Skip to content

David Brooks writes that technical knowledge—“the statistical knowledge you need to understand what market researchers do, the biological knowledge you need to grasp the basics of what nurses do”—can be “memorized by rote”

The popular New York Times columnist writes:

The best part of the rise of online education is that it forces us to ask: What is a university for? . . .

My own stab at an answer would be that universities are places where young people acquire two sorts of knowledge, what the philosopher Michael Oakeshott called technical knowledge and practical knowledge. Technical knowledge is the sort of knowledge you need to understand a task — the statistical knowledge you need to understand what market researchers do, the biological knowledge you need to grasp the basics of what nurses do.

Technical knowledge is like the recipes in a cookbook. It is formulas telling you roughly what is to be done. It is reducible to rules and directions. It’s the sort of knowledge that can be captured in lectures and bullet points and memorized by rote.

Brooks is citing Oakeshott, whom I haven’t read, so let me be clear here that I’m reacting here not to the original source but to the above characterization, “the statistical knowledge you need to understand what market researchers do,” etc.

In all seriousness, I think statistics and biology are like many other skills, such speaking a foreign language, playing a musical instrument, or hitting a softball: they are hard to learn, but if you put in the time you can be ok at it. To say that “memorizing by rote” is the way people learn statistics, biology, languages, music, or athletic skills—that’s just stupid. There are some things you do need to memorize, but learning is about making connections and practice practice practice practice practice.

It’s hard for me to believe Brooks actually believes something so dumb. I think what happens is that he has a smooth writing style and the words just come out sounding right, and he never goes back to see if they make sense. When I write for the Times, an editor reads my articles and goes in and makes changes, but maybe Brooks is too much of a big shot for that or maybe his editor gives Brooks’s columns a non-literal read. I can see why it makes sense to give Brooks free rein with his prose, in that I was amused by this sentence of his: “Are universities mostly boot camps for adulthood, where young people learn how to drink moderately, fornicate meaningfully and hand things in on time?” That phrase “fornicate meaningfully” is just perfect.

One reason to suspect that Brooks does not believe that technical knowledge is “memorized by rote” is that he later writes that mooks are the solution: “as online education becomes more pervasive, universities can no longer primarily be in the business of transmitting technical knowledge. Online offerings from distant, star professors will just be too efficient.” But if you’re just memorizing, why the need for star professors? Anyone can read from a script and give you things to memorize.

Instead, Brooks wants universities to teach “the wisdom a great chef possesses that cannot be found in recipe books. Practical knowledge is not the sort of knowledge that can be taught and memorized; it can only be imparted and absorbed. It is not reducible to rules; it only exists in practice.” That’s fine too. Most of our students are not going to be great chefs or even great statisticians, but we’d like them to be the best they can be at what they do. But I don’t think they’re gonna get there if they think their technical knowledge is something to be memorized by rote.

P.S. Also, what’s with the slam on “bullet points”? Suppose I have four sentences to convey. In some contexts, they can work best in paragraph form, in other cases as a numbered list, in other cases as bullet points. I use all three, and I think Brooks is foolish to dismiss bullet points as a mode of communication. Just cos Dilbert’s boss gives bullet-ridden powerpoints, it doesn’t mean they’re always a bad idea.

P.P.S. This is a good place to link to a criticism by Jay Livingston of an earlier column where David Brooks tells just one side of the story (“Brooks’s tour did not include a stop to chat with Nechemaya Weberman. . . .”)

P.P.P.S. Please don’t take all this a criticism of Brooks. I recognize that he’s a busy man. He makes mistakes, but so do we all. To say that Brooks has written something stupid is not to say that he is stupid. I’ve written stupid things too. I just recommend that in the future he read his words more carefully, watching out for things that sound good but don’t make sense on closer inspection. I also recommend that he acknowledge the errors he’s published in the past. One reason for this is it will establish a personal incentive for him to be accurate. If he knows he will have to correct his errors and suffer the resulting (small) embarrassment, maybe he’ll be more motivated to avoid the errors in the first place.

