I think the question about why books are so long is partly answered by Saffran's initial observation about "clock time and calendar time." Because a book is long, you sit with the info for longer than if it were a pamphlet, so you absorb more of it than if it were a pamphlet, even though you still don't absorb that much of it. The extra padding basically acts to increase retention of the most important bits.
The other component is that which parts are the "most important bits" will be different for different people.
1. I read almost all of your posts in full, because your essays are concise and rich in content. 2. As one who has written a dissertation, I suspect that the process of dissertation writing—in which every contingency must be covered, even the slightest influence must be cited, and at least one picky professor must be satisfied—is the cause of much nonfiction bloat. 3. A 30-page pamphlet is too long. The modern analogue is the online essay, whether in Substack or a news outlet. It is not infrequent that a title or headline inspires a click, after which I have to scan miles of blather before the essayist finally gets to the point. I appreciate essays that begin with an outline of the main points and get right to them.
I think the key thing with live classes is that you can stop the professor and ask a question, follow up on the answer, and do so in idiosyncratic ways. For myself, I always found the questions I had were never in a list of the Top 10 Common Questions, so it was very difficult for pre-recorded questions to cover them, and usually the common ones were ones I already knew the answer to.
Likewise, in person classes allow students to talk to the professor after class and ask more follow up questions, which might be where most of the learning really happens.
I agree too that the ability to read the room, see where students are all giving confused looks, or getting bored, etc. is a huge benefit.
I think your “key thing” may well be key in advanced undergrad classes. Especially for the more advanced students.
But it can’t be the answer for K-12 nor most undergrad intro classes, where that rarely happens. And it also can’t be the answer for most students, since almost by definition it is only a small minority who ask uncommon questions.
Separately, it seems to me that if you are correct, AI would be a great teaching mechanism, since as AK’s project shows, different people can ask uncommon questions and it scales and is efficient in that regard.
I disagree on the notion that it doesn't work for undergrad intro classes; that is what I taught the most, and students do get into things if they can (<40 students in a class, a teacher who will go off on digressions from script). Often what counts as an uncommon question is just seeing it from a slightly different angle, or having it related to something you understand in particular. I didn't intend to imply that it required being particularly clever or insightful, just that one is missing a different piece of the puzzle and needs that specific question answered. It requires a teacher that is willing and able to think on their feet and not be a slave to their slide deck.
Which makes me wonder about the K-12 aspect. I have never taught that level long term, but I wouldn't be surprised to find it requires a bit more ability to think on one's feet and relate to oddball questions since your class is 1: less selected for academic ability, and 2: has a rather less complete or more randomly incomplete set of knowledge to relate new things to. I find with my kids that it is sometimes surprising to realize what they don't know (or know that isn't so) so it feels harder to be able to anticipate their questions, and requires a lot more talking through ideas to make sure they really get it.
Regarding AI, I think there is a strong use case there, if the AI actually can reason along with students. The really difficult of dealing with questions in class and getting the student to understand is often just figuring out what the question is even about and why they don't understand, e.g. empathizing and reasoning through from their viewpoint and figuring out what is missing. That is very hard for most teachers, and why, I think, most teachers hate going off script and actually engaging in conversation in class. I am not sure if AI can do that well, either. In theory it will at some point, but I am not sure it can now.
It is also worth noting that there is probably a combination of "the teacher cares about me and wants me to do well" and "the teacher cares about the subject and is excited about it" that helps, too. AI can probably fake that better than most teachers, I suspect.
I said I agreed with you for the most part, then spent a lot of time disagreeing. I agree on the AI part, and I agree with the implicit suggestion that most teachers are bad at fostering the discussion aspect, and many students don't take advantage, which is why I think many students don't learn much of anything for all time they spend in class. I just also think they don't learn a lot when forced to sit and watch videos or read books on a subject. If they are motivated to learn that makes a difference, and often people seek out books and videos because they are motivated themselves, but a good teacher can help motivate you, too.
A rarely explicit, IMO quite important part of the discussion is whose outcomes/education we are most trying to improve.
If it’s the “No Child Left Behind” implicit students nearest but not quite at the bottom, I’m with you that literally very little can be done to significantly improve outcomes.
I’m totally with you and your approach if one is talking about the top 10% / those who are motivated / have “tiger moms”.
But personally for K-12 I’m more interested in what helps the students in the middle [30th to 70th percentile]. With undergrad I’m most interested in what helps the upper middle [~50% - 80% percentile].
