proximate cause is negligence. But for motivated last-mile training providers like Lambda, the problem is that employers aren’t responsive to the entreaties of a single school or bootcamp – not even one with 10,000 students, not even one with Ashton Kutcher. This means even if Lambda were willing to sacrifice scale and high gross margins, it’s hard to strike deals to custom-train for Google or any other employer.

There is a shortcut here: misrepresentation. Up until the New York Magazine attack, Lambda had prominently featured Google on its homepage as a destination for Lambda graduates. How many Lambda had Google actually hired? One. (Now Google is listed as one of many employers that have hired Lambda grads.) Beware the specter of the FTC, which pried a $191M settlement from University of Phoenix for similar tactics.

Despite the many things it’s doing right, without a course correction (pivot in Silicon Valley-ese), Lambda is on its way to having a for-profit college problem (sans squandering of taxpayer dollars). Talented entrepreneurs like Austen should leaf through an unabridged history of fast-growing education businesses that don’t deliver on promises to students. The paradox of job placement from a scalable, high gross margin online-only model risks taking last-mile training to a dark place. And that's not right, because without regard to margins or profits (many models will be nonprofit, or government-subsidized), employer-centric last-mile training models like apprenticeship have the potential to rekindle socioeconomic mobility in the U.S.

While I remain an admirer of Austen’s entrepreneurial verve and commitment to employment outcomes, it’s Lambda’s basic dance move that has it on the verge of being forbidden: if you’re asking students to take any financial risk at all to upskill for jobs where there is a skill gap – i.e., where there are willing payors for that upskilling (employers) – you have an unimaginative business model. And in Silicon Valley, that’s the cruelest dance of all.