Theories of Progress is a series from Education Progress building the intellectual framework for durable education reform: why the system resists improvement, what excellence actually requires, and what history tells us about how to build it.
Introduction
Bryan Caplan’s landmark book The Case Against Education defends the provocative signaling theory of education: the value of education, he claims, is about 20 percent meaningful skill development and about 80 percent signaling, and as such the education system is largely a waste of time and money. With its willingness to leap gleefully onto third rails, the book is a useful contribution to the conversation. But when I read it, I was disappointed.
Our schools ought to be cathedrals of knowledge and expertise, cultivating students’ skills in the best ways we know how. The task is to build them up. So I came in excited to see what Caplan would say about the nuts and bolts of education, the things that actually help students learn. But while he has a great deal to say about the job market and the pointlessness of asking kids to learn history, art, and languages, he exhibits much less interest in what makes courses effective or ineffective at achieving their stated goals.
When he does venture into pedagogy, he presents a superficial and incurious picture of how students actually learn. He enlists expertise research to support on-the-job learning as ideal when the field arose specifically to explore why people so often failed to learn essential skills on the job. His vision of the ideal — online education providing skills to the eager better than brick-and-mortar ever could — is catastrophically mismatched with how even eager students learn well in practice. And he treats education research, a field that has been plagued with sloppy work and ideological fads, as simultaneously inconsequential and solved.
I want humans to understand the world and to build the skill to navigate it well. If you want students to learn, you should teach them directly and at a pace and depth calibrated to their current level. I think it is good and admirable that we collectively pay to allow children to attend school for free. I care about education as a good in itself — in Caplan’s terms, a “merit good” — something worth getting right no matter what the things we learn are designed to prepare us for.
I do not intend to argue directly against all parts of Caplan’s book here. He is strongest when he stays within his core domain of economics, and his points in that domain are worth considering. A case against education that lacks a clear picture of pedagogy, though, is incomplete. Whatever your views on the economics of education, Bryan Caplan is wrong about pedagogy.
The Root of Our Disagreement
My dispute with Caplan goes, in some ways, all the way back to kindergarten.
Midway through the book, Caplan relates that he developed the core sentiments behind it shortly after starting kindergarten, when he realized both that he’d “never use most of the material [his] teachers taught” and that “a bright future was waiting” as long as he went through the motions. From there, it was just a matter of time until he landed on his signaling theory.1
I, on the other hand, was never unhappy with digressions into niche or economically unimportant subjects, nor ever terribly concerned about what sort of professional future my school was preparing me for. I loved the idea of learning, and I jumped gleefully into every academic competition I saw, no matter how trivial, be it math drills, spelling worksheets, art contests, or pursuing first chair among my middle school’s French horns. What I hated was not the drudgery in learning material I would never use, but the feeling that it was all being taught much too slowly, wasting my time with things I already knew. And yet whenever I got out of school, I couldn’t help noticing that no matter how much I loved learning math in the abstract, absent external structure I was much more likely to learn the minutiae of Pokémon chaining than algebra. I loved the idea of school, and I valued the structure. I just wanted it to work.
In some ways, then, I think it’s helpful to model the dispute between us as not just one between our adult selves, marshalling data and research and sophisticated arguments, but one between the kids we were, rooted in values, goals, and instincts. When Caplan got older and discovered libertarianism, economics, and signaling, they offered the answer to the questions that long plagued him. When I grew older, I found expertise research and the debates over the science of learning. He concluded we need less school; I concluded we need better school. Whatever evidence either of us marshals, it is that divergence in goals more than anything else that sets our views apart and explains the book’s central blind spot.
The Deliberate Practice Reversal
Caplan spends Chapter 2 building the case that education is mostly useless: “schools teach few job skills, transfer of learning is mostly wishful thinking, and the effect of education on intelligence is largely hollow.”2 In the latter two areas, I cleanly agree with him: learning is mostly domain-specific and does not meaningfully transfer between subjects, and education does not change the rank-order of students in intelligence or aptitude.
These are important insights, but they are also perfectly compatible with the project of making schools better at cultivating skill and expertise. The research on transfer is driven in large part by a concern for schools to focus on improving how they teach subject-specific skills; similarly, acknowledging that instruction primarily improves specific skills rather than intelligence motivates that same focus. The decisive question for Caplan’s case against education is how people actually learn, and it is here that he fumbles.
