chasing skills is a bad bet and bad policy

 Danielle Kurtzleben reflects on the myth of the STEM shortage and its analytic problems: 

[I]t’s not necessarily that there aren’t enough science and math scholars out there; it’s that there aren’t enough people out there with the particular skills the job market needs right now. Spending four years doing biology experiments is no guarantee for a job, and indeed might not go as far as a couple semesters of statistics or computer science….

There’s a key distinction to be made here: this doesn’t necessarily say that there is or isn’t a shortage of STEM workers — rather, it might say that there’s a shortage of business workers with sufficient STEM skills.

“If you’re an anthropology major and you want to get a marketing job, well, guess what? The toughest marketing jobs to fill require SQL skills,” Sigelman says. “If you can … along the peripheries of your academic program accrue some strong quantitative skills, you’ll still have the advantage [in the job market].” Likewise, some legal occupations (such as intellectual property law) and maintenance and repair jobs stay open for long periods of time, according to the Brookings report, if they require particular STEM skills.

Perhaps. Kurzleben seems to be a careful and smart writer who is genuinely committed to thinking these issues through without resorting to cliched, unsupported thinking. But it’s very frustrating that Kurtzleben, and essentially our entire elite policy media, doesn’t go a step further: trying to predict what particular set of discrete and limited skills will be useful in the future is a mug’s game. It’s a fundamentally risky way for an individual to behave, and for policy decisions that are supposed to be based on the most good for the most people, it’s incoherent strategy. Jobs in petrochemical engineering have been exploding, because of a largely-unpredictable boom in American fossil fuel reserves. Becoming a contracting engineer for a construction firm was a great idea in 1999, but by 2005, was a very risky proposition. Going to law school was the epitome of mercenary self-interest until, suddenly, it was the epitome of laughable, deluded foolishness. Teaching kids how to code Python now, when they’ll be hitting the job market 20 years from now, is ludicrous, especially in a world where there’s every reason to think that tech firms will continue to have very low employee to market cap ratios and where computers might take over the bulk of coding. Individuals can navigate the markets, if they’re smart, privileged, and lucky. But great masses of people never can. If you’re telling me that you know what every freshman should start studying in 2014 so that s/he can get a good job in 2019, I think you’re full of it.

Instead, we should return to the point of what education has always been about: to teach students skills, yes, but only as part of a larger, more important goal of teaching them soft skills, meta-skills, and habits of mind that enable them to adapt to an endlessly-changing labor market. If you teach a kid how to use a particular kind of database or programming language, you might get them employed for five years, maybe ten. But if you teach them how to think, how to acquire skills themselves, how to be critical interpreters of information, and how to exist as compassionate and ethical members of a democratic society, you may empower them to keep themselves employed for 40 years. We could stop mistaking education as the process of one person giving information to others and rediscover education as a process of mentoring and apprenticeship where teachers work closely with students to develop not just specific skills but a mind that’s capable of acquiring more skills, and of understanding how and why skills become valued in the first place, and of forming moral choices about how these decisions drive society.

Of course, if we do that, then we’re back to the ideals of a liberal arts education, and I’m afraid that embracing that long, illustrious tradition isn’t sexy in policy elite circles, and doesn’t sell books.

I’ve been writing about issues like this on this blog consistently now, because they’re an obsession of mine and because they’re deeply important for our society. I’ve written about how there is no STEM shortage generally, how there is no computer scientist shortage specifically, how the tech industry is profitable in large measure because it employs so few people relative to revenues, how humanities majors don’t underperform national averages, how you can’t possibly blame broad unemployment on people taking “impractical” majors, how we mistake the value of being a star in a field for the value of simply being in that field. As the data about specific questions like the farce of a STEM shortage becomes too obvious to ignore, the policy apparatus slowly evolves. What makes me worry is the possibility that the thought process will simply shift around a little bit, without confronting the actual central problem: a vast embrace of quantitative, scientific, and technology skills as the solution to all of our labor woes, under the false notion that those skills are more “futuristic” than broader ways of knowing and that they necessarily will result in better economic outcomes. That is essentially the calcified orthodoxy of our policy apparatus, and yet it has never been buttressed by much more than prejudice, assumption, and narrative. (And I say all of this as someone whose research is largely computerized and quantitative.)

