Is STEM Education Broken?

After decades of discourse about the humanities in crisis, some progress has finally been made: We’ve moved on to STEM in crisis. The sociologist John D. Skrentny’s new book, Wasted Education: How We Fail Our Graduates in Science, Technology, Engineering, and Math, purports to identify a growing problem with American STEM education, which is that college graduates who major in the sciences fail to stay in STEM jobs, fleeing to other industries even as employers bemoan a chronic shortage of skilled workers.

The Census Bureau’s National Survey of College Graduates shows “only a paltry 28 percent of STEM graduates working in STEM jobs,” Skrentny writes, “despite the massive governmental, philanthropic, and personal investment” in STEM education. His first explanation of what he calls “the exodus from STEM” seems sensible enough: He posits that graduates who major in the sciences leave STEM employers for better-paying jobs elsewhere. But this simple economic explanation isn’t enough for Skrentny; if that’s all that were going on, Wasted Education would be a short book. So he goes on to diagnose a series of disparate maladies that he believes afflict scientific workers, and offers some prescriptions.

To get clear on one thing right from the start: Wasted Education is not really a book about STEM at all: It’s a book about tech. While Skrentny makes vague gestures at “life sciences, energy, and chemicals,” they only serve to underscore his myopic focus on computer programming. While Skrentny stresses that STEM workers help to develop clean energy sources, new vaccines, and other innovative products that serve the public good, the book’s examples and statistics are overwhelmingly drawn from tech companies like Facebook, Google, and Uber, where the case for their benefit to society is, to say the least, less obvious. Even the pharmaceutical industry doesn’t get a look.

That omission poses a particular problem for Skrentny’s thesis because — as he himself notes, citing data from a report by Georgetown University’s Center on Education and the Workforce — it’s the dismal employment-retention rates for math-, physical-, and life-science grads (a meager 24 percent!) that seriously drag down the numbers. There is scant analysis of such jobs in this book, however, which focuses overwhelmingly on computers. When one considers computer-science graduates in isolation, the picture is quite different: Sixty-four percent of them go into technology jobs immediately after graduation. Ten years later, the number of computer-science grads in STEM jobs actually rises, to 72 percent. That level of retention is considerably higher than the 53 percent of business majors still in business jobs 10 years on. So … what’s the problem exactly?

Another question: Why should we, as a society, care whether or not STEM majors stay in tech-industry jobs? Even within the tech field, many more of the jobs Skrentny considers involve advertising analytics and streaming media than any loftier goals in the service of humankind. Strangely, Skrentny admits as much in his last chapter, where he acknowledges that “a lot of tech business models, their way of generating profit, is simply to sell stuff and not improve life.” The STEM-based skills that tech employers value most highly are those that can be used for “tracking, profiling, and selling,” pursuits that Skrentny characterizes as “morally stressful.” Thus, on page 158 of 180, we discover we have been reading a book about tech companies driving people away from their own intrinsically harmful, stressful, and toxic jobs. And this is supposed to be a bad thing?

The chapters that precede this strange conclusion are a mixed bag. Skrentny spends far too much time on problems — wages and efficiency pressure, increasing time demands, devaluation by American society — that aren’t specific to tech fields, or even to STEM fields generally. The book’s lack of analytic clarity is compounded by Skrentny’s apparent lack of familiarity with the tech sector itself. Skrentny’s background is in studying affirmative action and civil rights, with little focus to date on STEM or technology. His book relies almost entirely on secondary and tertiary sources, many taken from popular news outlets like BuzzFeed and The Verge. The acknowledgments thank many (primarily non-STEM) academics and public-sector employees but few (if any) STEM workers, and there is little evidence in the text of any direct interaction with the individuals the book purports to study. Unsurprisingly, then, Skrentny makes several errors, such as treating the “Google union” as though it were a real union, when it is in fact an informal group of employees and contractors not officially recognized by the National Labor Relations Board, much less Google itself.

Wasted Education is similarly inadequate when it comes to treating the discriminatory practices within tech workplaces. Racial discrimination is a very real phenomenon at tech companies, but it is also far more complicated than hackneyed popular press accounts portray it as. I speak from experience, as a former software engineer at both Microsoft and Google. In fact, tech was by far the most racially diverse field I have worked in, while simultaneously exhibiting racial bias in both more and less common forms. Google’s technical work force has been majority non-white since 2020 (though Black people only make up around 3 percent of that work force). The interplay between South Asian, East Asian, American, and European subcultures within tech companies is multifaceted, and discrimination is frequently not unidirectional. The industry’s gender imbalance — women make up only a quarter of Google’s technical employees — only complicates the picture further.