Toward the end of the chapter,3 for example, he asks how people get good at their jobs if they learn so few useful job skills in school. To explain, he cites the late K. Anders Ericsson, whom he accurately describes as the world’s leading expert in expertise, and describes Ericsson’s model of deliberate practice:
Novices improve as long as they are, “1) given a task with a well-defined goal, 2) motivated to improve, 3) provided with feedback, and 4) provided with ample opportunities for repetition and gradual refinements of their performance.” Before long, though, the benefit of mere practice plateaus. To really get good at their jobs, people must advance to deliberate practice. They must exit their comfort zone—raise the bar, struggle to surmount it, repeat.4
All of this is true and important, and I was excited to see Caplan raise it. But the way Caplan uses Ericsson would make the late researcher roll in his grave. Caplan describes how the labor market offers “plenty of subpinnacle opportunities [for deliberate practice]” and provides “ample time to get good in most occupations.”5 Per Caplan, people “become skilled workers by devoting years to their chosen vocation — by doing their job and striving to do it better.”
This is an inversion of Ericsson’s work — replicating, in fact, the very misconception that inspired much of Ericsson’s research in the first place. In many professional domains, people do not meaningfully improve given more time on the job. In his seminal 2008 paper on deliberate practice, Ericsson hammers a “virtual absence of a relation between amount of experience and objective performance in many professional domains,” in which people often turn suboptimal practices into rote, automatic habits, and describes how that observation shaped his research on deliberate practice and the necessity of coaching and careful, structured practice — not the regular day-to-day of job duties — to cultivate performance.6
In a 2015 paper more specifically focused on medical work, Ericsson notes that “research has not shown sustained benefits of longer professional experience by health care professionals after completing supervised training” and that benefits beyond the first few years of initial on-the-job experience are limited; he then argues for an overhaul of training processes, lengthening training and focusing on fundamental skill processes much earlier.7 Ericsson’s picture of deliberate practice is one that relies on purpose-built institutions that take the nuts and bolts of talent acquisition seriously rather than treating it as something that just happens on the job.
Ericsson’s work is only the start of expertise research, not the last word, and it may well be that Caplan disagrees with him about the utility of on-the-job experience. But Caplan doesn’t even present it as a debate, suggesting Ericsson’s work supports job experience as obviously superior for building expertise when it does not.
The False Paradise of Online Learning
Next, Caplan’s vision of education-that-works, and his justification for why it should lead to fatalism, is telling. He waxes poetic about the role of the internet in education — full of “not just stream-of-consciousness enlightenment, but outstanding formal coursework,” “free of charge,” a “Merit Machine” that is “swiftly making traditional humanist education policy obsolete.”8 To Caplan, the reality that most people don’t use the internet for intellectual enrichment shows that “eager students are few and far between,” that public education mostly serves to make ideas and culture “mandatory for everyone who isn’t interested,” and that low consumption of ideas and culture stems from apathy. To him, “online education has clear pedagogical advantages over traditional education.”9
This is in line with Caplan’s ideas about revealed preference in general. He takes a hardline stance on the topic, arguing that even mental illness is best understood as “an extreme, socially disapproved preference.” But people are not so simple as to ever have singular, unified preferences, and institutions at their best align our short-term whims with our long-term interests. The internet is a wonderful enabler of our short-term whims, and a mediocre pedagogue that very rarely provides the structure an optimal education demands. That people often waste time online even as they prove able to pursue high-effort goals in structured environments is damaging not to those who call for such environments, but to Caplan’s reliance on revealed preferences. Ericsson’s work and that of his successors in expertise and pedagogy research aim to build institutions that make it so that individual students do not have to cobble together programs in isolation, but can enter frameworks that make it straightforward for them to do things that require herculean willpower on their own.