[IMO those unmotivated we shouldn’t be optimizing for in K-12. And those not genuinely interested we shouldn’t be wasting any time on in undergrad, but instead should actively be seeking to weed out.]
In both cases the people who are *somewhat* motivated but not necessarily hugely so [although some near the bottom of my range in K-12 might have only minimum motivation].
So my point was not so much that a good teacher in front of a class "doesn't work" for large intro classes as that it materially benefits only a relatively small percentage of the students. And of course the smaller the intro class I readily concede the greater can be the value.
It just seems to me the AK-style AI approach could be hugely valuable for both groups I cite, and arguably better than even a very good teacher in front of a class for at least 75% of the hours the student spends / required for learning. And certainly far more cost effective.
I suspect we are very much in agreement re: AI and screens v. good teachers for the bottom and the very top.
I dunno, I think for the middle deciles of the K-12 range a good teacher is going to be better than screens except perhaps for those topics the individual kid is intrinsically motivated for. If we can’t scale good teachers (which I am not certain we can, but we also hardly seem to try) then yea, really good videos/AI might be the way to go. I am hopeful that AI will turnout to be really good here, but worried it will be used more than it should as a least cost but trendy option.
Re: your last point, I have no prediction whatsoever on what *will* happen, especially in the next few years.
And when I said K-12 I really meant 5/6-12.
To be clear on where we disagree, I'm suggesting a hybrid approach augmenting the best K-12 teachers where ~75% of the student's time is with the AIs and ~25% with said very good teacher as more optimal for those middle deciles.
Of course, this happens to fit my classical liberal model of fewer teachers who are more highly productive and better paid, so it could be a blind spot.
And having said I have no prediction of what will happen over the next few years, I surely would NOT predict that my preferred outcome will happen in that time period...
I can definitely see a strong possibility for value in students working with AI for instant feedback and learning, then with a teacher for more specific help or guidance.
I suppose it is also worth mentioning that we don’t really seem to know how much learning we can even expect from kids. The answer does seem to be “more than they do now” but part of me does think that we might be ignoring a severe trade off where the 60th percentile kid is really not going to learn much compared to the 90th percentile, so we are wasting his time with very general education and need to focus more. I definitely think that is true for college students, but maybe by 8th grade a lot of students are really maxed out on general knowledge. Maybe that is why the null hypothesis is so strong; not matter how good you are you can’t fill a 5 gallon bucket with 10 gallons of water.
Andrew Ng's Machine Learning course was better than a number of the University courses that I did. It cost nothing. I paid for the certificate at the end because it was just so good.
The one thing that online courses often don't have is really good assessment. Proctored exams are still necessary.
In Australia and the UK some Universities offer courses online with proctored exams that are taken in exam halls with 'proctoring as a service'. They also need hard exams that discriminate between students on how good they are at a subject.
If well done these can be as good as University courses.
Unfortunately there seems little push from Australian Universities to deliver these kinds of courses even though they can provide most of the value of a University course at a fraction of the cost.
They are not as prestigious and don't bring in as much money as getting students to come in.
One more point in favor of books: I have recently been reading a couple of books on the Shakespeare authorship question, books championing the Oxfordian view. The case cannot be made in less than a book, because it consists in countless small points, each a mere tidbit of information: here a fact about the Earl of Oxford that makes it more likely than you would have thought that he wrote something attributed to “Shakespeare,” there a fact about the man from Stratford-on-Avon that makes him less likely; plus an assemblage of minor remarks or actions by contemporary figures that slightly favor Oxfordism over Stratfordism. The case for Oxford consists in thousands of these tidbits, no one of which carries *much* weight. It requires a book.
I think that is an excellent point. Many modern non-fiction books seem to have one or two points couched in a whole lot of fluff (and some evidence sometimes), but comparing that to Wealth of Nations or Theory of Moral Sentiments is like comparing an old coffee can with a single bean rattling around with a 50 lb sack of beans bursting at the seams. Not just many small points, but also many, many more insights supported by many small points. Granted, not all old books are like that; the incentives to pack fluff when you are paid by the word are immense. Still, I can't help but feel that the modern scholar feels the need to have a book with their name on their shelf, even if the book could really have been a 30 page pamphlet, or a long Substack article or two.
