Quantitative science is (indirectly) relevant to the real world, also some comments on a book called The Case Against Education

Joe Campbell points to this post by economist Bryan Caplan, who writes:

The most painful part of writing The Case Against Education was calculating the return to education. I spent fifteen months working on the spreadsheets. I came up with the baseline case, did scores of “variations on a theme,” noticed a small mistake or blind alley, then started over. . . . About half a dozen friends gave up whole days of their lives to sit next to me while I gave them a guided tour of the reasoning behind my number-crunching. . . . When the book finally came out, I published final versions of all the spreadsheets . . .

Now guess what? Since the 2018 publication of The Case Against Education, precisely zero people have emailed me about those spreadsheets. . . . Don’t get me wrong; The Case Against Education drew plenty of criticism. Almost none of it, however, was quantitative. . . .

It’s hard to avoid a disheartening conclusion: Quantitative social science is barely relevant in the real world – and almost every social scientist covertly agrees. The complex math that researchers use is disposable. You deploy it to get a publication, then move on with your career. When it comes time to give policy advice, the math is AWOL. If you’re lucky, researchers default to common sense. Otherwise, they go with their ideology and status-quo bias, using the latest prestigious papers as fig leaves.

Regarding the specifics, I suspect that commenter Andrew (no relation) has a point when he responded:

You didn’t waste your time. If you had made your arguments without the spreadsheets—just guesstimating & eyeballing, you would’ve gotten quantitative criticism. A man who successfully deters burglars didn’t waste his money on a security system just because it never got used.

But then there’s the general question about quantitative social science. I actually wrote a post on this topic last year, The social sciences are useless. So why do we study them? Here’s a good reason. Here was my summary:

The utilitarian motivation for the natural sciences is that can make us healthier, happier, and more comfortable. The utilitarian motivation for the social sciences is they can protect us from bad social-science reasoning. It’s a lesser thing, but that’s what we’ve got, and it’s not nothing.

That post stirred some people up, as it sounded like I was making some techbro-type argument that society didn’t matter. But I wasn’t saying that society was useless, I was saying that social science was useless, at least relative to the natural sciences. Some social science research is really cool, but it’s nothing compared to natural-science breakthroughs such as transistors, plastics, vaccines, etc.

Anyway, my point is that quantitative social science has value in that it can displace empty default social science arguments. Caplan is disappointed that people didn’t engage with his spreadsheets, but I think that’s partly because he was presenting his ideas in book form. My colleagues and I had a similar experience with our Red State Blue State book a few years ago: our general point got out there, but people didn’t seem to really engage with the details. We had lots of quantitative analyses in there, but it was a book, so people weren’t expecting to engage in that way. Frustrating, but it would be a mistake to generalize from that experience to all of social science. If you want people to engage with your spreadsheets, I think you’re better off publishing an article rather than a book.

As a separate matter, Caplan’s “move on with your career” statement is all too true, but that’s a separate issue. Biology, physics, electrical engineering, etc., are all undeniably useful, but researchers in these fields also move on with their careers, etc. That just is telling us that research doesn’t have 100% efficiency, which is a product of the decentralized system that we have. It’s not like WW2 where the government was assigning people to projects.

Comments on The Case Against Education

This discussion reminded me that six years ago Caplan sent me a draft of his book, and I sent him comments. I might as well share them here:

1. Your intro is fine, it’s good to tell the reader where you’re coming from. But . . . the way it’s framed, it looks a bit like the “professors are pampered” attack on higher education. I don’t think this is the tack you want to be taking, for two reasons: First, most teaching jobs are not like yours: most teaching jobs are at the elementary or secondary level, and even at the college level, much of the teaching is done by adjuncts. So, while your presentation of _your_ experience is valid, it’s misleading if it is taken as a description of the education system in general. Second—and I know you’re aware of this too—if education were useful, there’d be no good reason to complain that some of its practitioners have good working conditions. Again, this does not affect your main argument but I think you want to avoid sounding like the proudly-overpaid guy discussed here: http://andrewgelman.com/2011/06/01/the_cushy_life/

This comes up again in your next chapter where you say you have very few skills and that “The stereotype of the head-in-the-clouds Ivory Tower professor is funny because it’s true.” Ummm, I don’t know about that. The stereotype of the head-in-the-clouds Ivory Tower professor is not so true, in the statistical sense. The better stereotype might be that adjunct working 5 jobs.

2. You write, “Junior high and high schools add higher mathematics, classic literature, and foreign languages – vital for a handful of budding scientists, authors, and translators, irrelevant for everyone else.” This seems pretty extreme. One point of teaching math—even the “higher mathematics” that is taught in high school—is to give people the opportunity to find out that they are “budding scientists” or even budding accountants. As to “authors,” millions of people are authors: you’ve heard of blogs, right? It can useful to understand how sentences, paragraphs, and chapters are put together, even if you’re not planning to be Joyce Carol Oates. As to foreign languages: millions of people speak multiple languages, it’s a way of understand the world that I think is very valuable. If _you_ want to say that you’re happy only speaking one language, or that many other people are happy speaking just one language, that’s fine—but I think it’s a real plus to give kids the opportunity to learn to speak and read in other languages. Now, at this point you might argue that most education in math, literature, and foreign language is crappy—that’s a case you can make, but I think you’re way overdoing it my minimizing the value of these subjects.

3. Regarding signaling: Suppose I take a biology course at a good college and get an A, but I don’t go into biology. Still, the A contributes to my GPA and to my graduation from the good college, which is a job signal. You might count this as part of the one-third signaling. But that would be a mistake! You’re engaging in retrospective reasoning. Even if I never use that biology in my life, I didn’t know that when took the course. Taking that bio course was an investment. I invest the time and effort to learn some biology in order to decide whether to do more of it. And even if I don’t become a biologist I might end up working in some area that uses biology. I won’t know ahead of time. This is not a new idea, it’s the general principle of a “well-rounded education,” which is popular in the U.S. (maybe not so much in Europe, where their post-secondary education is more focused on a student’s major.) Also relevant on this “signaling” point is this comment: http://andrewgelman.com/2011/02/17/credentialism_a/#comment-58035

4. Also, signaling is complicated and even non-monotonic! Consider this example (which I wrote up here: http://andrewgelman.com/2011/02/17/credentialism_a/):
“My senior year I applied to some grad schools (in physics and in statistics) and to some jobs. I got into all the grad schools and got zero job interviews. Not just zero jobs. Zero interviews. And these were not at McKinsey, Goldman Sachs, etc. (none of which I’d heard of). They were places like TRW, etc. The kind of places that were interviewing MIT physics grads (which is how I thought of applying for these jobs in the first place). And after all, what could a company like that do with a kid with perfect physics grades from MIT? Probably not enough of a conformist, eh?”
This is not to say your signaling story is wrong, just that I think it’s much more complicated than you’re portraying.

5. This is a minor point, but you write, “If the point of education is certifying the quality of labor, society would be better off if we all got less.” This is not so clear. From psychometric principles, more information will allow better discrimination. It’s naive to think of all students as being ranked on a single dimension so that employers just need to pick from the “top third.” There are many dimensions of abilities and it could take a lot of courses at different levels to make the necessary distinctions. Again, this isn’t central to your argument but you just have to be careful here because you’re saying something that’s not quite correct, statistically speaking.

6. You write, “Consider the typical high school curriculum. English is the international language of business, but American high school students spend years studying Spanish, or even French. During English, many students spend more time deciphering the archaic language of Shakespeare than practicing business writing. Few jobs require knowledge of higher mathematics, but over 80% of high school grads suffer through geometry.” I think all these topics could be taught better, but my real issue here is that this argument contradicts what you said back on page 6, that you were not going to just “complain we aren’t spending our money in the right way.”

To put it another way:

7. You write, “The Ivory Tower ignores the real world.” I think you need to define your terms. Is “Ivory Tower” all of education? All college education? All education at certain departments at certain colleges? Lots of teachers of economics are engaged with the real world, no? Actually it’s not so clear to me what you mean by the real world. I guess it does not include the world of teaching and learning. So what parts of the economy do count as real? I’m not saying you can’t make a point here, but I think you need to define your terms in some way to keep your statements from being meaningless!

And a couple things you didn’t talk about in your book, but I think you should:

– Side effects of Big Education: Big Ed provides jobs for a bunch of politically left-wing profs and grad students, it also gives them influence. For example, Paul Krugman without the existence of a big economics educational establishment would, presumably, not be as influential as the actual Paul Krugman. One could say the same thing about, say, Greg Mankiw, but the point is that academia as a whole, and prestigious academia in particular, contains lots more liberal Krugmans than conservative Mankiws. Setting aside one’s personal political preferences, one might consider this side effect of Big Ed to be bad (in that it biases the political system) or good (in that it provides a counterweight to the unavoidable conservative biases of Big Business) or neutral. Another side effect of Big Ed is powerful teachers unions. Which, once again, could be considered a plus, a minus, or neutral, depending on your political perspective. Yet another side effect of Big Ed is that it funds various things associated with schools, such as high school sports (they’re a big deal in Texas, or so I’ve heard!), college sports, and research in areas ranging from Shakespeare to statistics. Again, one can think of these extracurricular activities as a net benefit, a net cost, or a washout.

In any case, I think much of the debate over the value of education and the structuring of education is driven by attitudes toward its side effects. This is not something you discuss in your book but I think it’s worth mentioning. Where you stand on the side effects can well affect your attitude toward the efficacy of the education establishment. There’s a political dimension here. You’re a forthright guy and I think your book will be strengthened if you openly acknowledge the political dimension rather than leaving it implicit.

92 thoughts on “Quantitative science is (indirectly) relevant to the real world, also some comments on a book called The Case Against Education

    • Yeah, I was always a bit amused by the spreadsheet point, and the challenge from critics to engage with them. Like, if it personally takes you whole days to give a guided tour of these spreadsheets to each (presumably sympathetic and having some baseline of familiarity) friend, and they represent 15 months of dedicated effort… my suspicion is you’re not operating at the cutting edge of reproducible, open social science, with clear-cut and well documented workflows. And that folks avoid engaging the spreadsheets not because they’re afraid of quantitative rigor or the analysis performed is so ironclad, but that they’re 15 months worth of spreadsheets lol

  1. It’s your blog to run, Andrew, but I came for the “Quantitative science is (indirectly) relevant to the real world” and stayed for your incredibly interesting “some comments on a book called The Case Against Education”. Maybe a few courses in journalism would have taught you to write better headlines!

    I kid, but will defend Caplan (who does not need my defense) by praising his willingness to talk to just about anyone about stuff. My read on his work is that while he communicates honestly what he believes, he does have an incentive to skew “hardline” when framing his arguments, if only to generate buzz / discussion. I’m not always sure he resists the temptation to argue more than he wants to (deep down), but he is always there to do the arguing!