Benjamin Bloom proposed a three-stage model of expert talent development in every domain: an early exploratory stage where novices play and explore, a long and involved second stage focused on intense and strenuous skill development, and a third stage at the discipline’s frontier where a person is ready to contribute meaningfully to research or pushing the discipline forward.10 In Caplan’s vision of education, stage one is almost all there is. He praises preschool for its approach of “gently introduc[ing] students to a wide range of enrichment opportunities,” being happy if something sticks, and otherwise leaving students to do their own things.11 That misses one of the most important functions of institutions. Self-structured online learning rewards play and exploration and empowers original research, but in the strenuous stage needed for real talent acquisition, people benefit immensely from deliberate structure and real accountability.
When I was younger, I wanted to study Chinese. I started out by self-studying, progressing in leaps and bounds for a few months, before getting distracted and flitting away to other pursuits — as one does. Later, I joined the Air Force and attended Defense Language Institute, where I studied Chinese in a structured environment eight hours a day for some 15 months. Now I speak Chinese. Left to my own devices with just the internet, I would not.
Likewise, I was interested in the abstract in studying law. Amid my intellectual dabbling, I would periodically read one or another article about law, would develop opinions about major Supreme Court cases, and so forth. With just the internet, I would have kept dabbling in and around many other interests. But I attended law school, where I received rigorous, structured instruction on enormous chunks of the law in practice that I never would have bothered with from internet dabbling, courses that were both interesting and meaningful and that I was almost uniformly glad to take.
Caplan’s view leaves little room for the idea that students would prefer intellectual enrichment if given the right support. He might say that while I claimed to be interested in learning Chinese and the law, my usual behavior revealed instead that I really preferred watching speedruns on YouTube or arguing on Twitter. But my actual, in-practice preferences led me into one institution where I studied Chinese and went on to work for several years as a Chinese language analyst, then into another where I studied the law and am now preparing to swear in as an attorney.
I tried online education myself for a year in elementary school, having been frustrated by the slow pace of my courses in my regular school. I rushed through the provided curriculum, spending 1–2 hours a day on coursework to complete two years of material, then sneaking as much time as possible on Neopets while my mom was distracted. When my mom noticed how antisocial I was becoming, how possessive about the computer, she put me back in physical school part-time. Even in a near-optimal setting, in other words, with a kid who was ahead of his class and liked to learn, an online focus meant a lot of burning time and real side effects.
I share these to underscore that this is personal for me. Caplan may be happy for me to chase revealed preference into oblivion, but I am not.12 I wrestle with institutions rather than dismissing them because despite my frustrations, I benefit from their structure in real, important ways. Neither I nor the many students like me would be happier in a world where they were weaker. More important than anecdotes, of course, is the data on online learning and remote work, which lines up with the intuition that removing social structure tends to hurt. Online charters like the one I attended perform terribly on empirical outcomes, with researchers reporting the equivalent of between 70 and 180 days of instruction lost per subject—that is, in a 180-day school year, they were often learning functionally no math at all. During the natural experiment provided by COVID-19, students tended to lose the equivalent of at least a third of a year of schooling relative to expectations, and remote work tended to lead to lower productivity particularly among younger and less experienced employees.
This is true even in the domains best suited to remote, self-structured learning. You can learn chess from countless videos and online courses, you can play it any hour of the day online, and it’s fun to play. But virtually everyone who pushes toward the top does so with a human coach who spends substantial amounts of direct time and effort helping them. Grandmasters train under grandmasters. At every level of learning, the institutional and social structure matters.
Students benefit from structure. I benefit from structure. It is not mercy to leave people to their whims and their own ingenuity, even as a small number of highly capable and motivated people wind up doing extraordinary things when left to their own devices.
Pedagogy as Simultaneously Inconsequential and Solved
Caplan’s most substantive discussion of pedagogy comes at the beginning of Chapter 8, in which he defends the value of vocational education. He explains why he doesn’t focus on pedagogy and rejects the vision of “human capital enthusiasts” who want to “transform our schools from time sinks to skill factories”:
How could we make this happen? Finding better ways to teach students reading, writing, and math is the conventional path. Since an army of researchers and practitioners are already working on this problem, I have little constructive to add. Yet overall, we should be pessimistic about improving basic skills. Why? Because the goal has long been popular, the research has long been ample, yet basic skills remain mediocre. The logical inference is either (a) pinpointing ways to improve basic skills is elusive, or (b) schools spurn the methods that work.13
He goes on to give the example of firing more teachers as something with a solid intellectual case that nobody really expects to prevail.14
I confess, the incuriosity and fatalism here bothers me. The only example he entertains of a method to improve skills is the libertarian hobby-horse of firing teachers, which isn’t an example of an instructional method at all. He hand-waves the entire field, asserting at once that it’s received so much focus that he can’t add anything meaningful, and that nothing can come of it anyway. This from a man writing a book to argue for one of the most politically impossible ideas ever raised in education simply because he believes it’s the right thing! You can’t even get the public to back firing teachers, and you want to tear the whole education system down?