I had a rare experience recently, reading a book that had practically not a wasted word - each chapter was that pamphlet, in effect. In fact, the text was double-spaced, a relief to my old eyes. The author recounted some familiar things, in brief, but in a new manner/with a purpose that became evident as he built to his conclusion. So key were the earlier chapters as building blocks, that I went back and actually re-read several times when I got to the “As I described in Chapter 4”, etc.
The only passage at which I felt a little impatience was a digression about Dante, and that only 3 or four pages, easily forgiven especially since the author is Italian.
The subject was quantum gravity, which did not really exist and certainly wasn’t yet in the textbooks when I was in school.
The best part is the book was so persuasive it makes a whole lot of other books obsolete, for me, which is good because those books were much harder to understand. Supersymmetry, goodbye, it was good to know ya except I really didn’t and was never going to be able to.
The book is “Reality Is Not What It Seems” by Carlo Rovelli.
As for AI, I can honestly say that in the course of an indifferent public school education (not my fault), as well as a poorly conceived/chosen college curriculum (my fault) - I had my share of stupid teachers, but I would not choose cringe -inducing AI pablum over the least of them.
That's one hell of an endorsement for a book. Have you encountered the work of Joscha Bach? Reading the Amazon reviews, and then, yes, asking ChatGPT how Rovelli fits into an exploration I've been conducting with it regarding Bach, Wolfram and the views of computationalism and digital physics makes me think you would get a lot from engaging with their work.
At least for K-12 math, the combination of Socratic dialogue, gamification, and mastery learning with spaced repetition seems to make screen learning pretty damn effective for some kids. My anecdata on this is my son's positive experience with the Art of Problem Solving (and before that Beast Academy) self paced online courses, and my looking over his shoulder to get a sense of how they work. But it may not generalize to most kids or most subjects.
I read multiple parts in a row, but IIRC I didn’t go through it all in any one period.
I bought it for my daughter, who was reluctant to take an econ class at her very liberal arts college, but agreed to read this book when I “strongly” suggested she must do one or the other…
Re: screen learning. My company had excellent remote meeting capabilities well before COVID, but I insisted all meetings be in person for two reasons:
1. I found the folks who were just on a screen would be regularly and consistently not paying attention, whereas someone not paying attention in a room with other people gets noticed and easily called out.
2. I found that the side discussions after the meeting were invaluable. Too many of my staff wouldn't seek each other out, but since there were already there together, there were plenty of "oh, I've been meaning to talk to you about..."
I've seen this with my high school aged kids as well. If they can skip paying attention because the lecture is on a screen, they will (even with their cameras on, it's easy to not pay attention). And when I ask them what they learned, it's usually something that involved their classmates.
So if I were overhauling college education, I'd try to get rid of as many of the multi-hundred person lectures classes as possible (where it is still possible to fall asleep in the back row without being chastised for it) and add as many 8-15 person in person discussion group classes as I could get away with.
1. Screen learning is a "thin" experience, leading one to be easily distracted, versus a "rich" (and respectful) classroom experience that has required resource expenditure to attend.
2. Serious reading requires an engaged brain or "flow" experience, versus the pablum of legacy media.
3. Vygotsky's zone of proximal development, which effectively promotes "flow" experiences, plays a role in here somewhere. For example, the older one gets, the more one seems to appreciate history, which happily tends to attract good writers/story-tellers.
Hi Dr. Kling. I suspect there are still live classes because tradition is powerful--students who go to college still expect to be in a room listening to a lecture. Also, having just written a book on modern approaches to learning and development, I have come to know that many college professors today employ a more interactive approach than in the past. They invite students to own their own learning far more than previous days.
I think it might be useful to differentiate between classes that focus on knowledge transfer and classes that use engagement and involvement tactics to optimize learning. The former can be recorded and replayed, as you suggest. The latter benefits from having a learned practitioner in the facilitator's seat.
Why do we read books? It strikes me that much depends on your purpose for reading. If all you want is the "key insights" then a 30-page summary might be all you need. However, the key insights I'm seeking might be different from yours, or from the conventional wisdom that an LLM will generate. If, instead, you're seeking to forge deep interconnections between the new material and what you already know, then 300 pages might not be enough. Similar considerations apply to gaining an understanding of historical nuance and detail, or to reading a classic text like The Wealth of Nations or Democracy in America in an exploratory fashion. With all that said, many newly-published books feel like padded articles to me, and I tend to read them using something like the Arnold Kling method: skip right to the conclusion, use the index to see if the book provides new insights on topics I'm already familiar with, etc.