  2. Quantitative social science is barely relevant in the real world – and almost every social scientist covertly agrees. The complex math that researchers use is disposable. You deploy it to get a publication, then move on with your career. When it comes time to give policy advice, the math is AWOL. If you’re lucky, researchers default to common sense. Otherwise, they go with their ideology and status-quo bias, using the latest prestigious papers as fig leaves.

    This is just a transparently untenable position to take in 2022, after 2 years of debate about leading and lagging indicators, test positivity rates, infection fatality rates vs case fatality rates. Schools were literally opened and closed based on a numerical threshold. I’m not saying all the quantitative policy was good or done without political considerations, some was bad and some were enormous successes. But even before covid, I guess bus routes are determined by throwing darts at a map, and fund allocation is completely independent of population, and the 4th most valuable company in the world by market cap doesn’t make 90% of its money on targeted advertising and generalized vickrey auction theory. I suspect what he really means is really “when the correctness of a quantitative strategy is ambiguous, or when an issue is politicized, quantitative analysis is rarely decisive,” which is true, but a lot less fun to say for a Bryan Caplan type.

    It’s just epically solipsistic to state “nobody engaged with the spreadsheets in my book; therefore, numbers don’t matter to anyone in the world.” Takes some real cajones to post a statement like that on the internet.

    I’m very sympathetic to the case against a college education for everyone, and do believe the race for credentialism is reducing our total social welfare, but maybe Caplan’s hyperbolic contrarianism and revelry in the resultant controversy is part of why people aren’t taking him all that seriously.

  3. I believe I made this comment on the earlier post. I routinely engage with people’s spreadsheets when available. But when I looked at Caplan’s, they were primarily spreadsheets he constructed derived from data rather than the raw data itself (perhaps I just didn’t look hard enough). I am immensely interested in raw data, but not so much in the manipulated data, particularly when manipulated by someone with a political agenda.

  4. The utilitarian motivation for the social sciences is they can protect us from bad social-science reasoning. It’s a lesser thing, but that’s what we’ve got, and it’s not nothing.

    If you take a step back, you’ll see that marketing and propaganda is now one of the most powerful industries across the world. That is the use-case for social science, and it has been very effective.

    • ” marketing and propaganda is now one of the most powerful industries across the world.”

      I am not sure what makes something an “industry,” even less the “most powerful” one, but marketing and propaganda have been powerful for quite some time, and social science has not been very effective antidote. “propaganda” is a useless term, it just means whatever the person using it does not like.

      • I don’t think he’s saying social science is an antidote… He’s saying It’s the dote itself, in essence the thing that social sciences like psychology and economics have given us is incredibly well educated highly effective scammers.

      • Marketing is influencing others to buy your product. Propaganda is influencing others to support some political position. It is pretty much a subset of marketing.

        I’m not sure why social science would be expected to find an antidote for its greatest success.

  5. I should probably keep my comments to myself because I haven’t read the book, but I particularly liked the comment asking what, exactly, is meant by “the real world”? Is it that place where Responsible People debate and “study” infrastructure and housing options on a geological time scale? The place where Genius CEOs boldly decree that every department must cut back by some arbitrary percentage? The place where armies of lawyers and consultants spend their careers arguing opposite sides of perennial policy issues? What does it mean to make a contribution in that place?

    In the exceedingly unlikely event that I wandered into those spreadsheets, what kind of analytical framework would I be entering? How is value defined? Wait, let me guess: I think the phrase “return to education” is pointing to lifetime earnings as the value metric!? If that’s it, then I suppose a spreadsheet would help me improve my understanding of which kinds of people “benefit” more or less from education at one level or another (and it would also help me understand how well-supported these conclusions really are). I do think it’s helpful to know this stuff a bit better, and it’s certainly valuable to help disassemble the thoughtless (and often harmful) truism that ever-more education is the silver bullet for everything, so I appreciate Caplan’s contribution to the public discourse (Bonus: he seems to have stumbled on a major value of social science research after all!).

    I guess my first paragraph was just my struggle with issues that simply aren’t what this book is about. Rereading the original post now, I think a lot of Andrew’s comments can be read as, “be more clear about the contribution this book is trying to make, and then try to stick to that path”. Here, here!

    • Josh:

      I think Caplan’s fundamental principle is that he’s politically conservative and educators are mostly politically liberal, so he’s receptive to arguments that disparage education. This doesn’t make his arguments wrong; I just think it helps explain where he’s coming from, and, in particular, why he seems to only look at arguments from one direction here.

      Along with this is, as you discuss near the end of your second paragraph, there are lots of bad arguments out there in favor of education, so if Caplan is coming into this with an anti-education perspective, he can make some real progress, at least for awhile, shooting down bad pro-education arguments.

      The trouble arises when the arguments swallow the big picture. For example, referring to education as “signaling” is useful in the same way that it can be useful to refer to marriage as a “business deal” or to refer to drug addiction as “rational”: in all these cases, making an extreme counter-intuitive claim has shock value and can shake some of the dust off our traditional assumptions. (If marriage really is just a business deal, then what does it mean when people “love” their spouses?, etc.) But the existence of these contrary viewpoints does not, or should not, make us forget the value in traditional perspectives (education is teaching people things, marriage is about building a loving family, drug addiction can ruin people’s lives).

      Perhaps Caplan is taking a division-of-labor perspective here, in which the big picture is that society will make decisions on education, and his role is to provide an advocate’s view of one position. From that perspective, he would want to just present arguments in favor of his anti-education take, with the understanding that people on the other side will present their view, and the rest of us can integrate all this information and decide. That’s one of the perils of advocacy: you have to decide how much you want to integrate all the information yourself and present a full position, and how much you want to play the role of partisan in a public debate.

      • “For example, referring to education as “signaling” ….making an extreme counter-intuitive claim has shock value…”

        As far as I can tell Caplan’s claim that education is “signaling” is the empirical conclusion – that is, that’s how it **actually** works out. It’s not a claim for shock value. It’s the **actual** net of what education produces.

        I’d suggest it works out that way because, in fact, most people recall only a small proportion of what they’re taught, and effectively apply far less than they actually learn. The net effect is that they retain very little *actual education* and most of what they get is a degree on their transcript.

        This could be established with certainty pretty easily: test people about their discipline five years after they complete their degrees. Or just look at GRE subject and general scores.

        Andrew, part of the reason you don’t see it this way may be because you work in a discipline that selects for both the most intelligent and most motivated people and students. But if you worked regularly with graduates from across the academic spectrum, you’d be pretty hard pressed to claim that people take a lot of “education” away from their undergraduate degrees.

        • Chipmunk:

          Regarding your point, “most people recall only a small proportion of what they’re taught, and effectively apply far less than they actually learn,” see point #3 in my above post. Even if a particular skill is not used later on, that does not mean it was prospectively a bad idea to learn it. You could think of a college education in which you study 10 different subjects as analogous to planting 10 seeds and you’re happy if just one or two grows up and blooms.

          But, yes, I take your point in your last paragraph. My students are not typical, first because they’re studying statistics, which is a pretty clearly-defined subject, and second because they’re selected in a competitive admissions process, and third because they have a lot of choice in where to work after they graduate. So I should be careful about generalizing from my experiences (just as I said in point 1 of my above post that Caplan should be careful about generalizing from his own experiences).

        • This is standard sophomoric fare. I was waiting for the “Bill Gates did not go to college” punchline.

          Who claims that “people take a lot of ‘education’ away from their undergraduate degrees.”

          What would testing people “about their discipline five years after they complete their degrees” even look like?

        • Andrew said: “Even if a particular skill is not used later on, that does not mean it was prospectively a bad idea to learn it. ”

          The problem is that most people don’t learn it. There is no seed. Or on a population-wide basis, the number of seeds planted is very small compared to the number of education consumers. That’s what I think Caplan claims to observe. Personally, if I thought your seed example actually reflected the real world, it would be a great point. Yes, education should teach you to generalize beyond the specific application that was used to introduce some set of principles. But mostly it doesn’t happen. That’s my observation. People don’t learn even the specific application, much less the general principles.

          I’m glad you agree about the bias, on top of all the other things you’re at the number two school in the US News ranking! :) I’d also say the more technical the education – physics, chem, math, stats, bio, geo, engineering etc – the more difficult it is for people to fake their way through without learning anything, so that’s another bias you face: just in the subject matter you teach, people who can’t learn the material are weeded out and disappear.

        • GB:

          “Who claims that “people take a lot of ‘education’ away from their undergraduate degrees.””

          Well it would be pretty stupid to have people spend tens of thousands of dollars and take very little education away, now wouldn’t it?

      • Chipmunk –

        > It’s not a claim for shock value. It’s the **actual** net of what education produces. It’s not a claim for shock value. It’s the **actual** net of what education produces.

        That’s a very limited concept of what education “produces.”

        I’m very critical of our standard educational paradigm, but ideally education “produces” students who know how to learn. Looking at merely what subjects students studied
        and whether they remember or regularly utilize discrete facts is only metric if educational outcomes.

        In general, I think our educational institutions could do a much better job than they do of helping students to become better learners. Unfortunately, a relatively high percentage of students leave their educational experiences with a view that leans, more than I think they should, towards viewing education as learning to do what they’re told, or viewing it as basically a passive process of jumping through hoops to get credentials.

        I’m very critical of our schooling in the tendency to be a social sorting mechanism of perpetuating the status quo class structure.

        But the world isn’t binary and sub-optimal does not mean worthless. There’s a reason why societies, in the whole, prosper in parallel with the levels that they educate their populace. There’s a reason why metrics like amount of education track with reducing poverty, reducing hungry children, reducing suffering from ill-health.

        The idea that education means nothing other than “signaling” is a cute political slogan but it doesn’t stand up to even a modicum of scrutiny.

        While I wouldn’t argue that “signaling” is irrelevant to educational outcomes, the idea that signaling is “the net” of what education produces is simplistic and vapid. Of course it’s a claim for shock value. It’s surprising to me to see Andrew seems to think your simplistic political reasoning has merit.

        • I should add…while I say this: “There’s a reason why societies, in the whole, prosper in parallel with the levels that they educate their populace. There’s a reason why metrics like amount of education track with reducing poverty, reducing hungry children, reducing suffering from ill-health.”

          I don’t mean to imply that the correlation necessarily means a causal relationship, let alone a causality that runs in one direction only; I”m sure that societies that are more prosperous for a number of reasons tend to offer more education, and that the causal mechanism is multifactorial. But I think there’s plenty of evidence to show that the argument that the “net product” of education (whether in general or in our society in particular) is nothing other than signaling, can just be dismissed and merely, well, political signaling. As if there’s only one way to measure the “net,” and that signaling is it.