Is the case for “human capital enthusiasts” building “skill factories” so bleak? No. Pedagogy is neither so well-understood nor so hopeless as Caplan would say. Education has never been a mature science, and outsiders working from first principles often make jumps that would be unthinkable in more developed domains. Looking at a zoomed-out view of a complex system, you see a mix of progress and regression. But as soon as you get concrete and narrow, defining the problems you care about with a seriousness about solutions, there are always improvements individual “human capital enthusiasts” can make.
When advertiser Siegfried Engelmann took an interest in pedagogy in the 1960s, he developed a comprehensive instruction model, Direct Instruction, that dramatically outperformed peer curricula in the largest government-funded research program of the 1960s and ‘70s, continued to show sizable effects in subsequent research, and, despite heavy resistance, built out a serious niche, particularly in special education.
When college student Richard Rusczyk noticed student scores were frustratingly low on a mathematics competition he tried to run in 1990, he spent decades developing a high-level mathematics curriculum, Art of Problem Solving, from scratch. Now, nearly all US participants in the International Mathematics Olympiad are Art of Problem Solving alumni.
Mississippi, fed up with its former bottom-tier education performance nationally, implemented a sweeping set of phonics-oriented reforms in the early 2010s, including wide-scale curriculum-specific teacher training and accountability measures. Now it’s one of the best-performing states nationwide on 4th grade reading.
It’s worth dwelling on that last one a bit. Public schools are often the most resistant to change, with policy fights and uneasy compromises between many people who have mutually contradictory goals. They are also the schools of choice for some 80% of American K-12 students, including in states with robust school choice ecosystems. Caplan gives readers sympathetic to the unglamorous work of effective reform an excuse to concede: nothing can be done, it’s all a waste of time and money, go find something else to do. But even someone who shares his absolutist stance that public schools should not exist must acknowledge that there are millions of kids attending them who benefit from positive change. Abdicating that battlefield only leaves kids reliant on those combatants who choose to fight on.
Each of these initiatives responded to real problems, faced serious headwinds, and navigated the education system as it actually exists to improve school in concrete ways. People with specific visions for what they want in schools can build new schools, as with recent projects like Alpha School’s Montessorium, North Carolina’s Telra Institute, and Kelsey Piper’s Oakland LEARN; they can build new curricula or programs filling gaps they see in the system, as with Math Academy, Science is WEIRD, or Mentava; or they can look for levers of influence within existing schools.
None of this is hopeless. Much of it is hard. All of it is worthwhile.
This is what frustrates me, really, when I read Caplan on pedagogy. He painstakingly documents every concrete detail he can find on something like the sheepskin effect, warranting bold, counterintuitive claims with the patient meticulosity that boldness requires. Then, when it comes to everything that concerns how subjects are actually taught, he glazes over the topic, botches the particulars, and treats the entire thing as almost unworthy of notice.
Where Caplan’s View Wins
All is not lost for a collaboration between the schools of thought of Bryan Caplan and Education Progress. What I talk and care about and the positions I take are hardly the default among people who study or think about education. Here’s where I must confess to some envy. Those who want to pursue economics, like Caplan, enjoy a functioning, serious discipline that maintains high standards for research, good schools, and good courses. Indeed, that corner of education is one Caplan distinguishes from the rest, noting that “there’s no magic” when he trains Ph.D. students: they want to do his job, and he shows them how it’s done.15
As I developed my interest in education, I looked for discerning voices in the field. I kept noticing, almost without fail, that their perspectives on schools of education—the institutions that train teachers and lead education research—ranged from something approximating “the root of all evil that must be burned down with merciful fire” (to paraphrase critics like Engelmann and knowledge-rich curriculum pioneer E. D. Hirsch) to “Look, they’ve never been good, per se, but they’re really not powerful enough to be the source of all ills” (to paraphrase education historian David Labaree). Carl Hendrick, one of the leading experts on the science of learning and the co-author of How Learning Happens, describes how much of his work was motivated by the realization that much of the really important information on learning he found, useful technical topics like research on the limitations of working memory, had never really found a home in education schools, which tend to be dominated instead by low-grade fads.