And of course, admitting that we shouldn’t educate to provide specific and limited skill sets robs neoliberal media types of one of their favorite pastimes: blaming individuals for their own unemployment. “You’d be employed if you got a computer science degree instead of studying French poetry!” combines the preening moralism and the vague, useless embrace of “the future” that are so intrinsic to our media elites, as well as reflecting unquestioned assumptions that are wrong as often as they are right. The truth of the matter is that the world of work has undergone massive changes in a short period of time, that jobs were lost by the hundreds of thousands thanks to destructive and immoral behavior of financial elites, that our problems have been far more about a lack of aggregate demand and the merciless march of automation, and that individuals cannot fairly be blamed for their own precarity. The supposedly impractical Millennial theater major has been the butt of constant invective for years. But she’s not guilty; she’s merely a convenient receptacle of hate for a policy apparatus that has failed her and her generation by the millions. What’s needed is a redistributive public policy that shields people from the random vagaries of the job market, and a return to the definition of education as a public good that teaches our citizens not just how to make money, but how to think, how to be free, and how to live.


  1. Freddie, do you know off hand of any resources for research related to teaching this soft skills, vs. education based on curriculum that’s more specific and content/information oriented?

    1. There’s a lot out there. Unfortunately I’m on several deadlines and am leaving early tomorrow on a five-day trip. But I can write a post about it when I get back, fair enough?

  2. Pushing back slightly, I’d say there are STEM meta-skills. Application of the scientific method, some level of statistics, etc. I don’t think that’s really at odds with the classical idea of a liberal arts education, indeed I’m sure many core curriculum’s address that sort of thing.

    My experience of STEM education as actually taught is more in line with the principles you mention than what you rightly skewer the pundits for calling for. The old Univ. of MD (College Park) computer science curriculum ran you through a gauntlet of different program languages. The objective wasn’t to learn Ruby or what have you, it was to learn how to code in a variety of different systems. I suspect we weren’t an outlier in this.

    That said, this actually makes it a costlier meta-skill than just learning a single coding language which is probably a good reason to not consider it anywhere close to a core skill.

    Anyways, C.S. is a strength of Maryland, so maybe we aren’t representative. But do you think STEM as taught in your average state school is more in line with the liberal arts ideal or the brittle suggestion of the pundits?

    1. See, that to me is success– the point is never “don’t learn STEM,” the point is that people in elite media shouldn’t think about STEM learning or skills in a reductive and unhelpful way. I love the sciences and technology, but I love them in part because they aren’t what policy elites seem to think they are. And a society cannot have a healthy labor market if the expectation is that everyone should be invested in the same types of knowledge.

      1. “the same types of knowledge”

        STEM is enormously broad. I bet $100 that all those petroleum engineers don’t have degrees in that particular discipline. But they know how to think like engineers. Speaking for myself, I’m a self-taught programmer who knows how to think about CS. I’ve never coded Python, but I have no doubt I could quickly pick it up and be productively contributing to a project within a week.

        Further, STEM is not the end, it’s the beginning. Many STEM folks go on to get an MBA or start their own companies. It’s almost impossible to start a tech company without having a fairly deep understanding of how the technology works. A few outliers have done it, like Steve Jobs, but they usually have a Steve Wozniak at their side.

        I don’t think we should push people into STEM per se, but this music major sure wishes he’d taken some CS classes between rehearsals.

        1. By “all” I meant “some.” There are lots of trained petroleum engineers on the scene, of course, but I’m positive there are lots who studied something else.

    2. Following-up on what Greg said, I think one of the main advantages to exposing students to programming early on is to teach them the core concepts of computer programming, not the details of implementing some specific algorithm in a specific language. Once they have that foundation, they can build on it in the future.

      Programming has a pretty steep learning curve. Until you understand the basics of what you’re doing, you can’t even get a program that runs. With a bit of help you can get over that barrier and then learning more isn’t very difficult. At the high school level, I think the goal should be to make sure that many students (say, the same number that take trig) learn enough to get over that initial barrier.

      Also, (at least in my experience) programming isn’t very similar to anything else — without having done it, there’s no real way to know if you would like it.

      1. I had a similar experience– I was never a great programmer, but I think the time I spent messing around with Basic on my Commodore 64 taught me as much or more about rigorous thinking as any math class ever did.

        I also spent about ten years teaching science to urban middle school kids, and insofar as my teaching was good and communicated something worth knowing, it was about joy, community, curiosity, and patience, none of which are “skills.”

        The problem education policy will always have is this one:
        Growing up means growing up into a world and a place within that world. For millennia, that process took place very directly, by children imitating the adults of their intended social station. The educational system developed not (merely or primarily) because the modern world depends on literacy and internalized self-control, but because the adult workplace became inhospitable to children. Meanwhile, the work of the household lessened sufficiently that, first, adult women could manage it without assistance from children, and now, it is a tertiary responsibility to work and “childcare,” both of which children themselves cannot participate in.