          Oy. Methinks that comment was pretty much on the level of where I explained how vapid a previous comment of yours was:

          https://statmodeling.stat.columbia.edu/2022/09/23/hey-whats-up-with-that-method-from-1998-that-was-going-to-cure-cancer-in-2-years/#comment-2085192

        • Joshua – I’m with you on this one. I have many criticisms of higher education (and lower) – more severe concerns than most people have. But I believe education is more than signaling, and needs to be so much more. My biggest complaint of late is that our whole education system is geared towards people providing answers to questions. We all speak “critical thinking” but when we evaluate students it is on their answers. What a largely obsolete skill! Asking good questions is far more important, and in my mind, much closer to true “critical thinking.”

          I can agree with Caplan (and Chipmunk) that (too) much of education is signaling. But they leap from that to wanting to reform education so it is more oriented to careers and “practical” skills. This often takes the form of recommending that we abandon non-STEM majors, reduce required courses (which I actually agree should be reduced, if not eliminated), and a focus on retained knowledge. This sells education short in my mind. I do like the idea of measuring the impact of education 5 years after they take courses – but I wouldn’t focus on whether or not they retained the knowledge from those courses. I want to know if the way the view the world has changed, whether a quest of learning and critical thinking has been enhanced, and whether they have gained an appreciation for alternative views, ability to work effectively with others, and developed a sense of pride in their own work. These are not easily measured and often conflict with the calls for education to be more career focused and “practical.”

          So, I see our education system as fundamentally flawed, agree with some of the criticisms, but I reach quite different conclusions about the required changes. Signaling is a symptom of the problems but doesn’t really point to the direction of the things that need to change.

        • “Ideally education “produces” students who know how to learn. ”

          Excellent! We agree very strongly on that point! That’s good to see.

          However if students achieve the ultimate goal of learning how to learn, then effectively applying their knowledge to a common application should be a slam dunk.

          “The idea that education means nothing other than “signaling” is a cute political slogan but it doesn’t stand up to even a modicum of scrutiny.”

          First, he does not claim that education means ‘nothing other than “signaling” ‘ He claims that *much* of education is just signaling.

          Second, the “signaling” claim does stand up to at least a modicum of scrutiny. Part of Caplan’s evidence for that claim is that jobs that years ago required only a HS diploma or not even that frequently require a BA today. Why? Because there are tons of excess BAs to be had. There’s no need for the BA in the job. The BA “signals” something about the job seeker.

        • Dale said: “But they leap from that to wanting to reform education so it is more oriented to careers and “practical” skills. ”

          It’s great to have an education for the sake of an education. However, and education is quite expensive, so there has to be some payoff for investing in it. I mean if you can show in some way that an enhanced quest for learning and gaining an appreciation for alternative views in some way generates the funding that the next generation needs to obtain an education, I’m all for it. But I doubt you can show that. My bet is those things have to be bonus points that ride along with the more productive aspects of education. If you focused education on those things alone, the system would collapse, because it wouldn’t generate enough funding to maintain itself.

        • Dale –

          I do agree that these are complicated issues.

          For example, I think a greater openness to focusing education on practical skills could be valuable, as we look at individual students to assess their goals and their strengths.

          By way of example: it used to drive me nuts when I’d be working with a student who was struggling to stay afloat, and I’d ask their teacher why they had to master a particular piece of material at a particular time, and the teacher would answer on the order of “Because that’s what they’ll need to know when they’re at the next level of their educational path.” I mean of course there’s some logic to that, but unfortunately for many students it basically just ensured failure.

          Sometimes what was most important was that the particular student needed to see the meaning behind what they were learning. Their material needed to be relevant to them and their lives. And the goal for me when working with that student, was for them to learn how to identify what was important for them to learn, and how to go about realizing that objective. The idea that there’s a fixed curricular scope and sequence that basically all students would need to follow, in a particular order, does many students a disservice.

          There’s absolutely nothing wrong with engaging these issues and looking at our educational systems with a critical eye. But we can get a lot further when people aren’t trying to leverage the interrogation to advance a particular political agenda.

        • Chipmunk –

          > However if students achieve the ultimate goal of learning how to learn, then effectively applying their knowledge to a common application should be a slam dunk.

          It’s not a binary. Students grow as learners. I think that overall we could do a *much* better job of helping students to grow as learners. But that doesn’t mean it doesn’t happen at all.

          > First, he does not claim that education means ‘nothing other than “signaling” ‘

          Here’s what *YOU* said:


          It’s the **actual** net of what education produces.

          If you’re walking that back, that’s great. Then there’s something to talk about.

          > The BA “signals” something about the job seeker.

          I think our educational system, on the whole, would be much better off if it were organized around criteria-referencing metrics than the currently predominanting norm-referencing metrics. But a degree has some relationship to a criteria-referencing framework. Reducing it to “signaling” is intended to imply that it’s useless or meaningless – like “virtue signaling.” Again, I think at some level that’s a valid criticism. But shock value polemics is an inefficient path to a productive exchange.

      • “Caplan’s fundamental principle is that he’s politically conservative and educators are mostly politically liberal, so he’s receptive to arguments that disparage education”
        Not to (very common) ones that academic leftists are brainwashing their students. That would require them to have at least some effect!

      • Counter-intuitive?

        I recall, long ago, a friend recounting the statement of a dean to him and his fellow first-year students in a well-regarded ivy-league graduate program. That statement was something like the following. “Our process here is like the Department of Agriculture’s food inspection process—We stamp PRIME on prime meat. We don’t raise the beef.”

        The perception that signaling is a big benefit of education has been held by many for a long time.

        Bob76

  6. I agree with every word in this passage, but I’m not so sure it’s a bug in the system:

    “The complex math that researchers use is disposable…. When it comes time to give policy advice, the math is AWOL. If you’re lucky, researchers default to common sense. Otherwise, they go with their ideology and status-quo bias, using the latest prestigious papers as fig leaves.”

    Policy makers (and scientists, for that matter) are *supposed* to check surprising findings against common sense. Common sense provides valuable protection against wasting time taking nonsense research results too seriously. If the results are real, hopefully they’ll move in to the mainstream (after replications, further academic- and policy debate, and so on), which can be another way of saying they become absorbed into ideology.

  7. To your point 3, Andrew, Professor Sidney Burrus was an advocate of a broad education. “He told of counseling a electrical engineering student on what courses to take, and this student didn’t want to take quantum mechanics because “I’ll never use quantum mechanics.” His answer was simple and straightforward: “The only way to know you won’t use quantum mechanics is if you don’t know quantum mechanics.”” (https://ricehistorycorner.com/2021/04/07/to-sidney-burrus-teacher-friend-colleague/)

    • Bill,

      I’m glad I learned quantum mechanics, because, as we used to say in college, it’s “culture”: just something important that everyone should learn if they have the opportunity to do so. Also I needed to take 2 semesters of quantum mechanics to graduate with my degree in physics! And when I was studying the subject, I was expecting to do further work in the area, as my interest within physics was solid-state physics, which is pretty much entirely quantum mechanics.

      But I have only very rarely used my quantum mechanics knowledge in the decades since, and I haven’t at all used the problem-solving skills I learned in my quantum mechanics classes: solving Schroedinger’s equation and all of that. That might be a sign that my quantum mechanics classes were mostly a waste of time. But I could also be easily convinced of the opposite position, that my real mistake was not learning these methods better. I say this because I do lots of work on statistical computing, exploring probability distributions through Hamiltonian Monte Carlo, etc., and I suspect that a better understanding of quantum mechanics—along with my existing statistical intuitions and experiences—could make me a much better researcher in this area. As it is, I can try to collaborate with people who know more of this math and physics, and this is possible, but there’s no substitute for having all the understanding within oneself.

  8. Let’s play along for a bit (with the large caveat that I have not read this book). What does the world without education look like? People will still have babies; those babies will still be born not knowing much (pace Andrew’s sister, perhaps!). Is the plan that nobody teaches them anything, and civilization dries up? That their parents have sole responsibility for teaching them things, i.e. there is no joint social project of maintaining the bones of an agreed-upon culture? Or that we do still teach them things collectively, maintaining some kind of culture, but those things differ from what is taught now? Does Caplan lay out his preferred future in this book?

    I have definitely wondered whether asking individuals to spend hundreds of thousands of dollars collecting a fairly vague but non-optional career credential might have been oversold as a path to social mobility, but it sounds like Caplan’s argument might go in a different and rather more radical direction.

    Agreed with the crowd here on “spreadsheets.” If he’s looking for critical engagement from numbers people, he might get farther with Github.

    • It’s been a few years since I read Caplan’s book, but he’s not against education per se — he speaks highly of education as being beneficial to the individual, both practically and in terms of intellectual enrichment. He is, however, opposed to the *institution* of (forced, or almost forced) education.

      I liked Caplan’s book a lot, despite there being parts I disagreed strongly with. I felt that his case against K-12 education is weak; the case against higher education, at least as it currently exists, is far stronger. The basic idea that most (not all!) of higher education is signaling, once you start to think about it, is hard to dismiss.

      • I dunno, I think I started wondering about the signaling function of higher education about fifteen years ago, in a fit of cynicism that came with the dawning realization of how little faculty are meant to care about teaching undergrads; and then I actually did sort of dismiss the idea mentally after a time because I felt like it wasn’t leading me anywhere interesting. Now that I see it spun out into K-12, which strikes me as completely insane, I’m thinking I must have missed some poison in this idea previously, and I’m glad I put it down.

        • I’m actually working on some analysis of standardized test scores in CA at the moment, for a series of videos I’m going to do on using Julia in data analysis.

          One thing that’s interesting is that looking at the local school district, a lot of kids never get much beyond about 6th grade competency in math. It’s not like they stop taking math class, it’s just that for a lot of students they get the same statewide test score in 11th grade as they got in 6th grade.

          This suggests that for the most part, for a lot of people, education in math beyond 6th grade may have very marginal benefit.

          To me, that’s pretty depressing. My kids are in 6th and 7th grade. Both of them test in the mid 2600’s on the CA state math test. That’s above average for 11th grade in any of the high schools in the district which all test around 2500-2550 or less.

          I consider my kids to be pretty quick at math but very very under-educated for an 11th grader. For example neither of them knows how to solve a quadratic yet, neither of them knows anything about trig functions, or anything much about calculus or probability even basic dice type questions.

          So, let’s face it. I’m all for what we’re doing up to about 8th grade, kids increase their math scores continuously through somewhere in the range 6th to 8th grade. But after that, we’re not obviously serving the purpose of increasing numeracy and mathematical literacy and such.

          I’m with Caplan in that I think **actual education** is important… People should know a bunch more math and a bunch more science and a bunch more health topics, and a bunch more history and geography and such than they do.

          But I’m against what we **call** education, because I don’t think it **is** education in large part.

          Seriously, even among the people here who are incredibly well educated compared to average, how many know the basic concept of how an mRNA vaccine works, or know anything much about how an Air Conditioner works?