I always learned most efficiently and completely by application of what I was reading. For me, it was a matter of simply doing every single problem a textbook included. This works very well for STEM textbooks of all kinds. Even in the age before the internet I didn't need the class lectures at all- not even a little bit. I suppose AI might have helped me in the study of subjects that are non-STEM such as literature, philosophy, history, and languages.
My prediction about the effect of AI on learning is that it will become a crutch rather than mechanized exo-skeleton for the vast majority of students.
Riffing on your oft-repeated "we choose what to believe by choosing who we believe", I would suggest books tend to be longer than absolutely necessary because the author has to convince the reader, who he knows is highly likely to be somebody unfamiliar with him, that he is somebody we should believe. That means a lot of otherwise extraneous information is included to demonstrate believability. A book, unless it is from an established author, doesn't have the institutional backing of a newspaper or other regularly published media that can provide a shortcut to establishing (or disproving, as the case may be) that believability. That doesn't mean authors can't also be disorganized, wordy, or poorly edited as well.
I don't think we'll get away from in-person face-to-face classes without super-immersible 3-D virtualization of ourselves as avatars. As you suggested, there is just too much information we communicate nonverbally from students to the instructor, among students, and of course from the teacher to make Zoom or any current chat platform a viable alternative.
I was going to suggest the same. The 10% we remember of a book is the conclusion. The 90% we forget is the evidence.
It's like a murder mystery. The conclusion and the clues leading to it are all that we remember. Well, there's the tone of the writing, whether we enjoyed the reading, but we don't remember that as the words themselves.
Two points on your wise questions raised in the essay:
“If we were to try to read ten 30-page pamphlets instead of one 300-page book, our minds would be overwhelmed and we would learn nothing.”
For analogy, scrolling and reading various essays on Substack is similar to reading the ten 30-page pamphlets. And perhaps the learning results might be alike. Or maybe not since a lot depends on reader comprehension and focus.
This brings me to second point about screen learning only, apart from human interaction. A small class, say less than ten humans as opposed to the large auditorium, seems more conducive and aligned with our evolutionary background for learning. A learned tutor (mentor) might suffice. But either of these likely will not pass financial muster in the current educational milieu.
"This raises the meta-learning question of why books are so long."
Depends on the subject and the time period.
My personal experience is that it's easy to be concise with people who are more or less friendly and open-minded and can be expected to have a sufficient intellectual foundation such that the "inferential distance" from their starting point to my claim is fairly low.
But it is impossible to be concise when facing a "hostile peer review" of critical adversaries always interpreting everything in as malicious a manner as possible, posturing as willfully obtuse, and looking to exploit every little mistake or poke any possible hole in every possible argument or point out any omission of detail. A lot of books - especially on controversial topics likely to get one in hot water unless phrased and supported just right - end up spending more time and effort on throat-clearing disclaimers and forestalling disagreement and larding things up with countless endnotes than they do on actually preparing and presenting the core of their content.
This is what happens in a modern adversarial trial, which is part of why trials are now so extremely burdensome and the records of cases are now so extremely long. The total length of text in all the concurrences and dissents in SCOTUS and appellate court opinions has got to be two orders of magnitude larger than it was a century ago. There is also the typical kind of too-clever word-twisting legal argument that has to be long by necessity because it is, at heart, completely incoherent nonsense. In the law it's easy to be coherent and brief and usually unnecessary to be coherent and crushingly comprehensive. You can often get away with incoherence buried under a mountain of confusing phrases, but it's hard to get away with the curtly incoherent. So, given that the law has been a total mess for a long time, you get a lot of LOOOONNNNNGGGG incoherent opinions.
Another reason books used to be long was because arguments were presented verbally instead of graphically, and a picture's worth a thousand words. Speaking of "The Wealth of Nations", the late P.J. O'Rourke read through the whole thing and in his attempt at summary made that exact observation, that to the modern impatient reader it was obvious that several pages of Smith's text could have been captured in a simple cartesian diagram or chart, but that was not easy given the publishing tech and costs of the time, and also, readers of that era were experienced in and used to the verbal, not so much the graphical, especially regarding topics distant from geometry and pure mathematics. NB: Internet comment tech also doesn't let me make arguments in pictures, so I'm forced to give you a thousand words, sorry.