Skrentny’s own accounts of the tech sector’s corporate culture, however, tend toward caricature. In his discussion of diversity, he quotes a litany of stereotypes from Chaos Monkeys, an autobiography by the tech entrepreneur Antonio García Martínez, who describes a hypothetical “soft-spoken Indian or Chinese engineer, quietly competent but incapable of the hard-charging egotism that Americans almost universally wear like blue jeans.” I know some hard-charging Indian and Chinese engineers that Skrentny — and, apparently, García Martínez — should meet. (And yes, I even know some soft-spoken American engineers, hard as it may be for some to imagine.)

It’s even more questionable that Skrentny has chosen to cite García Martínez as an evidential authority, given that, elsewhere in Chaos Monkeys, you can find passages like this:

Most women in the Bay Area are soft and weak, cosseted and naïve despite their claims of worldliness, and generally full of shit. They have their self-regarding entitlement feminism, and ceaselessly vaunt their independence, but the reality is, come the epidemic plague or foreign invasion, they’d become precisely the sort of useless baggage you’d trade for a box of shotgun shells or a jerry can of diesel.

Such chauvinist balderdash should have provided a hint that García Martínez was perhaps not the most reliable source on discrimination in tech culture.

In any event, Wasted Education utterly fails to explore the complex ethnic and gender dynamics of engineering with any subtlety or insight. “Especially in tech and engineering,” Skrentny writes, “employers maintain organizational cultures that are based on the assumption that workers are young and male, and typically White and American.” With this statement, he contradicts his earlier assertion that tech employees are equally divided between whites and Asians (a data point that causes him to exclude Asians from the category of “underrepresented minorities” in his analysis).

Worse, Skrentny doles out praise where it is not due, as when he compliments Uber (infamous for its toxic work culture) for “having women sit in on interviews [to allow] interviewers to see how workers and prospective hires treated women in a professional setting.” True, this is not a bad idea, but Google bettered it 20 years earlier by mandating a woman interviewer on all engineer hiring loops. (I do not know if this is still the company’s policy.)

Skrentny’s lack of familiarity with the tech sector also manifests in his discussion of “retraining.” He is undoubtedly correct in saying that tech firms tend to prefer hiring recent graduates with relevant skills over retraining existing employees, but errs in treating “retraining” as a single homogeneous enterprise. Skrentny presumes a roughly equivalent capacity for productivity, flexibility, and learning among the workers independent of what particular skills are being considered. But not all “skills” and “retraining” are equal: Jumping from C to Python is very different from jumping from R to TensorFlow, for example. Moreover, retraining can be informal. I didn’t take any official “training” during my time as an engineer, but there were many occasions where it was understood that I was taking the time to learn this or that particular subsystem or utility because the problem at hand required it. That, too, is training. On the other hand, I know many tech workers who have endured hours of unnecessary and useless on-the-job training.

Perhaps the most significant blind spot in Wasted Education is political. Skrentny takes for granted that much of the innovation that he wishes to see (ecological, primarily) will emerge from the private sector itself. His call for change is oddly muted, essentially hoping for companies to reform themselves:

Encouraging firms [through federal laws and government regulation] to choose the high road on their treatment of workers and in their business models could help keep STEM grads in STEM jobs. Cutting off low-road business models could also help keep STEM grads in STEM jobs and thus improve the return on investment in STEM education by driving investor dollars to more useful companies, creating more job opportunities at these companies, and making it less likely that STEM grads will say their job sucks, is embarrassing, or (in the worst case I’ve discussed) makes them feel as though they have blood on their hands.

Indeed, it is devoutly to be wished that capitalist firms would “choose the high road,” and that investors would take more of a rooting interest in “useful companies.” But there’s little elsewhere in Wasted Education to suggest that the current system, even with regulation, will redirect consumer profiling investments into clean energy investments, at least not in the absence of a clear profit motive. Even were that profit motive to appear (and it may well do so, as climate crises accelerate), Skrentny does not convince me that most of today’s STEM workers could be retrained into the more socially beneficial fields he imagines. Nor does he persuade me that such “useful companies” would treat their workers any better than tech companies do today, that profit motives wouldn’t redirect seemingly beneficial efforts in less effective or more harmful directions, or that STEM workers wouldn’t still jump ship to more lucrative, less stressful jobs if given the chance.