        • Daniel asks:

          “how many know the basic concept of how an mRNA vaccine works, or know anything much about how an Air Conditioner works?”

          I’d guess that by now everyone here knows that mRNA vaccines hijack the cell protein production machinery to make virus proteins to prime the immune system, and I’d guess that most everyone here knows that air conditioners are heat pumps that steal heat from the cold side and dump that heat (plus the energy lost due to inefficiency) on the hot side, usually by using an evaporating liquid to steal said heat.

          Now one thing I’d bet most people here can’t do is describe how phase diagrams change with pressure…

          One thing that hasn’t been mentioned is that one thing that college does (or should do) is expose the students to stuff they wouldn’t have seen otherwise. Freshman chemistry had a Materials Science option, and although I had learned to weld at the age of 8 or so, that’s a whole intellectual space that most people haven’t seen.

          Although it doesn’t always work as expected: my father once said “The best thing you learned at MIT was juggling.”

        • David: you know those things, but seriously how many grad students in Econ reading this blog because they want to know more about quantitative methods in social sciences know those things? And how many can take it just a little farther, like what is RNA ? and why use it to “hijack” the cell?

          And how does an AC pump heat with a fluid? Like, what happens to that fluid inside the machine?

          I learned to stick weld when I was about 22 or so. How many people can even describe what happens with current flow in a US house with 110 vs 220 volt outlets?

          How about how sound was recorded onto film?

        • I’m confused. Is your position that more people will understand mRNA vaccines and HVAC systems if we end mass education? Or is it simply that it bothers you that someone can lay claim to various educational credentials and either not understand or have forgotten these things?

        • It bothers me that you can get through 12 th grade and not really know what RNA is or how heat flows in an AC (or any huge number of similar things, like maybe what was the Battle of the Bulge or what country occupied Vietnam before the US entered the war there, or what does the federal reserve system do or what is a torque or why is water called a polar solvents etc etc?)

        • Mainly in response to Erin Jonaitis, “Is your position that more people will understand mRNA vaccines and HVAC systems if we end mass education?” but also more generally:

          One of the amazing things about our current age is that the answers to these questions, at all levels from a few-paragraph Wikipedia summary to a months-long course, are freely available to anyone who is curious. I think Caplan would argue that learning about mRNA and being a student in an educational institution are largely orthogonal. (You don’t need the latter to do the former.) This, I think, is an exaggeration, but I’m often struck by how, at universities, we *don’t* actually assess what people know or what they’ve retained, and are more (increasingly?) keen on rewarding sitting in a classroom and checking the boxes of participation.

        • Daniel asks:

          “How about how sound was recorded onto film?”

          In analog, optically! But that’s getting seriously obscure. (Although it raises a pet peeve of mine: transfer of old film to new media should not involve insanely wrong frame rates (playback could be off by the ratio of standard recording frame rates at the time to standard playback rates at the time, but by no more than that), and they should have no more frame-to-frame misregistration than the cameras of the time introduced (off hand, I’d guess essentially no misregistration at all; having run a 16mm projector for the MIT Film Society, my gut feeling is that it’s the projectors that are flaky), but current digital technology ought to be able to recognize and correct such misregistration easily (if they tried).)

          Another historical obscurity: how did pre-touch-tone telephone switching work?

          Speaking of obscurities, how does digital TV manage to squeese so much more information into a limited bandwidth (it would be interesting to know what percentage of people here know this: it ought to be quite high, I’d guess), and why was the NHK (Japan’s BBC) first high-def TV standard analog and not digital (hint: there was a good, but (historically, not technically!) wrong, reason)?

          But your question makes my latter point: good education isn’t so much about teaching the chain rule or integration by parts as it is about pointing out that there’s a lot more out there than what pre-college education can possibly provide*. If you read the blokes agonizing about the difficulties of teaching calculus in college, what you’ll realize is that they accept (well, grudgingly admit) that high schools can’t necessarilly (or don’t) teach even enough pre-calc. So they figure out how to do it themselves by providing remedial courses. And figuring out how to teach calculus less painfully. (Although this area has more than it’s share of ranters and ravers, it seems. Still, they’re trying.)

          Another point, though, is you need to be careful what you wish for. The Japanese high-school system does very good job at teaching a reasonable number of pre-STEM kids through calculus. (Most junior year high school exchange students to the US from Japan can earn a US high school degree that year.) But even the Japanese system has it’s critics. Among other criticisms, mine would be that the nominally hyper-democratic college admission system ends up only working for kids from well-off families and perpetuating, not ameliorating, class distinctions.

          *: One thing I see too much of in Japan, is idiot reporters asking children what their dream is, and the kids respond from the set of things they’ve been exposed to, which is a boring list of the usual suspects (albeit a slightly different one from the states), and then not questioning, or understanding that there are other things in life besides the usual suspects. And blubbering on and on about how kewl kids’ dreams are, when they’re actually naive and _uneducated_.

        • @Daniel Lakeland

          Since you keep plugging Julia in every comment, you make, I wonder what you think of this

          https://yuri.is/not-julia/

          I honestly feel a little similarly. I was briefly very enthusiastic about multiple dispatch and a potential solution to the two-language problem, but also found the ecosystem to be just shy of ready in terms of bugs. I also don’t feel like it’s moving anywhere on that front; it’s always expanding into new, often experimental features rather than bug-proofing what’s already there. In addition to the cultural and governance issues mentioned in the above post, I think there need to be more static analysis tools for identifying issues before runtime, if not a full on algebraic type checker like functional languages have.

          But I’m curious about your thoughts as an evangelist.

        • Hi somebody, thanks for clarifying your tone, it can be hard to express tone in written text.

          I read this blog post you mentioned in the past, but I’m skimming it again… Unfortunately there is no date on that post…

          The difference in the experience of using Julia between say 2018 (when 1.0.0 was released) and today is dramatic. The ecosystem and Julia core language itself has improved dramatically since I first started using it for real things in about 2019. There’s a lot of bug fixes and documentation improvement since 2018 in DataFrames, DataFramesMeta, CSV, GLM, Turing, Plots/StatsPlots, NLOpt, Optim, DifferentialEquations.jl and various other core data analysis and computing libraries.

          Prior to Julia 1.0.0 the language itself was unstable. People’s experiences from before 1.0.0 are not really relevant to today any more than the issues you’d have encountered in pre 1.0 Python are today. In fact, I don’t consider Julia really fully usable until mid 2019 or so which would have been about 1.2.0 or 1.3.0 mainly because of several enhancements that hit around that time, including enhancements to threading.

          A lot of what he links to is secretly bugs introduced by using OffsetArrays (but you have to click through his links to figure this out). This is kind of a known issue and it’s a bit of a hand-grenade into the Julia ecosystem. My advice is **do not use OffsetArrays** at this time. Also be aware that Julia is a flexible enough language that most of Julia is written in Julia so you can kind of fundamentally alter the language itself by just writing Julia. Something you can’t do in R or matlab or Python. This is a double edged sword you should be careful of.

          However there’s a reason most of his bugs are related to OffsetArrays and that is that such core breakage is rare in the actual ecosystem. People aren’t putting out special floating point numbers that can only represent integers divisible by 3 or Linear Regression code that can only represent slopes that are integer multiples of pi or whatever other weird stuff you could imagine. (You could in fact create all of those kind of things directly in Julia… but people don’t because it’d be silly… OffsetArrays isn’t silly, but it is niche at the moment)

          A lot of his bugs are also about disabling bounds checking using @inbounds. Something you should never do except in performance critical code that already runs correctly and is verified. Note also that Julia has a pretty great profiler and you can in fact profile and optimize Julia code in a way that you just can’t in R or Python because your performance critical stuff is always going to be inside some C/C++/Fortran library.

          The thing to remember about Julia is that it’s a young ecosystem. A lot of people compare it to say R. But I can tell you that back in 1998 when I first started using R version something like 0.62 it had lots of issues. It would just bomb in the middle of calculations because you ran out of RAM for example. You had to guess how big the RAM should be an pass arguments on the command line… and that’s even though in fact R is mainly a reimplementation of a language that was already 20 years matured at that point in time (S/SPlus started in about 1976).

          I also started using Python in about 1997 and at that time it had some weird issues which have been broken by later versions of Python… Python was originally created back in about 1989. Julia 1.0.0 was released Aug 8 2018
          https://github.com/JuliaLang/julia/releases?page=7

          So Python is basically 30 years older than Julia. the S language was first developed in about 1976 so it’s ~50 years older than Julia.

          The question is how much do these gotchas matter for everyday things one usually does when one isn’t doing something particularly weird (I consider the use of OffsetArrays pretty much unusual)?

          In my experience, using Julia for everyday stuff is reliable, easy, fast, and enjoyable. Also it lets me do stuff I just wouldn’t do in R or Python because it’d be crazy slow. Something as simple as writing a loop to run through a dataset and pick out certain special values. Let’s say something like you remember what the value of the X variable is for the last 10 rows and if in any of those variables it was X* then when the current row has X* too output that row… Good luck coding that in Python you’d write loops and your own queue data type and have bugs and the whole thing would take ages to run.

          In Julia you’d hit up the DataStructures.jl library. You’d create a Queue, you’d start a loop, run through each row of the dataset, you’d push stuff in the queue, you’d look in the queue to see if the current row matches any of the last 10 rows, and then you’d move forward, popping the oldest stuff from the queue. The whole thing would be readable and run at warp speed compared to Python. Maybe you’d do a couple billion rows in a few seconds compared to maybe a few hours to a few weeks in Python.

          Now, here’s the thing… If you’re someone who is almost exclusively a user of plug-together analysis…. like someone elses R+Tidyverse pipeline combined with a linear regression and a ggplot2 plot and that’s all you do… Then you won’t see the advantages of Julia. If what you do is use a computer like a very fancy pocket calculator which has special “pre made” buttons for doing pre-canned analyses… then pick a language that has that pre-canned analysis and use it. No problem.

          If what you want is a language which can do some pre-canned analyses, but also gives you the flexibility to do at high speed your own custom calculations… Then Julia is really the only game in town. The fact is that with built-up ecosystems from decades of development, a lot of people have gotten used to the idea that R or Python or something just works to do the kind of canned analysis they want, and if they want something not canned they’ve just gotten used to the idea that you can’t really do it, and they’re often ok with that. Matlab is similar, a huge array of high quality pre-canned “solvers” or “toolkits” Or whatever from a company that has a huge revenue stream to fund development and documentation.

          In fact, people have internalized this so much that they don’t even realize it. They may have never worked with a language that doesn’t have this kind of limitation. How many undergraduates graduate today having never used C/C++/Fortran/Common Lisp or any other of a number of languages that compile to actual machine code? Probably the vast majority. Even if you took a few computing courses for your Stats or Econ or Engineering degree, it was probably in Matlab or Python or R.