And I have to wonder—how would Caplan feel, if economics departments had the output and reputation of education departments? Would he shrug and say an army of researchers and practitioners are already working on economics research, that there were other productive things to do, and he had little to add? Or would he feel more like the way I feel about education—that there is something fascinating and important there, and that the failure of the discipline to capture it was an indictment not of the topic but of those who neglected and damaged it?
In truth, as Fields medal–winning mathematicians weep because they see artificial intelligence annihilating their field, I have no confidence looking anyone in the eyes and telling them education will continue to provide their children with useful job skills. But I believe deeply that, no matter what the future holds, institutions to push humans toward their potential belong in it. Whatever the job market looks like, when Caplan wants to tear down all public education, I stand against him.
But tearing down education schools, with a determination to build something worthy of the subject in their place?
That’s a project I could get behind.
Caplan, Bryan. The Case against Education: Why the Education System Is a Waste of Time and Money. Princeton: Princeton University Press, 2018. Pp. 217–218.
Ibid., p. 63.
Ibid., pp. 63–64.
Ibid., p. 63.
Ibid.
Ericsson, K. Anders. “Deliberate Practice and Acquisition of Expert Performance: A General Overview.” Academic Emergency Medicine 15, no. 11 (2008): 988–94. Portico. https://doi.org/10.1111/j.1553-2712.2008.00227.x.
Ericsson, K Anders. “Acquisition and Maintenance of Medical Expertise.” Academic Medicine 90, no. 11 (2015): 1471–86. https://doi.org/10.1097/ACM.0000000000000939.
Caplan, op. cit., p. 242.
Ibid., pp. 219–220.
Quoted by Justin Skycak, https://www.justinmath.com/files/the-math-academy-way.pdf, pp. 32-33.
Caplan, op. cit., pp. 283–284
Caplan does not dismiss the idea of “meta-preferences,” as he puts it, entirely. In talking about revealed preferences, he notes: "Almost everyone reading this probably wishes they were less lazy, more patient, more outgoing, more loving, more ambitious, or more persistent. But you still are the preferences you really have." That may well be! But the preferences I really have change based on the institutions I am in, and my meta-preference is to build institutions that shape the preferences I really have toward the preferences I really want.
Caplan, op. cit., p. 225
On the merits, I think firing teachers is a tool well-suited to specific, somewhat narrow circumstances that’s often presented as a panacea. As I detail in my account of Hobson v. Hansen, there are widespread problems with teacher competency in many districts, particularly in poorer urban schools. But the problem is non-trivial. The kids need someone in charge; there is no contingent of well-trained, eager teachers ready to take the place of dismissed ones; and teachers rightly object to impossible standards like the Clark Plan’s approach of “teachers must teach kids who are several years below ‘grade level’ in heterogeneously grouped classes, and if their kids underperform on standardized tests, they will be penalized” or to waves of know-it-all reformers trying to upend longstanding systems in ways they know will be abandoned in a year or two. In general, I see current teachers, like school systems in general, as realities of the system to be worked alongside to the extent possible, not obstacles to defeat.
Caplan, op. cit., p. 13
I think Caplan's arguments are strongest when it comes to university education. There's still something of a cult around university education, where there's an idea that if we have the right subsidies and culture, everyone can go to university and be upper middle class. But for a lot of fields, e.g. your field of education, a Bachelor's really isn't providing much in meaningful skill training and is instead a costly waste of time. It's just signaling.
I think it is possible to greatly improve university education for a lot of fields, but in the meantime we should greatly reduce subsidies so money isn't wasted on a zero sum status race.
For elementary and highschool, the focus should just be improving education, and your ideas for improving it are wonderful.
Nicely argued. Alas, Caplan's green visor is vastly more compelling than your rose-colored glasses.