For History, it's because history is really complicated and full of both mysteries and important details and and interactions, and adequately "setting the stage" for anything requires a lot of previous history. It is hard to narrow things down in a way that doesn't lose sight of the big picture and how all the puzzle pieces fit together, and it likewise hard for authors of history to avoid the procrustean exercise of forcing the messy facts to serve a narrative and fit into the presentation of a particular ideologically-flavored theme or lesson. You can read 50 huge and high quality books with mostly independent content about WWII and still get important new insights from the 51st.
On the other hand, some forms of communication have gotten lots shorter. It used to be something of a joke to point out the extreme duration of speeches given by current or aspiring Communist Dictators, but not too long before the Lenin-to-Castro era, comparably long political speeches and rounds of debates were not uncommon in the free western countries, where they were often well attended and with every word clung to by members of the general public in a way that Hartley's phrase "the past is a foreign country" does not even begin to express with adequate intensity. Most people probably think that change may have been for the best. On the other hand, those of us who watched it from the beginning have a feel for what the tendency toward more short text-messages and tweets did for the discourse, and in that regard I suspect that the average judgement about it is much less positive.
I think the question about why books are so long is partly answered by Saffran's initial observation about "clock time and calendar time." Because a book is long, you sit with the info for longer than if it were a pamphlet, so you absorb more of it than if it were a pamphlet, even though you still don't absorb that much of it. The extra padding basically acts to increase retention of the most important bits.
The other component is that which parts are the "most important bits" will be different for different people.
1. I read almost all of your posts in full, because your essays are concise and rich in content. 2. As one who has written a dissertation, I suspect that the process of dissertation writing—in which every contingency must be covered, even the slightest influence must be cited, and at least one picky professor must be satisfied—is the cause of much nonfiction bloat. 3. A 30-page pamphlet is too long. The modern analogue is the online essay, whether in Substack or a news outlet. It is not infrequent that a title or headline inspires a click, after which I have to scan miles of blather before the essayist finally gets to the point. I appreciate essays that begin with an outline of the main points and get right to them.
I think the key thing with live classes is that you can stop the professor and ask a question, follow up on the answer, and do so in idiosyncratic ways. For myself, I always found the questions I had were never in a list of the Top 10 Common Questions, so it was very difficult for pre-recorded questions to cover them, and usually the common ones were ones I already knew the answer to.
Likewise, in person classes allow students to talk to the professor after class and ask more follow up questions, which might be where most of the learning really happens.
I agree too that the ability to read the room, see where students are all giving confused looks, or getting bored, etc. is a huge benefit.
I think your “key thing” may well be key in advanced undergrad classes. Especially for the more advanced students.
But it can’t be the answer for K-12 nor most undergrad intro classes, where that rarely happens. And it also can’t be the answer for most students, since almost by definition it is only a small minority who ask uncommon questions.
Separately, it seems to me that if you are correct, AI would be a great teaching mechanism, since as AK’s project shows, different people can ask uncommon questions and it scales and is efficient in that regard.
I think I agree for the most part.
I disagree on the notion that it doesn't work for undergrad intro classes; that is what I taught the most, and students do get into things if they can (<40 students in a class, a teacher who will go off on digressions from script). Often what counts as an uncommon question is just seeing it from a slightly different angle, or having it related to something you understand in particular. I didn't intend to imply that it required being particularly clever or insightful, just that one is missing a different piece of the puzzle and needs that specific question answered. It requires a teacher that is willing and able to think on their feet and not be a slave to their slide deck.
Which makes me wonder about the K-12 aspect. I have never taught that level long term, but I wouldn't be surprised to find it requires a bit more ability to think on one's feet and relate to oddball questions since your class is 1: less selected for academic ability, and 2: has a rather less complete or more randomly incomplete set of knowledge to relate new things to. I find with my kids that it is sometimes surprising to realize what they don't know (or know that isn't so) so it feels harder to be able to anticipate their questions, and requires a lot more talking through ideas to make sure they really get it.
Regarding AI, I think there is a strong use case there, if the AI actually can reason along with students. The really difficult of dealing with questions in class and getting the student to understand is often just figuring out what the question is even about and why they don't understand, e.g. empathizing and reasoning through from their viewpoint and figuring out what is missing. That is very hard for most teachers, and why, I think, most teachers hate going off script and actually engaging in conversation in class. I am not sure if AI can do that well, either. In theory it will at some point, but I am not sure it can now.