P.P.P.P.S. I’ve added a paragraph above (“Brooks is citing Oakeshott, whom I haven’t read . . .”) to be clear that, when using the term “technical knowledge,” Brooks is referring to a particular meaning of the phrase. I still don’t think that “memorizing by rote” has anything much to do with understanding what market researchers do, etc., but I was mistaken in criticizing Brooks’s use of the term without making this clarification.


  1. Rahul says:

    @Andrew: One suggestion is that you should ease off critiquing snippets and paragraphs and try to tackle whole articles. It is almost impossible to write a piece where every sentence can stand strong on its own, context free, immune from criticism.

    • Andrew says:


      I’m criticizing the whole article. As far as I can tell, Brooks really is saying what I am saying that he’s saying.

      I just quoted some of Brooks’s article because that’s what seemed relevant to the discussion here. I thought that was standard when quoting. But I could quote the whole thing. . . .

      P.S. I quoted the whole thing, then someone else said that might be violating some copyright, so I thought I better not do that! I compromised by adding a bit to the quotes in my post above.

      • Anonymous says:

        Such a long quote is not a breach of copyright?

      • Rahul says:

        Sorry! I didn’t mean to ask you to quote the whole thing.

        I just assumed you were critiquing the part you quoted. If you are indeed criticizing the whole article its a different matter. To me the article seemed much better than that part you quoted.

        • Andrew says:


          I agree that there are some reasonable aspects of Brooks’s article. Certainly the article’s average level of quality is higher than its lowest level, and of course it is the lowest level that bothered me.

  2. Seems to me like he got it backwards. Technical knowledge is more or less a holistic set of connections about a topic, things like what it means to think statistically, how does probability work, what does inference mean, why do we need “representativeness” and how can randomization help us with achieving it… etc. You can’t memorize this, because you’re learning logical connections and how to reason in a certain way.

    Practical knowledge is things like what commands do I need to invoke in R in order to actually get numerical results, plot graphs, do the nitty gritty. This stuff can be more or less memorized to a large degree. In fact much of it has to be, since it’s arbitrary and determined by the specifics of a program.

    From the standpoint of the food analogy, what a great chef does isn’t fiddly bits of cooking that no-one else can do, in fact that’s what sous chefs are for, what a great chef does is put together ingredients in quantities that are destined to taste great, and it’s more or less reasoning holistically about the process that’s essential. You don’t learn to be a great chef by rote memorization of some other chef’s cookbook, you do it be experimenting and building a model in your head of what is more or less likely to work together. part of that experimentation might be to repeat others recipes, but if you do it purely by rote, without thinking about what makes the recipe great or how it could be altered to give alternatives, then you’re pretty much just learning the recipe and not the process of being a chef.

  3. Stuart Buck says:

    I’ll defend Brooks, or at least a version of what he could have meant:

    1) Technical knowledge in statistics would be memorizing the formulas for the means of various distributions, the standard formula for beta in an OLS regression, etc., etc., etc.

    2) Practical knowledge in statistics would be the art of putting the above formulas to work in a thoughtful way — coming up with a meaningful model of a real dataset, for example. For this, you do need to learn how to make connections, practice, etc.

    Is that a useful distinction in thinking about what is possible through a MOOC vs. what’s not possible?

    • Andrew says:


      I think a mook (or any other sort of course) that tried to teach statistics via the rote memorization of formulas would be pretty close to useless.

    • Steve Sailer says:

      My impression is that a good statistician is good at both recipe-following and at pattern-recognition, and those two skills are not all that highly correlated. I suspect that a lot undergrad students who have a knack for noticing patterns in society often struggle with the recipe-following aspects of introductory courses in statistics and thus lose interest in the field, while students who are good at recipe-following are less good at coming up with interesting hypotheses to apply their technical skills to.

      In my case, I did fine in the introductory and midlevel statistics courses that concentrated on recipe-following, but I didn’t find them terribly interesting. Finally, when I got to the most advanced statistics courses offered by UCLA’s B-School, I find them utterly fascinating because I could finally apply my pattern recognition skills to new problems in the real world.

  4. Brian says:

    Isn’t the idea that science is defined by rules a fairly old one? It’s long been one of the central concepts of methodology – the idea that in mature sciences, there are generally agreed upon rules for designing experiments, interpreting data, evaluating theories, and so on.