        Our world is incomparably more humane than the world of the past in many, many ways, but it often doesn’t “make sense” to children or to adults because there is, largely, no “real life” for which school is preparing us for, no social world for us to live within, to join, to grow up into.

        1. To expand on this for a second, for most educated adults (and many less educated adults), the “real world” they experience, the world of relevant social opinion and participation, is online, held within a glowing rectangle we carry around with us or sit down before for hours a day. To be a parent these days is very often the frantic and largely unsuccessful attempt to invent an alternative, contained and safe social world that kids can participate in, outside of that glowing rectangle. This is why school is so very important, entirely apart from whatever is in the Core Curriculum guide– all the kids are there, adults know it’s important and care about it, it’s the “realest” place in both kids and their parents’ lives, the only site of in-person community that many adults interact with on a regular basis. It’s important because it’s the place that children spend their lives, because having a good, safe, fun, interesting, exciting place to spend your life is important, because the intellectual and social life is as much a part of being a person as the physical life, because there’s no conceivable shift in our economy that will make workplaces into dynamic, supportive, comprehensible places to grow up into. But the world of work will increasingly exist (and in many ways already does exist) in order to allow and facilitate the world of school, not the other way around. This maybe has always been true– to paraphrase Paul Goodman, a good society is the society that it is good to grow up within. But it will be even more true in a world after work.

  3. I’ve spent thirty years writing programming/admin books for hi tech companies. What prepared me for this work? Strong writing and teaching skills, ability to work alone and do research, and the study of Latin/Greek. Learning FORTRAN or c after learning Latin is child’s play. How did I get this training? By pursuing a Ph.D. in English.

    Most of the people who built the foundations of “computer science” did not have a cs degree. They were physicists, egyptologists, philosophers, english majors, mathematicians, etc. The field didn’t get “professionalized” until the second generation.

  4. Great post, but how would you respond to these two pushbacks?

    First, is the increasing relevance of STEM majors at all motivated by the effects of a diminished core curriculum or the fact that distribution requirements have become increasingly relaxed? In principle, I have no difficulty imagining a French Poetry major handling a tech heavy job, but the fact on the ground is that many (I don’t have exact numbers) English majors, or many other humanities and social sciences, now skate through college with only the most introductory mathematical instruction available. So that sort of major is going to predict poorly for a job where general mathematical skill is considered at least a strong advantage. This would not be as much of a problem if there were a strong push within the university to guarantee broad education rather than self-guided exploration of the curriculum. The inverse of this is that all the engineering students writing at a high school level or lower would greatly benefit from being forced to take rigorous English classes with a significant composition and textual analysis component.

    The second pushback is that there really is a lack of decent maths education in America, but it is not in the area the STEM proponents think. The fact that a large portion of our college-bound students have taken calculus but will be bamboozled by the most trivial proof is astonishing. Basically, we have failed to teach maths at the foundational level, set theory, mathematical logic, real analysis instead of “calculus”, and groups and fields instead of “algebra.” These are incredibly useful skills for developing broad critical thinking, as writing a proof is a highly technical exercise requiring clear understanding of any possible weakness in an argument. You can imagine how this sort of habit would benefit arguments beyond the mathematical domain also. Interestingly enough, a lot of CS programs do expose you to that sort of thing, as they require a fairly advanced understanding of discrete mathematics.

    I value liberal education greatly, in fact that is why I chose not to attend a technical school. I deliberately majored in philosophy and CS to get as much out of the broad knowledge available in such an environment. But I think we need to recognize the reforms necessary in liberal education today before suggesting to retain it as a model.

    1. What Guarino said. I double majored in CS and Mathematics in the late ’70s at Massachusetts Tool and Die, and really the mathematics was far more valuable to me, in a programming position, than the CS. All the programming I learned by doing (and a couple of algorithms courses). But learning what a real proof is, which I learned from taking lots of courses (complex analysis, topology, various logic courses), was far more valuable, especially when writing concurrent programs.

      Not that you need to know what an compact space is to write a correct program, but you do have to know what a proof is.

  5. What’s happened is that employers have attained too much power in our society. That certainly isn’t to say they should have no power. Obviously employers’ needs are important. But currently they control too much: workers’ leisure time, healthcare, and the content of education from ages 10 through 25. This wouldn’t necessarily be a bad thing if employers could be counted on to always want the “right” things. But collectively employers tend to be extremely short-sighted. App-programming and social media are hot now, so they want the country’s entire youth trained in a set of narrow programming skills now — never mind what technological conditions will be like in 5, 10, or 20 years’ time.