Ultimately, Wasted Education fails to offer a coherent etiology for the problem it purports to diagnose. Yes, STEM workers are quitting en masse, but is that because of corporate culture, capitalism, consumer-oriented technology, or existential despair? These four factors are hardly independent, but neither are they identical, and Skrentny slides from one to another over the course of the book without setting down lines. Just as Asian workers slip in and out of the “underrepresented” bucket over the course of the book, Skrentny’s implied culprit is a similarly moving target.

The larger context for STEM workers’ present ennui, which goes unmentioned by Skrentny, is the massive shift from public-sector to private-sector investment in science and technology in the late 20th and early 21st centuries. Skrentny, like many others, fails to recognize how profound that shift has been. The primary driver of STEM research for much of the 20th century was the U.S. federal government; the reason was the Cold War. Tons of money was poured into highly speculative, and frequently unprofitable, research on the grounds of keeping pace with the Soviet Union. This had the twin impact of providing large amounts of grant-driven public-private partnerships as well as placing universities at the heart of such research. Such government investment gave us a tremendous amount of early AI research, as well as, of course, the internet.

With the end of the Cold War and the rise of the personal computer, the balance drastically seesawed in the other direction. Not only does STEM investment today come overwhelmingly from the private sector, but the immense technological and material requirements for a good deal of research, particularly on AI and pharmaceuticals, are such that universities do not even have the resources to perform such work. This means that Ph.D.s who want to do innovative research have nowhere to go but the private sector.

The seesaw could always tip back. With the right combination of societal upheaval and economic and ecological crises, the federal government may once again assume a more aggressive role in directing the advancement of research. Granted, the case of the Cold War shows that the compromises in public sector-driven research can be as troubling as those in the private sector. But I am surprised that Skrentny’s focus remains so narrow and his ideas for reform so limited, especially given his stated belief that most present-day STEM jobs aren’t ethically justifiable.

Such a shift back toward government investment in science and technology would redefine the role of STEM education in a salutary way, but Skrentny seems unable to even imagine this. Throughout his book, he assumes the sole purpose of a STEM education is to provide vocational training for a future industry job. Indeed, Skrentny does not give much indication that a STEM education could be of value in itself, beyond its value as preparatory training. In reality, however, learning the fundamentals of computer science, biology, physics, or a host of other scientific fields can give one generalized cognitive tools in much the same way as studying philosophy or literature can. I would argue, in fact, that the conceptualization and rigor demanded by many of these fields would have made Wasted Education itself a better book.

One doesn’t have to be a technologist to see the value of the sciences in and of themselves. Plato required that the Guardians of his ideal city study pure geometric mathematics for 10 years, in spite of the seeming impracticality of the material. Theorists of education from William James to John Dewey to C. P. Snow have stressed the need for a degree of scientific and technological literacy among the populace at large, and this is only becoming more important as technology colonizes more and more of our lives. It is a sad comment on our view of STEM and education more generally that even advocates of science and technology like Skrentny proceed from the assumption that STEM is a vocational toolbox and nothing more. It doesn’t have to be this way.

My own experience may be instructive here. Computer science, as it was taught to me in college, was not intended as vocational training per se. The emphasis was not on learning any one particular programming language or domain, but on foundational principles for conducting future research in applied and theoretical computer science. Programming was sometimes — not always — a part of such research, but it was expected that students would learn many such things on their own. This is far from Skrentny’s view: His reigning assumption is that STEM majors learn STEM skills for STEM jobs.

The greater problem, of which Wasted Education is one symptom, is the increasing lack of the notion that knowledge can be generalized at all. We seem to have convinced ourselves that higher education must always serve some instrumentalized purpose, whether occupational or political. Reading literature with a myopic focus on “decolonization” is actually not so far from being taught computer science with a myopic focus on software engineering, or chemistry with a myopic focus on pharmaceutical development. It clamps down on the very idea of the interplay of knowledge across disciplines, a value central to the project of the Enlightenment and to scientific research itself. Seen in this light, Skrentny’s many instances of the “failure” of graduates to stick to their original discipline are not failures at all. As a STEM graduate myself, I don’t see my departure from the tech industry to become a writer as a failure: On the contrary, it’s a triumph of the totality of my education that it offers me a greater multitude of possibilities and ideas. It’s that larger sense of possibility and capaciousness that we need to bring to STEM education, not Skrentny’s narrow concern over training and job retention.