    Popper seemed to subscribe to it (see Latakos’ famous critique of Popper’s account in “The role of crucial experiments in science”)

    This is a fairly old school view now (outside of positivist philosophers). But I’m not sure it’s fair to ridicule it – although Brooks is probably giving a fairly crude account of it.

  5. John says:

    Speaking of snippets, I’m here because of your gross lumping of Boston’s population into a “Creative Class,” a “Service Class,” and a “Working Class.” Your characterizations seem to be making severe value judgements about relative wealth, with people (like you) being “creative,” and people (not like you) being “non-creative” worker bees.

    Do you actually know people in these subservient classes? Are they friends? What if some of them, apart from their low-wage day jobs, are creative artists and writers? What if, given actual opportunities, many of them could and would become business owners?

  6. Gary says:

    We should get Brooks a “Got Reification?” t-shirt.

  7. Wayne says:

    I think he’s trying to say,

    “A knowledge of sampling, regression, stratifying, matching, and other techniques, helps you to understand the tools a market researcher employs. It tells you the kinds of specific actions the researcher will do during a day. It is the ‘what’ of a market researcher’s activities, which Oakeshott calls ‘technical knowledge’.

    “But how the market researcher attacks a problem, how and when he uses these tools and how he interprets their results is what Oakeshott calls ‘practical knowledge’. It is the ‘when’, ‘how’, and ‘why’ that take much more effort to grasp than the ‘what’.

    “Technical knowledge is ‘technique-al’ knowledge: knowing how to implement various techniques. ‘Practical knowledge’ is ‘practice-al’ knowledge: knowing what to look for, what techniques are applicable, and how to interpret what your techniques reveal. Techniques can be memorized, practical knowledge — some would call it “theory” — is slowly acquired by working with people who have already acquired a practical knowledge and by purposeful practice.”

    Technical knowledge is relatively simple, and teaching assistants or wiki pages can disseminate it. It can be disseminated to mass audiences, and requires little feedback. But practical knowledge is valuable and requires a craftsman and a dedicated apprentice to transfer.”

    Or something like that.

  8. Professor of Etiquette says:

    Brooks wants colleges to become finishing schools and places to learn tact, manners, and earnest-seeming postures. As a Professor of etiquette, I couldn’t agree more.

  9. I’m pretty unhappy with both Brooks and Gelman in this discussion.

    One way of characterizing the (questionable) dichotomy Brooks set up is to see it as the difference between a liberal arts and a vocational education. This is why Brooks, justifiably, chooses the word technical, as in technique, which is a sort of a tool and not a deep comprehension. Certainly, to effectively utilize many tools requires something outside pure technique, something that is at minimum moderate comprehension and often deep comprehension. But not always. In many contexts, not often. There is a difference between being an organic chemist and an auto mechanic, although both are steeped in technique.

    Brooks goes one way, Gelman the other. Brooks thinks that universities these days are for most students vocational schools, imparting mostly techniques they don’t really comprehend but will utilize effectively in their careers.

    Meanwhile, Gelam thinks that universities, insofar as they are preparing students vocationally in teaching them techniques, are necessarily imparting deep comprehension.

    Both are wrong. Anyone outside academia will tell you that in the vast majority of vocations for which four-year universities have supposedly trained their students, students enter the workforce with only a small facility with a limited portion of the techniques they’ll be using on a daily basis. As vocational education, four-year university education is in most cases, not.

    On the other hand, academics like Gelman are quite likely to tell us that deep comprehension is what happens in grad school relative to undergraduate education. In any of the sciences or similar vocations requiring post-graduate education, it’s well-known that students enter graduate school with mostly a set of rudimentary skills and knowledge with for the most part no deep or strong connections between them. Half the discussion here include some grousing about undergraduate statistics coursework and its inadequacies specifically within the context of teaching technique with little comprehension.

    I have been dumbfounded by much of the discussion I’ve read lately in response to online initiatives — I guess because I read a lot of academics, who otherwise are far above average in intellectual integrity, suddenly I find justifying the necessity of a lecture hall packed with 300 students because of the supposed quality education that occurs there.