    The whole point of having an education sector which is at least somewhat detached from employers’ short-term needs is that this way the needs of the long term still get their say. Certain skills and abilities transcend these brief technological fluctuations: skills like reading comprehension, articulate writing, critical analysis, foreign language, and an internalized, cognitive understanding of the basics of history, geography, science and math.

  6. Ever notice those saying we all need to be in STEM and/or people are idiots for not doing it are invariably not from STEM themselves and in fact are so innumerate as to be laughable?

  7. Unrelated to STEM, but maybe the same could be said of the humanities? I liked my undergrad education in history, but found that the program too often promoted excessive country or region specific knowledge rather than the historical skills which I think were the real value.

    The undergrad syllabus had dozens of courses on “Germany in the 1700s” or “Modern India,” and in their own ways the Profs would do their best at teaching the core historical skills, but that was usually incidental to the course’s purpose of surveying whatever happened in Country X over Time Y. At best you’d write one or two papers per course, and get feedback from a Prof/TA of varying quality.

    And its a shame, because I think most humanities courses have the potential to impart very valuable skills from research to writing to critical thinking and more. Unfortunately, my program (at least) didn’t exactly make those skills the point of things. The downsides were that it was quite common to get upper year students not understanding the basics of things like a introduction. As well, there’s a general impression that the humanities teach ‘fluffy’ things like Art History and “when will knowledge of things like gender issues in Meiji Japan really ever be of use”? If some humanities programs made more of a point of highlighting, teaching and assessing the skills they fostered, perhaps there would be less of a drumbeat against the field.

    Law programs may be a useful counterexample. I don’t think a typical J.D. would be seen to have much specific expertise in any one field of law, but achieving the J.D. should display that the student has gone through the basics of legal practice and has certain skills, which are then confirmed by various bar exams. Presumably something similar could be done with fields like history, where graduates could be shown to have achieved certain standards of skill, rather than exhaustive knowledge of any particular case.

  8. I can imagine a future where service industry employers cut costs and increase profits by not providing uniforms. Still, uniforms will be required for the job. Articles will be written about how the key to finding a good job is investing in a good uniform. People will speculate about which uniforms are the best bets now for the labor market to come. Studies will show the earnings gap between those with uniforms and those without. We’ll read about good-paying positions sitting there unfilled because none of the applicants have the right uniform.

  9. Actually there is a skills shortage in CS, but it’s a shortage of people who are good. The average programmer is mediocre and doesn’t even realize it. Ironically the same thing has happened to CS courses as has happened to education generally. More and more schools are teaching the ‘hot’ skills, rather than the hard skills (how to reason, styles of programming, decomposition of problems and composition of solutions). That stuff is hard, but employers don’t value it (well your google’s do – but that’s a tiny fraction of the market).

    However the average job (including many of those ‘hot’ lauded jobs that currently pay well, but won’t in 5-10 years) are really just technician jobs. Most programmers aren’t very good at their job, but they don’t really need to be (and the corporate culture is sort of okay with that).

    1. Well, isn’t the fact that processor power, storage and bandwidth are so damn cheap now, along with the error-resiliency of modern operating systems, highly connected to the “shortage of people who are good”? Except for something like a Google or very specialized software, nobody needs to know this stuff so well because the consequences of sloppiness are so much lower. Who needs skill just for skill’s sake? That’s like putting truffle oil on McDonald’s fries: expensive and inefficient.

  10. In the early 90’s, I was studying Computer Science. The upper-level course work was a lot of C coding in the Unix environment: networking, operating systems, compilers, …

    I was a fairly early user of Linux: the kernel I started with was 0.99, patch level 13.

    Boy, how things have changed in 20 years!

    Today I’m … coding in C on Linux firmware.

  11. I think this is an extraordinary compelling analysis. I am 51. I have a PhD in the humanities (history of medicine). I have spent the past 15 years outside of academe — working for a hedge fund, a pharmaceutical company, and a consultancy before becoming self-employed. Although I excelled in both math and science as a teen, I lost interest and studied neither in college or graduate school (excepting statistics which I love). My true love is literature. But … I earn my living working with extraordinarily accomplished scientists and physicians — the overwhelming majority of whom are — according to my admittedly snobbish standards — illiterate and incapable of writing a compelling sentence let alone a full paragraph. It’s not that they’re not brilliant. Many are. It’s just that language and narrative are not their things. That’s where I come in and how I earn my keep. And I can do that thanks to my splendid liberal arts education.

Leave a Comment

Your email address will not be published. Required fields are marked *