          Let me just give you an example. Suppose you have an expensive function to compute, and every time you compute it you want to first check to see whether you’ve computed it for these inputs before and then just use the previously computed value… this is a “memoized” function, a standard concept in Computer Science. Suppose we’re doing this in R. Then just f*** you, you can’t. I mean, ok, sure recently someone implemented the “hash” package in CRAN so people could finally have hash tables… only 50 years after the introduction of S at Bell Labs. https://cran.r-project.org/web/packages/hash/index.html and they’re doing it by kind of abusing a native data structure called an “env” (environment). So I guess you can now maybe, as long as the arguments to your function can be put into a hashable form. But it’s a huge kluge.

          Now in Julia, you do:

          using Memoize.jl

          @memoize function myfunction(myargs)…

          end

          seriously, you just put the macro @memoize in front of your function definition and it becomes a memoized function. There are some tunings you can do, for example you can use a “least recently used” cache so it only remembers the last N function evals etc etc… but at its core, the language is powerful and general purpose, so stuff can just work. Programmer defined data types are first class in Julia, so even if you’re passing incredibly complicated structures to your funciton, that structure will happily go into the hash table used under the hood by Memoize.jl

          Here’s what it comes down to, Julia is a language for people who have taken at least a couple semesters of computer science classes, know how to program, and want to do scientific computing and data analysis.

          R is a language largely designed with its core purpose (back in the origin days of S in 1970’s) to let statisticians who have no real programming skill at all to run canned routines written in FORTRAN77 without needing to learn FORTRAN… Python is a language that was designed originally, back in 1989, to be interactive and easy to use for first year undergrad CS students to get used to core ideas in computation… and then got a lot of special fast C/C++ code to do numerics and other special things bolted on… I remember when numpy was the new hotness… that was like 2005.

          Both of them are the origin of the idea of the “two language problem”. Ie. there’s one language you work in (R or Python) and one language that actual computing is implemented in under the hood (C/C++/Fortran).

          It just depends on who you are. Are you a person who find the “two language problem” to be a problem? ie. you want to be able to write your own fast code from scratch? Then Julia is for you.

          Are you a person who is honestly kind of afraid of programming? Maybe you’re a biologist who spends 90% of your time doing wet-lab stuff, and then every few months you have to run your wet lab data through some pipeline and get some canned answers? Or you do regulatory affairs and you have prescribed specific analyses you must by law do every week? Then R/Python isn’t a problem for you at all it’s a solution.

          if you’re someone in the middle, where you do data analysis, and every so often you get a situation where you really need to do something kind of custom that there doesn’t exist a pre-existing package to do, and you need it to run over a huge data set and you need it to be fast, and you’d really not like to have to pull out C and muck about with casting your function pointers to void * to hand to a publicly available C based optimization library, and etc etc… then you should be looking at Julia.

          I just happen to think that having a language that is capable of being both fast and flexible/interactive is a huge huge advantage. The people who are like me hang out on discourse.julialang.org basically. There are a bunch of us.

        • These days, a python user who needs fast differentiable computing would probably use JAX, which you might be interested in (and if they need slow differentiable computing, they’d use pytorch)

          https://github.com/google/jax

          which is essentially numpy with a JIT compiler and an algebra of dual numbers. But, since it has its own special syntax, you can’t always just import functions and move them into a jit compiled block or otherwise do that naive “pythonic” thing. Not quite a solution. Also not up to 1.0 yet either.

          And yeah, I struggle the two language problem. Without giving too many identifying details, we need to fit a model’s parameters in one framework, then optimize some other variables for a business utility function keeping the parameters fixed, but the framework doesn’t have much in the way of nonlinear optimization. Also, I/O is slow, especially if you need to copy data over to the GPU (keeping my fingers crossed that a unified memory architecture catches on). But in any case, Julia doesn’t solve the two language problem for just everyone yet — its interfaces for cloud deployment and other web technologies are a little lacking compared to its numerical capabilities.

  9. In pedagogy there is the principle of the most important correction: an instructor should focus on the most important mistake the student is making, because its hard to change several things at once as you do them. I follow a similar principle in my academic critique: there is no point quibbling about whether footnote 23 cites the latest research if the main argument is fallacious.

    So if people are not responding to the details of Caplan’s spreadsheets, maybe they are not the weakest points of his argument. I agree that submitting his argument about the cash RoI of education for peer review would be one good way to get focused criticism.

  10. “research doesn’t have 100% efficiency, which is a product of the decentralized system that we have”

    I’d argue its the nature of research. If you can reliably estimate what it will cost to answer a question or make something, and what effects that answer or thing will have, then its not research but engineering or another practical art. Applied research often fails or has unexpected results too.

  11. I only have one complaint here. You say, “Some social science research is really cool, but it’s nothing compared to natural-science breakthroughs such as transistors, plastics, vaccines, etc.” The problem here is that you are looking at engineering applications of natural science, not the science itself. The theories of special and general relatively are cool, but you can read and understand them and still have no idea how to build a nuclear device. That took a massive engineering project that could easily have failed, just like the German and Japanese efforts did. The science indicated that a nuclear device was possible – that’s why Einstein wrote the letter – but the Bomb didn’t appear until a variety of practical problems had been solved. Same, btw, with “… transistors, plastics, vaccines, etc.” Most natural, life, and social science has absolutely no immediate and apparent use at all. Except, of course, that it helps us understand what’s going on around us. And that can become very useful. More often in the natural sciences then in the life and social sciences; it’s sort of a sliding scale. But that doesn’t mean that the research itself shouldn’t be done.

  12. It seems like most people here (certainly me) agree that our formalized education system leaves much to be desired. But missing are explanations why that is the case. Part of the reason has to do with bureaucracies and the institutional incentives within education. But another role – often neglected in my view – is that our society doesn’t really want too many educated people. Politics and business depends on having customers/voters who are incapable of much critical thinking. In that sense, education (yes, schooling is a better term) is doing what it is supposed to do. Only the elite emerge capable of critical thought. Politics and business succeeds based on the masses, not the elite. How else can you get people to succumb to idiotic political stances and how else can you get people to purchase things they don’t need for prices they can’t really afford?

    Am I too cynical? (I don’t think so). And the movement to make education more practical and applied – which I agree with – will not change what I am thinking. Yes, make education more relevant and practical. I do this in my own teaching and it is far too rare – witness the typical US textbook or syllabus to see just how far removed much schooling is from being relevant and practical. But let’s not abandon the idea that education should do more than that.

    Universities spend a lot of energy designing their core curricula (general education requirements). Of course these have more to do with filling the classes of those less popular subjects and keeping tenure track people in their jobs than with the real purpose of broadening and deepening our students’ thinking. I’d reverse the sequence – start with the major and add those other dimensions when some degree of competence has been acquired. I believe a capable engineer would gain much from a good philosophy course after becoming somewhat proficient as an engineer. The idea that they should study philosophy first, before focusing on engineering, strikes me as very strange. The usual justifications are that they should be exposed so that they choose the right major – that is one form of signaling. The other justification is that it will make them more well-rounded and better capable citizens. The purpose is admirable, but delaying study of what they are most interested in by forcing them to follow an indirect path seems like a weak justification to me.

    • “Politics and business depends on having customers/voters who are incapable of much critical thinking. ”

      I vehemently disagree!!!

      Both political parties and businesses – as well as many other types of organizations like labor organizations, NGOs, etc – are intensely competing for your money. In a world where highly educated people make the most money, it would be suicide to presume a stupid customer!

      “a capable engineer would gain much from a good philosophy course”

      I don’t think the average engineer would get anything out of any course that they are forced to take. Academics have freedom to research whatever they want and they claim this is a great benefit to society, yet they want to take that same benefit of freedom to learn away from other people!!!! This is a common problem in society: people who get theirs but then want to deny the same opportunity to other people.

      • Intense competition for my money:

        that’s why insurance coverage is so easy to understand?
        that’s why cable TV options are so transparent?
        that’s why most financial products sold to retail investors are so easy to understand?
        that’s why so many extended warranties are sold (and bought)?
        And, of course, that’s why so many political stances are so sophisticated and well thought out?

        I could go on, but apparently I live in a different world than you do.

        On your other point, I don’t favor forcing anybody to take any courses. My qualifier “good” philosophy course was meant to convey a course that provides relevant content for their audience. Engineers require and care about ethics too.

        • “I could go on, but apparently I live in a different world than you do.”

          Not at all. :) Why do you think they go to all that trouble? Because people are dumb? :)

          But, sincerely, I don’t get your point. I don’t have too much problem understanding financial products, tax instructions, legal documents, cable contracts, insurance coverage etc. And all of these products and services are so stupidly cheap it hardly matters. If you could offer a modern internet connection to a family living in 1965, what would they be willing to pay for it? Probably a lot more than the $75 I pay today! Even a cable subscription with HBO from 1985 would have been a crazy impossibility to a family in 1965.

          And financial instruments!!! JHC Dale, in 1995, to keep track of your investments you had to buy expensive newsletters, and a single stock or fund trade cost thirty-five-1995-dollars and you had to mail the goddamned order!!!! Today you can log on to your Fidelity account, compare thousands of stocks, mutual funds and ETFS, and buy anything you want for exactly zero dollars.

          I don’t know what you’re complaining about. You have a very narrow and immediate scope.

        • chipmunk
          You are obviously much smarter than me (seriously). But I know I am much smarter than average, so the idea that the contracts you cite (legal documents, financial products, insurance coverage, cable contracts) are simple to understand is clearly incorrect as an empirical observation. I’ve spent many hours trying to decipher these things, usually with little success. As I said, you are obviously much smarter than me – but your empirical reality is way off.

        • He is not saying he understands them, he is saying that whatever amount he’s being cheated out ($70 for internet) of is nothing compared to the overall improvement in private services over time.

        • somebody
          Thanks for the clarification. But now I have no idea whatsoever what chipmunk’s comment means. If we accept that we are so much better off than in the past (I’m not sure I agree, but this isn’t the place to debate that), what does that have to do with critical thinking and the purported “intense” competition for customers?

        • Dale Said:
          “the contracts you cite (legal documents, financial products, insurance coverage, cable contracts) are simple to understand is clearly incorrect”

          I didn’t say “simple”. I said I don’t have a problem understanding them. Legal, financial, insurance and tax instructions take time. They are complex. However I’ve found the wording is usually fairly clear when followed carefully. Recently I had a row with my sibling over some legal documents. Here’s what I found: sibling didn’t know things what were simply and clearly stated in the documents. They weren’t too complex at all, sibling was just distracted and disinterested while reading them.

          “what does that have to do with critical thinking and the purported “intense” competition for customers?”

          You said companies need dumb customers. I say they couldn’t survive if they assumed their customers were dumb.
          Most financial products, insurance, cable contracts, cell phone deals etc all boil down to about the same service for about the same price – it has to be that way because they all have the same costs. If some company came along and tried to offer the same service for 20% more, no one would buy it! People aren’t stupid.