    I’m suspicious of these online education initiatives for many of the same reasons as so many others (ostensibly) are, but let’s not pretend that the average graduate of a four-year state university has anything even remotely similar to a scholarly comprehension of their field of study. Likewise, let’s not pretend that they’ve been adequately trained vocationally, either, because they haven’t been. It’s worth noting, as a retort to Brooks’s argument, that it’s not those areas where American post-secondary is widely recognized to succeed in vocational training, those technical two-year college programs, where there’s a drive to online coursework. It’s the crowded lecture halls for survey courses satisfying core requirements.

    The status quo of four year undergraduate education in the US is a weird aspirational, egalitarian fantasy of some strange melding of liberal arts and vocational training that, sadly, does a piss-poor job of both. It’s aspirational in that whatever it’s worth qua education, it’s increasingly requisite for everyone because we don’t want to leave anyone “behind”. It’s egalitarian in that it refuses to choose between vocational and liberal arts education because other educational structures used (and many still use) such a division to enforce class distinction and reduce mobility.

    And maybe these are worthwhile goals to continue to pursue.

    But the cultural necessity to presume they’re worthwhile and justify the status quo has included a willful blindness to the failures of the typical undergraduate education. Both Brooks and Gelman are weirdly counterfactual or utopian in either asserting that it’s already doing a fine job at technical education and this can be safely split from the deeper education, leaving more room to focus on the latter (for everyone!); or that, really, all contemporary four-year education is already deep education.

    • Andrew says:


      You write, “Gelman thinks that universities, insofar as they are preparing students vocationally in teaching them techniques, are necessarily imparting deep comprehension.”

      I’ve never said that, nor did I write it, do I think it. I don’t think we do a lot of rote memorization. That does not imply that we are imparting deep comprehension (or for that matter that we are imparting anything at all). Sometimes we do, sometimes we don’t.

      • Okay, on a re-read, I can see that.

        But then I’m at a bit of a loss because it’s clear that you’re very strictly defining rote memorization. Which is fine, but I think it’s a misreading of Brooks because he’s using it more broadly.

        If you’re not asserting that the status quo four-year education is imparting the sort of comprehension that Brooks says it ought to be imparting, then I don’t much have an argument with you. (Except insofar as you’re implicitly supporting the status quo.) My beef, then, is more with Brooks — who seems to get almost everything wrong he could possibly get wrong.

  10. lemmy caution says:

    The idea that you can’t learn to cook from a cookbook is pretty silly.

    Here is a summary of Oakeshott on technical and practical knowledge:

    A similar distinction is now made between explicit and tacit knowledge:

    That distinction is from Michael Polanyi in the 1950s. Oakeshott was earlier and his distinction is more politicized. Practical knowledge includes the traditional unarticulated reasons for things. Oakeshott is explaining why you should not change things too much even if you can come up with good technical reasons to change things.

    This comes up in the gay marriage debate. there is no good technical reasons against gay marriage, but conservatives still fear the consequences since it goes against the practical knowledge embedded in tradition.

    There is also a similar left wing distinction made between contemplative knowledge and active knowledge that was popular in marxist theorizing:

    • April Galyardt says:


      The choice of the words “technical” and “practical” are somewhat poor word choices, and as lemmy caution points out those words have special meaning within the community they came from. The same way that “significant” has a special meaning in statistics that it does not have in plain english.

      It’s better to think about “codifiable” and “uncodifiable” Codifiable knowledge can be written down as Brooks states, like a recipe in a cookbook. If we can write it down, or summarize it easily, then it’s codifiable. Uncodifiable knowledge is stuff that’s exactly that, like social norms. It’s the stuff that you learn only by being part of a community, and participating in the activities of the community.

      The biggest thing that seems to have upset Gelman here is that Brooks put statistics in the codifiable category. A large part of it is. How to do a t-test is codifiable. How to chose between variable selection methods is less so. How to design a good data visualization is even more uncodifiable, we can critique what we like and don’t like about certain displays, but it’s hard to write down how to make a good one. Believing statistics falls completely in the codifiable category is a mistake I’m willing to chalk up to public ignorance of our discipline (a completely different and more serious problem).

      Uncodifiable knowledge is almost always transmitted through community interactions. And that makes Brooks right. The value that an on-campus experience offers over a MOOC is that it provides an opportunity to learn the uncodifiable as well as the codifiable.

      • Andrew says:


        Brooks wrote that “the statistical knowledge you need to understand what market researchers do, the biological knowledge you need to grasp the basics of what nurses do” can be “memorized by rote.” I don’t buy it. Knowing the formula for the t-test does not give you much understanding of what market researchers do. I don’t know much about nursing but I can’t imagine that rote memorization will help you grasp the basics of what nurses do.

        In either case, I imagine students would learn a bit by watching videos of good practice. This isn’t rote learning either, but it could be done in a mook. (I don’t know that mooks on statistics have this sort of video, but it would be possible.)

  11. EvanZ says:

    Knowing how to use the lm() function in R does not make you a statistician, although maybe the case is that many non-statisticians, including David Brooks, think that it does?

  12. jonathan says:

    If you accept the terminology, which I would do only provisionally not because I agree, then I can give some examples. One of my kids went to high school for a term in a high-end Chinese school – for Chinese kids, not expats or visitors. Math was taught at roughly the higher BC track calculus, but was done by rote not as a series of problem solving exercises. You learned whatever the material was and a set of applications in specific problems and repeated those back. This was not a way to learn deeply, to learn how the systems work, but it had the advantage of teaching a relatively large number of kids the nuts and bolts of calculus.

    I can argue the Chinese method is better or worse. It is better at teaching more people a general facility. It is worse at teaching underlying understanding. If that is what technical and practical means, there is some validity in the basic idea.

  13. MikeM says:

    David Brooks is just being David Brooks. He often uses the general “we” to describe what Very Serious People believe, in this case about university teaching and learning. In statistical terms, he talks about the mean (or what VSPs believe it to be) as applying to everyone in the distribution rather than noting (and celebrating) variation. No two classes I taught (mostly statistics, in a social science curriculum) ever was run in the same way; each had a different mix of students, and my goal was to reach them in whatever way they found most congenial.

  14. Chris G says:

    > Please don’t take all this a criticism of Brooks. I recognize that he’s a busy man. He makes mistakes, but so do we all. To say that Brooks has written something stupid is not to say that he is stupid. I’ve written stupid things too.

    No. “He’s a busy man.” does not get him a pass. Writing two columns a week that don’t suck is his job. When he puts forth a column that sounds good but doesn’t make sense upon inspection then he’s screwing up at his job. That is not okay. If you have a job then endeavor to do it well. Yes, everybody makes mistakes. That does not mean that mistakes are “okay”. They acceptable to the extent that they are survivable and that we learn from them so as not to make them again. Brooks scores low on the latter account.

    > I just recommend that in the future he read his words more carefully, watching out for things that sound good but don’t make sense on closer inspection. I also recommend that he acknowledge the errors he’s published in the past.

    Yeah, about that “things that sound good but don’t make sense on closer inspection” one might say that Brooks’ brand. See Sasha Issenberg’s piece on him, Boos-Boos in Paradise, from way back in 2004:

  15. ceolaf says:

    I think you are missing where Brooks’ idea is really coming from.

    When many students take technical courses (i.e. also known as “math,” “statistics” or some sort of science) they memorize (and also history and various other elements of liberal arts). Memorization is one of the chief skills of the high achieving student in high school, and far far too many students in college.

    Brooks’ exposure to learning so-called technical subjects, and perhaps some non-technical subjects, has showed him that memorization is a success strategy.

    He’s right.

    Students can get VERY far with memorization and playing the game. Do the problem set and hand it in on time, and you’re doing ok. Go to TA hours when you have a question and apply the lesson algorithmically.

    This is a failure of our educational system, yes. But it is also a failure of each and every student whose idea of success is a high grade. How many students would prefer a higher grade to increased learning? Almost all of them, right?

    We’ve all had students who, “just want to know what I’m supposed to do.” They want the algorithm, and the rule for when to apply it.

    Read Brooks’ column as evidence for how HE approached these subjects, and how HE thinks college operates.

    What we need to move away from the acheivement and attainment models of success in education to a learning-based model.

    What is Brooks missing? Well, he so little understand what the technical subjects are about that he doesn’t even realize that there is more to success there than memorizing algorithms and rules of thumb about their application.