          Sometimes companies sell products or services that some people appreciate and others don’t. I think paying more for organic or “natural” food is stupid. But some people value organic food. Are they stupid? Or am I stupid? How do you know?

          I prefer Legos over Megablocks. I’m willing to pay more for them. Am I stupid? Does Lego rely on my stupidity?

        • response to chipmunk (again) – and in the spirit of Andrew’s post today, this is the type of interaction better suited to blogs than to twitter, at least until Andrew admonishes us to cease this exchange:

          I’d suggest taking a look at https://www.deceptive.design/ just for some examples of how businesses intensely compete for us smart customers. More importantly, here is one example: my courses right now are working on an assignment for simulating the comparison between gold, silver, and bronze health care plans obtainable through the federal health care exchanges. It is not an easy problem and they struggle quite a bit. But it is amazing how many struggle due to a basic lack of understanding how insurance deductibles work. I’ve simplified the assignment quite a bit (to omit the hundred pages of details in each of the plans and reduce them to their most essential financial elements) – but an amazing number of students just add the deductible amount to total costs, regardless of whether or not those expenses have been incurred, and a number of other students apply copays to the expenses before the deductible has been reached. I should mention that these are working professional graduate students.

          I would maintain that most large businesses and almost all politicians gear their messages precisely for the stupid population (stupid is a poor word to use here, so perhaps uneducated would be better). Chipmunk, you apparently see things very differently. But I do not see the evidence that it is “suicide to presume a stupid customer.” Take a look at the number of businesses whose main source of revenue is selling extended warranties. Or look at the number of people who lack flood insurance and don’t understand what their homeowners’ policy does and does not cover. Or look at the number of politicians that feed the public economically moronic policy positions. Or….. (do I really need to go on?)

        • Dale:

          As I pointed out in my family story, my sibling was too lazy to read the details. That’s what’s up with your students. They’re not being deceived. They’re too lazy to read the details. The information is right there. They just don’t want to bother. How is that the company’s fault?

          I suggest that, instead of condemning companies because your students don’t want to bother reading the details, you should look at the business case for the companies providing the plans and try to understand from their perspective why the plans are written as they are.

          But also you’re ignoring the competitive aspect of the industry. There are dozens of non-profits providing health care coverage. If the for-profit companies are so bad, why aren’t the non-profits running them out of business?

          That’s the general answer to most of your claim. Companies that are too deceptive or offer poor products eventually get outcompeted by companies that offer better products and services. Because people are smart, they migrate to the better products. That’s why we have smart phones instead of flip phones; computers instead of typewriters; Hondas instead of GMs etc etc.

          I agree that companies sometimes use deceptive business practices or advertising to varying degrees. Customers do the same thing when trying to get refunds, exercise warrantees, recover losses through insurance companies etc. I’m sure your many of your students will one day try to cheat their health insurance. that’s one reason why the language of contract is sometimes complex: people will take every opportunity to cheat.

          That’s enough for me down this rabbit hole!

          Cheers Dale

  13. A common notion when this discussion around education pops up is that education is not about memorizing facts but about fostering critical thinking (learning how to learn). I think this is generally used as a defense of a liberal arts education and also as the justification behind changes in the way certain subjects are taught, primarily I’m thinking of a shift in the way math courses are taught. Personally, I’m very sympathetic to this argument, but I have been wondering if people (particularly educators) see/feel the results of this. In many countries a higher percentage of young people have some amount of university education than ever before. So posters/commenters living in these places do you notice any difference with younger cohorts when it comes to critical thinking or the ability to learn?

    • “In many countries a higher percentage of young people have some amount of university education than ever before. ”

      The Flynn Effect: “The Flynn effect is the substantial and long-sustained increase in both fluid and crystallized intelligence test scores that were measured in many parts of the world over the 20th century.”

      https://en.wikipedia.org/wiki/Flynn_effect

      • The Flynn Effect seems to have peaked in the 90s. I’m also more interested in the qualitative experience commenters have had. Many posters here are older and some are professors, I’m just curious if they can notice a change in the cohorts. A couple of days ago there was an NYT article where an Organic Chem Professor claimed that his recent cohorts were less focused than previous cohorts. Many people in the comments section additionally claimed or supported the idea that students entering university today are less prepared than previous cohorts were. To me this is very strange given the trends I highlighted earlier. At the same time, people have a tendency to romanticize the past. I’m too young for there to be a huge difference in the education I received compared to the education my juniors in the workplace received, so I’m just curious what the experience of others has been.

        • Blackthorne:

          The chemistry professor in the news used to teach at Stanford and then more recently was teaching at NYU. I would expect that chemistry students at NYU would be on average less well prepared than chemistry students at Stanford. Maybe if he was teaching film studies it would be a different story.

        • “The Flynn Effect seems to have peaked in the 90s. ”

          People have said that but the wikipedia article says otherwise.

          But perhaps students have not changed but faculty perceptions are actually correct: that could happen a greater proportion of the population goes to universities, with most of the increase coming from the middle and lower part of the ability distribution.

          BTW, it’s pretty well established that “grade inflation” is a real thing.
          https://www.gradeinflation.com/

      • Backthorne –

        > Many people in the comments section additionally claimed or supported the idea that students entering university today are less prepared than previous cohorts were.

        I think that the “kids today” complaint has around as long as there have been kids.

        Bawss on that, I’d say that over so many generations of the complaint, running around in caves dressed in skins should be considered an unusually high standard of living.

    • “do you notice any difference with younger cohorts when it comes to critical thinking or the ability to learn?”

      This is a tough question. I’ve been a professor at a public research university since 2006, and I often teach “general education” courses that sample a wide range of non-science-majors. If I have to give a simple response, it would be that in this period, these skills (critical thinking and the ability to learn) have declined on average, but their variance has increased. The amount that one can expect the average student to read has gone down, severely. Math skills are probably at the same level of mediocrity as ever. University writing courses (at least here) focus on expressing ideas rather than forming coherent text. All of these make up “ability to learn.” In tandem, student expectations that their grades reflect self-assessed “effort” rather than mastery of material have gone up (see the recent NYU Chemistry case). On the plus side, the top end of students is as good as ever, and those that take advantage of the many free, high-quality resources out there — whether it’s to learn topics or conduct research — can do amazing things.

      I think many of the comments here reflect perceptions formed much longer ago. The typical university student is vastly more likely to be a business major or a psychology major (together > 20% of degrees here) than any other major, and questions about the value of higher education should be considered through this lens, not whether the chemistry or mechanical engineering major’s degree is mostly “signaling.”

      • Indeed here’s recent graduation stats:

        https://educationdata.org/number-of-college-graduates

        Here’s a pie chart partway down that page:

        https://educationdata.org/wp-content/uploads/76/bachelors-program-graduates-degree-disciplines-1.webp

        Over half of degrees are: Business, Healthcare, Social Sciences + History, Psychology, and Bio+Biomed sciences.

        Now, how many of those people have had a course in mathematical analysis of dynamical systems… I’m not talking a course for math majors, I’m just saying courses that discuss “given what’s going on this year, what might we predict for next year?” like discrete time systems? Predator-prey, or sales forecasting, or whatever. Just a whole bunch of examples of “given the world is like this at time N, and some approximate rules… what will it look like at time N+1”?

        Aren’t time-series really important for business, for health care, for economics, for biological growth?

        How about basic ideas in computer programming? Say iteration, loops, basic data structures (dictionaries, trees, queues, priority queues), basic algorithms (sorting, hashing, whatever)? Can we pretend those aren’t all critically important for doing good business analysis? for economics? for designing decision trees in healthcare diagnosis? For genomic data processing in Biology?

        How about basic ideas in micro-economics like optimizing outcomes subject to a budget constraint? Business, Healthcare, Econ, even questions like how biological organisms spend their energy budget or ecological competition?

        In fact, what about the core idea of optimization and some basic algorithms of optimization? Ever heard of least squares? or trying to maximize the health care outcomes while minimizing risk of death?

        How about probability? Can you tell me that doctors don’t need to understand issues about probability in differential diagnosis of unusual diseases? That business people don’t need to understand that not everything goes according to the most likely plan in your spreadsheet? that biologists need to understand variation in behavior from one organism to another?

        People keep talking about how we need to make school more “practical” but honestly there’s nothing quite so practical as a good applicable theory. You can save a lot of time and billions of dollars in waste by just realizing early on in your product development cycle that if you use the idea of optimization you might be able to avoid doing a lot of useless crap, and if you understand what a SQL database is and something about Codd Normal Forms you could hire the right kind of person to make one that would let you answer a lot of questions in the future that without the answers you’d otherwise keep wasting time doing the wrong stuff.

        Not everyone needs to know the **details** of everything… but the purpose of high school and undergraduate education should be to give you enough foundation that you can know **what questions to ask** and how can you even know what questions to ask if you know essentially nothing about many **Core ideas of the modern world**. If there were a single core idea in the modern world it’d basically be some kind of amalgamation of time-series analysis, databases, computer processing, optimization, and probability/risk.

        My wife has masters students in biology who have never ever even once used a command line to do anything, much less have read in a dataset to R or Python and made a graph.

        Seriously, all the way through high school and undergraduate education in biology and have never used a computer to make a graph other than clicking in Excel following some instructions in a canned lab writeup.

        There’s nothing modern “education” is quite so good at as siloing knowledge between different groups of people. The CS people learn algorithms, the Information Systems people learn SQL databases, the Biologists learn PCR, the Econ people learn STATA regressions, the Math majors learn optimization, the Engineering majors learn strength of materials…

        • Daniel:

          You write that the math majors learn optimization. If so, that’s great. When I was a math major it was 100% theory. There were some applied math classes but there was a general understanding that the applied-math track was just there for the math majors who couldn’t hack the theoretical stuff. This was at MIT, where most kids were engineering majors, and math and physics majors were considered to be impractical majors, kinda like how it must look at other colleges to major in poetry or something.

        • Yeah, I guess mileage might vary. I did a kind of mixed theory and applied math major. I did a major thesis on the use of genetic algorithm optimization techniques to approximately deconvolve audio signals to create flat frequency response audio reproduction systems. I also had a computational numerical methods course at the 400 level which discussed some optimization methods.

          This was at a not-prestigious state school in the midwest. It was a great education honestly. What it was NOT was a great “signalling” method. If it were me and a Caltech grad applying to the same job the Caltech grad would probably get it, regardless of whether they were good or not.

        • Daniel:

          Yes, these were different eras. When I went to MIT, I think they were accepting 1/3 of all applicants. And, even so, I almost didn’t get accepted at the time. I’m still not quite sure what they were looking at in terms of ideal applicants: maybe students who didn’t look so much like “nerds”? In the early 1980s, that was before nerds were cool.

        • Hah! I have met faculty in biology who were similarly unconfident in working with data in that way. I assumed it was a function of disciplinary priorities.

          I work as a statistician in a medical research group and siloing is a fact of my life, not just an incidental one but part of its skeleton. I find it invigorating, if also intimidating, that the people I work with don’t all know the same things. Is it good for science when coauthors don’t all speak the same metaphorical language, I have no idea. I am sure there are inefficiencies and that inaccuracies get through. But I don’t know how else you’d do it in a practical sense. There’s just a lot to know, now.

          I spent a lot of years getting formally educated, but I have forgotten far too much of what I studied for the tastes of commenters on this thread. Oh, well. I will say I see education entirely differently now that I am a parent. We parents make a trade: we give up some control over what and how our kids get taught in exchange for some reassurance that they will be welcomed into society as adults, so that when we are gone, they will still be safe. What my fifth grader will retain in forty years about Mesopotamia or identifying leaves is going to be a function of her interests, her life experiences, and the vagaries of these frustratingly fallible organs we have to keep our memories in. I hope she remembers enough of it to be a good participant in society. I suppose someone with Bryan Caplan’s advantages doesn’t worry about that very much at all.

        • “Over half of degrees are: Business, Healthcare, Social Sciences + History, Psychology, and Bio+Biomed sciences.”

          You say that as though that’s a problem. It’s not. Really. It’s not.

          Some factoids.
          (1) US median income has been rising steadily for the last 25 years; we’ve been doing fairly well by the “median American”.
          (2) Since the beginning of the Obama recovery, the US economy has been create gobs of really good jobs; jobs in the 25, 35 and more dollars per hour range. Year in, year out. Usually well over 100,000 new jobs a month.
          (3) More and more Americans are graduating from college.

          I think all of that is really really good. Including the last bit. Because those jobs require, if not a 4-year degree, then at least some post-high school study. And college grads live longer, healthier lives than high school grads. Even if you don’t think current college education is perfect.

          But the final point is that every single one of those degrees involves learning things you personally don’t know. That is, your definition of what people are defective for not knowing is arguable. Here in Japan, the MIT Alum club is dominated by Sloan School grads. It’s boring. I got irritated with it all and attacked one of them. “This “corporate culture” stuff is inane.” I said. “No, it’s not. Here’s an example: A Japanese supertanker builder decided to diversify into DRAM memory back in the bubble days. It didn’t go well. Why? The company made money by building one supertanker every ten years. Now it had to sell a million products a month. They weren’t good at it.”

          That’s Business. Then there’s Healthcare: you personally would be real unhappy if you found youself being cared for by someone who didn’t have a “Healthcare” degree. And every other item on that list is something most people here think is of value.

          I submit that you’ve got this whole thing quite wrong: College is just fine, they’re really trying to do a good job. Sure, there are improvements to be made, and it costs way too much. But throwing the baby out with the bathwater is a right-wing thing due to the “problem” that US college grads are smart enough to read Ayn Rand and realize she was an idiot (and a really terrrible writer).

        • David in Tokyo: if I understand correctly, Caplan would say you are confusing correlation with causation (ie. college grads live longer and earn more than non-college grads, but that might be because the traits which let them enter and graduate college also increase health and income). I’m not inclined to see his book and judge for myself because his kind of person gives me a headache, but I agree that a lot of discussion of the benefits of education is not very rigorous.

          I agree with our host that a lot depends on what you consider a benefit. Universities are complex systems that serve many functions.

      • Erin,

        The details will always be siloed… no one can be an **expert** in everything. But far too much of even the basics are siloed now. A biologist should know something about making graphs and reading in data. An economist should know something about biology. You don’t have to be an expert, but a bio-statistician such as yourself should know for example how RNAseq works, something about optimization of general purpose functions, some computing and SQL databases stuff, a bit about the chemical reaction involved in PCR, a little about how the immune system works… etc etc. When a client describes their biological experiments you should be able to follow along and ask reasonable questions. Probably you can! otherwise you’d likely not be doing what you’re doing.

        I’m sure you know a lot more than you are willing to really take credit for. I also am sure that there are actual immigrant gardeners who never even entered high school in their native Mexico or central american country who don’t know the difference between Acetaminophen and Ibuprofen, or the chemical differences between gasoline and diesel fuel other than you put one in one kind of vehicle and the other in the other kind. some things we take for granted that we don’t even know that we know.

        Still I think there are a lot of really ignorant people even those who have undergrad degrees. Its because we’re not teaching the same level of stuff.

        • @David in Tokyo, and others: You might wish to actually read what Bryan Caplan writes, not what you imagine he writes. Points like “college grads live longer, healthier lives than high school grads” are true, and are actually discussed in his book. Similarly, the reasons for the correlation between education and income, which would occur for both “signaling” and “skills development” models, is discussed at length.

        • Short of reading the book, reading this interview might be instructive.

          https://www.vox.com/conversations/2018/2/16/16870408/public-education-libertarianism-democracy-bryan-caplan

          There’s quite a bit with which I agree (in particular his point related to how little we know about how to “teach” thinking).

          Nonetheless, I think that overall his conclusion suffers from binary thinking, and is ultimately more political than empirical (which, to his credit, he pretty much acknowledges)

        • It would be interesting to have the data on those who are saying that there’s a case “against” education (and that as Caplan says, our entire educational infrastructure should be dismantled and abandoned) and compare that set to the set of people who have focused a great deal on how much harm has been created by schools closing throng the COVID era.

          Methinks there’s a lot of overlap between those sets.

          It troubles me that there seems to be a lot of people who have difficulty holding together diverging factors, such as that schools are clearly sub-optimal but that they still have a net benefit, or that closing schools during the pandemic has clearly been sub-optimal but we don’t really know what the damage would have been had they all remained open throughout.

          Counterfactuals are hard, yet it seems that many people are very willing to make counterfactual assumptions at the drop of a hat when it serves their ideological orientation to do so.

      • David, you have misread me entirely I think.

        I don’t think it’s a problem those are the degrees being taken, I just think that because that’s 50% of the degrees it’s worthwhile to look at what those particular people are learning. I’m not convinced that what they are learning in undergraduate education is worth even 1/4 of what they’re paying for it *as education*. Now are they getting their money’s worth as “signalling” thereby leading to greater income? Perhaps. I don’t know. I don’t really care. I don’t buy that in the long run our society benefits from that. Kind of like I don’t think China’s society benefits from making 30% of its GDP from building shell buildings that must be knocked down soon and will never be occupied and their main purpose is to be a shell game of musical chairs where you only win if you’re not sitting in the chair when it gets the wrecking ball.

        Your factoids are I think very wrong. It’s not easy to define “real” income, but by most “standard” measures median real income in the US has stayed basically flat since the late 1970’s with some recent increases that are about to be wiped out by inflation. Of course, that’s based on what I think is a flawed measure of “real income”. What I personally think we should measure this by is the following:

        what is the ratio of household income of a 2 parent 2 child family divided by the cost to pay taxes + a mortgage on a median house purchased today + loss of income from stock market investments sold to make down payment (or alternatively rent on a rental 3 bedroom house) + transportation to and from work + food for 4 people + child care + health care + education all amortized over a 25 year time horizon from family initiation to graduation of the youngest child from college debt free (so including the cost to pay down the college debt)

        By this measure, I am fairly confident you will find that things are dramatically worse off than they were in 1970, which is why we are at record low fertility rates and marriage rates and record high suicide rates and mental health crises and drug abuse.

        Now I have no problem with the idea that people learn things I don’t know, and I learn things people don’t know, and we all get to mix together and support each other. That’s fine. But I think there are some things that pretty much **everyone** needs to know. For example my friend with a PhD in biology was recently ripped off by an Audi dealer who charged him $250 to “install a cabin air filter”. He later discovered that this literally means open the new air filter from its cardboard box ($25) and then pop the hood, reach in and pull out the old filter, and slide the new filter in… Worth absolutely no more than $27 total.

        now, in general, people just need to make decisions about things in their life, health decisions, investment decisions, business decisions, education decisions, etc. If people don’t know anything about anything except their little corner of specialty, then they are going to get raked over the coals in every decision. My same friend recently bought a whole home solar energy system. The first system which he almost decided to do cost $65k, after my urging the final system he decided on is like $25k.

        So, everyone’s real income goes down, because now every goddamn thing they are getting overcharged for by a factor of 5-10x because they dont’ know any better. And instead of making progress in the world, we just have scams utilizing asymmetry of information to its maximum to extract the maximum rent from everyone with the absolute minimum of “consumer surplus”. I think that describes the situation today quite nicely, and I think a lot of it comes down to the average person is vastly ignorant of basically a vast sea of things they need to know including things like how heat pumps work or finance calculations amortizing a solar system over its lifetime, or how ARM mortgages with 0 interest initial periods work, or whether it makes sense that Tesla is a bazillion dollar company that loses money on every car it sells and only makes money on its secondary market in government carbon credits, or how internet security works and whether it’s a good idea to buy a Ring camera and what information about you might be exposed under what circumstances… and what the hell is an NFT anyway and should I put $200,000 in bitcoin to try to get incredibly rich or not?

        there’s just a TON of stuff people actually REALLY NEED to know or they will lose out big financially or health wise or whatever, and they don’t know these things and they’re paying people to teach them other stuff which is largely crap but which “the system” tells them is important to learn because it’ll “get them a good job”.

        So, quite honestly, I am all for education = acquiring useful knowledge. I’d love it if we started doing it.

        • +10
          Although for some of these things, one can educate themselves quite easily by going on YouTube or something. I work on my own cars, do major home repairs like removing and installing a new tub and tiling shower surrounds, etc. I can learn that on YouTube….anyone can learn how to replace their car cabin filter.

          But great points. I would say most of university education was useless. In fact, I’ve had to teach myself Bayesian methods because all I learned (other than the most basic examples) in grad school was frequentist.

          >whether it makes sense that Tesla is a bazillion dollar company that loses money on every car it sells and only makes money on its secondary market in government carbon credits

          I have to look this up!

        • jd. Yes you can learn a lot by watching some YouTube videos, reading some books, and googling… IF you have the background knowledge needed to know what is good information and what is crap, and how to interpret that good information… Core background knowledge is required.

          For example, suppose you’re trying to figure out whether solar power system A or B is better for you? Up front costs are very different, one has batteries and the other doesn’t. They produce different amounts of power. They have different subsidies involved. The battery puts you in a different tariff than just solar. The power consumption varies by season. The power prices vary by season. The future power prices will vary due to inflation which has a bunch of uncertainty associated with it. The two systems have different loans associated with different interest rates… the monthly payments are different and the time periods are different… the batteries may or may not actually run your house during a power outage…

          How do you even approximately understand how to calculate a life-cycle cost associated with all of this if you’ve never even been taught anything about discounting cash flows, or electricity? How many PhDs in biology have had a course in which they learn to discount cash flows? This is NOT optional information you only need if you’re an Econ or Accounting student. This is core information you will need throughout your life to avoid been raked over the coals by scammers upselling you on bullshit. Also, you may have plans to install various equipment, like a pool pump, or a heat pump system to replace an aging HVAC. How will you decide how to size your system now to accomodate the future plans if you don’t understand wattage, current, voltage, and the relative power consumption of what you might be considering vs what you have already?

          And how are you going to even get anything out of that finance class if you haven’t got enough math background to understand exponential functions? How will you decide how much power your future house systems might consume if you haven’t had some kind of physics class involving electricity?

          The answer is you will be maybe $500k or $1M poorer by the end of your life because you are letting other people make the decisions and they are always making the decisions to maximize the amount of money they can extract from you without alerting you to the fact they’re selling you bullshit.

          You don’t have to be an expert, you don’t even have to actually be able to carry out the calculation. What you need is sufficient background information to understand when the expert advises you, whether they’re shitting you and you’re getting $27 worth of filters for $250 or whether they’re being honest with you and advising you in a reasonable manner.

        • I’m only skimming the discussion here, but I wanted to point out the inaccuracy of a few early claims made in this post (particularly given the “Your factoids are I think very wrong” statement).

          The claim that median real income in the US hasn’t risen since the late 1970s is demonstrably wrong (https://fred.stlouisfed.org/series/MEPAINUSA672N). You suggest data like those are unreliable (a not altogether unreasonable idea), and in response, create an artificial consumption bundle you imply would be more accurate in capturing real income. (That said, some of the things included in your bundle–like college tuition–would appear to be a bit asymmetric in terms of effects, given the increase in secondary education rates since the late 70s.) Some of those costs are significantly higher than in the late 1970s (housing, tuition), some of those costs aren’t particularly significantly higher (cars, food), and some of those things are difficult to compare (healthcare, for example, since it simply wasn’t available to many people in the late 1970s, and while costs have increased, so have medical capabilities).

          It should be noted that there are many costs that have decreased in real terms, and many goods/services that simply weren’t available in the 1970s, but that’s debating your chosen consumption bundle.

          Finally, scams intended to extract consumer surplus are not new, and I’m hesitant to believe they’ve increased in frequency or resource-extraction.

  14. This is a very interesting post and collection of comments (so far).

    While I was in college, it seemed to me that most of my fellow students (not the best ones) would be better off in a master-craftsperson and apprentice system. College is a structured opportunity to learn, at a time when most people don’t know what they should learn; a tremendous opportunity which I for one didn’t take full advantage of. Like most things, if I could have redone those years via a time machine I would have gotten much more out of it. Still, at about $6000 a year in 1960’s dollars (as best i dimly recall) it was well worth it.

    When I joined a GE engineering department in 1968, GE was all about training. I probably learned as much in GE training courses and from expert fellow workers as I did in college. One of the dumbest of Jack Welch’s many dumb ideas was to say, “Why are we spending all this money on training? Why not out-source the training to the colleges and just hire engineers with Master’s degrees?” Learning is a necessary part of progress. When we stop learning anything new we will stop making progress.

    My dumb idea of how to improve learning is to combine formal courses with relevant working experiences as I did at GE. Four years of several hours of classes a weekday seemed like a long time to me. I was glad to get to work rather than continuing to grad school.

    As for signalling, that is the hirers’ problem, it seems to me: people like Welch who want a cheap and dirty way to sort through job applicants. The two best managers I ever had gave me very detailed, searching reviews of my capabilities before selecting me.

  15. I read Caplan’s book when it was published. I had been a regular reader of his blog posts and while I am opposite to him on the political spectrum, I thought it would be useful to see what he had to say. At the end of the day I was pretty much in accord with Andrew’s points above and many of those were on a five page set of notes that I sent him after my reading (his response was not even thank you to a careful reader and a response to what I thought were valid criticisms. The response back was ‘I’ve moved on to other things.’ I don’t know the background of a lot of those who commented above but my most of my working career post-PhD was in the biopharma industry. I had the chance to manage a department and hire junior people. Signalling was never an issue and I had not heard of the term until reading Caplan’s book.

    I found it odd that the book was little reviewed and those who did were not impressed; neither was I.

  16. I can’t speak to how it used to be, but I can speak to how it felt to apply to college a little less than a decade ago. In America, the higher up the parental income bracket you go, the presumption is that you will go to college or fail to do so. Exceptions are extremely rare, motivated by something very specific, and even still often have lingering self doubt about the decision. So the primary motivation for the majority of students is to not be a failure, the second would be as a prerequisite to a good job, and a distant third would be curiosity. As for college itself, behold my subjective taxonomy of college students:

    1. The plurality of students get to college with neither a particular goal orientation nor a general curiosity, and drift into a major that is well placed for the white collar work of the day. The most determined a half decade ago would go into computer science or have been admitted as engineering students, after that would be economics or some kind of business school. They would make up the bulk of classes in their major requirements, span the range of performance but forget each subject as soon as the class is over even if they did extremely well, and phone it to the extreme in their general education requirements.

    2. Next would be a collection of pre-pre-professionals—people gunning for medical school or law school, also often not out of a particular interest but because it is a default, prestigious career path neatly already designed for them, guaranteed to pay well if they work hard enough. Whether one chooses law or medicine usually hinges on whether they felt more oriented towards STEM or the humanities. Their major is typically political science if they are pre law or biology if they are pre med. These would do very well in all the classes they took, though typically they will choose the easiest classes they can outside of what looks good on applications and if given the option to perform well without understanding many will take it. They are often extremely busy with some kind of extracurricular activity, again something that looks good on a resume.

    3. Next would be general drifters who want a degree, but are neither generally curious nor motivated by any career path. They pick a major essentially at random but pull away from the ones with a higher minimum difficulty (where I was, it was anything with calculus, English, history, philosophy, and foreign languages if you aren’t a native or heritage speaker). Performance depends essentially on how much they believe employers look at your GPA on a resume, and even if they do very well in a class, they will forget it the next semester.

    4. Next would be people who really want to be PhDs in a particular field. These usually excel in their major requirements, and do their best to really understand what’s happening. Increasingly, to remain competitive with other applicants, they are taking PhD classes in their field as undergraduates. Where I was, at this point the top 10 or so students of this kind probably have a full first year grad student courseload done by the time they graduate. They get involved in research or other work for a faculty member as soon as they can. For general education outside what’s necessary for research work or that signals to a graduate program, they will usually phone it in. If a

    5. Finally would be the generally curious without a plan, who commit to doing well, understanding, and remembering the content of all their classes, and don’t hesitate to take classes that challenge themselves. They might drift into a plan, but will continue to take classes outside of that which interest them and commit effort towards those.

    Now, for some subjective interpretations on top of subjective observations:

    It’s very disheartening to me how little the top performing students on the academic track care about their general education, but also, what else can we expect? If a good applicant to a physics program these days has completed a first year of PhD physics, PhD classes in topology, algebraic geometry, and partial differential equations, and has 2 or 3 first author publications in, can we really expect them to have cared about lofty goals in “education?” From a research productivity perspective though, does this race for more and more impressive students at the top end translate to faster progress? Is it worth the trade off in risk taking behaviors and cross pollination?

    At an elite university, there’s not a whole lot of separation in the skill of kids going in just by the nature of selection. There’s different areas of proficiency of course; some kids are better with numbers, some are better writers, etc. But by the highly competitive, numbers and transcript driven nature of college admissions, and the highly structured nature of K12 education, almost nobody was truly terrible at any one subject going in. Once in the elite university, everyone pulls apart. A studenf who struggled to ace AP computer science but was seeking a PhD pulled way ahead of peers who breezed past it in high school but were in it for any decent job. Many students would actually regress, especially outside of their major. Engineering students’ essay writing deteriorated to the point where many of them truly were terrible. While vast majority of my peers had aced AP Calculus in high school, by the second year of college maybe a third of them could differentiate a power law.

    It’s tempting to praise the generally curious, but in the sense of survivability in the real world they are idiots. They don’t have a plan that requires difficult classes or that they do well at it, but commit hard to it anyways. If their grades have some kind of signaling value, they’re risking that by challenging themselves, and if not, they’re stressing themselves out for no reason.

    In the spirit of remaining nominally on topic, Bryan Caplan is a contrarian attention seeking baby, incapable of self reflection, and is rightfully not taken seriously.

    “Hmm, I believe A, but A sounds too boring. Instead, I’ll state EXTREME A, which has the same flavor as defensible claim A, but is more fun to say and will attract more attention.”

    “Wow, look how mad people are at me, even though what I said was *essentially* A, a perfectly defensible claim. Obviously, they should all know I meant A! It must be because I’m such an independent thinker, willing to cross societal taboos.”

    • Two last observations:

      1. If I went back to college as an adult for another bachelor’s degree in another subject, I’d feel pretty bad about it. I don’t think it’s all just in my head. College is considered to be a stage of life, and both societal valuations of its worth and the design of the system itself make the strong assumption that it’s the 18-24 year old stage of life.

      2. While I do feel like most college graduates could do the jobs they’re doing without their college courses’ specific subject matter, that doesn’t necessarily mean they could have done it without college. College can be just an incubation period for people’s general development before they’re ready for to enter the white collar workforce, with some extra stimulus in a consequence free environment. Collaborative activities to improve the emotional development, some frequent reading and writing to mature those general abilities or at least certify them, some time living apart from their parents to make sure they can cohabitate with other adult humans and understand the basics of buying and selling and laws and rent. I personally don’t buy this argument, but it’s there.

    • I like somebody’s taxonomy I think it still pretty much applies.

      I’d say that it’s easier to praise the motivated curious student when they’ve chosen a cheaper path. To go to a couple years of city college then transfer to a govt subsidized 4 yr institution and then on to a master’s degree graduationg with very little debt makes a lot of sense. Hitting Reed college at $50k/yr while browsing challenging courses not really leading to any clear major would be a stupid path.

  17. I think of methodological criticism as having three stages:
    1) Are you asking the right question?
    2) Does your analysis set you up to answer that question?
    3) Did you do the analysis correctly?

    Most people will primarily engage with steps 1 and 2 because they’re more significant and also much more straightforward to inspect (and a good chunk of ‘hot new conclusion’/grandiose ‘societal’ research fails significantly in those first two stages). Nobody wants to spend a week looking at spreadsheets to validate the end result of some analysis that is flawed to begin with.

    Too often the defense against step 1/2 issues is ‘this is the data we have access to’ or ‘this is how we have to run the experiment/analysis’, which is fine but does mean that you shouldn’t be offended when nobody cares about your conclusions or trusts them to answer the question you really wanted to answer.

Leave a Reply

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