It is also worth noting that there is probably a combination of "the teacher cares about me and wants me to do well" and "the teacher cares about the subject and is excited about it" that helps, too. AI can probably fake that better than most teachers, I suspect.
I said I agreed with you for the most part, then spent a lot of time disagreeing. I agree on the AI part, and I agree with the implicit suggestion that most teachers are bad at fostering the discussion aspect, and many students don't take advantage, which is why I think many students don't learn much of anything for all time they spend in class. I just also think they don't learn a lot when forced to sit and watch videos or read books on a subject. If they are motivated to learn that makes a difference, and often people seek out books and videos because they are motivated themselves, but a good teacher can help motivate you, too.
Thx.
We do indeed agree more than disagree.
A rarely explicit, IMO quite important part of the discussion is whose outcomes/education we are most trying to improve.
If it’s the “No Child Left Behind” implicit students nearest but not quite at the bottom, I’m with you that literally very little can be done to significantly improve outcomes.
I’m totally with you and your approach if one is talking about the top 10% / those who are motivated / have “tiger moms”.
But personally for K-12 I’m more interested in what helps the students in the middle [30th to 70th percentile]. With undergrad I’m most interested in what helps the upper middle [~50% - 80% percentile].
[IMO those unmotivated we shouldn’t be optimizing for in K-12. And those not genuinely interested we shouldn’t be wasting any time on in undergrad, but instead should actively be seeking to weed out.]
In both cases the people who are *somewhat* motivated but not necessarily hugely so [although some near the bottom of my range in K-12 might have only minimum motivation].
So my point was not so much that a good teacher in front of a class "doesn't work" for large intro classes as that it materially benefits only a relatively small percentage of the students. And of course the smaller the intro class I readily concede the greater can be the value.
It just seems to me the AK-style AI approach could be hugely valuable for both groups I cite, and arguably better than even a very good teacher in front of a class for at least 75% of the hours the student spends / required for learning. And certainly far more cost effective.
I suspect we are very much in agreement re: AI and screens v. good teachers for the bottom and the very top.
I dunno, I think for the middle deciles of the K-12 range a good teacher is going to be better than screens except perhaps for those topics the individual kid is intrinsically motivated for. If we can’t scale good teachers (which I am not certain we can, but we also hardly seem to try) then yea, really good videos/AI might be the way to go. I am hopeful that AI will turnout to be really good here, but worried it will be used more than it should as a least cost but trendy option.
Re: your last point, I have no prediction whatsoever on what *will* happen, especially in the next few years.
And when I said K-12 I really meant 5/6-12.
To be clear on where we disagree, I'm suggesting a hybrid approach augmenting the best K-12 teachers where ~75% of the student's time is with the AIs and ~25% with said very good teacher as more optimal for those middle deciles.
Of course, this happens to fit my classical liberal model of fewer teachers who are more highly productive and better paid, so it could be a blind spot.
And having said I have no prediction of what will happen over the next few years, I surely would NOT predict that my preferred outcome will happen in that time period...
Pessimists are rarely disappointed :)
I can definitely see a strong possibility for value in students working with AI for instant feedback and learning, then with a teacher for more specific help or guidance.
I suppose it is also worth mentioning that we don’t really seem to know how much learning we can even expect from kids. The answer does seem to be “more than they do now” but part of me does think that we might be ignoring a severe trade off where the 60th percentile kid is really not going to learn much compared to the 90th percentile, so we are wasting his time with very general education and need to focus more. I definitely think that is true for college students, but maybe by 8th grade a lot of students are really maxed out on general knowledge. Maybe that is why the null hypothesis is so strong; not matter how good you are you can’t fill a 5 gallon bucket with 10 gallons of water.
Have you done many online courses ?
Andrew Ng's Machine Learning course was better than a number of the University courses that I did. It cost nothing. I paid for the certificate at the end because it was just so good.
The one thing that online courses often don't have is really good assessment. Proctored exams are still necessary.
In Australia and the UK some Universities offer courses online with proctored exams that are taken in exam halls with 'proctoring as a service'. They also need hard exams that discriminate between students on how good they are at a subject.
If well done these can be as good as University courses.
Unfortunately there seems little push from Australian Universities to deliver these kinds of courses even though they can provide most of the value of a University course at a fraction of the cost.
They are not as prestigious and don't bring in as much money as getting students to come in.
One more point in favor of books: I have recently been reading a couple of books on the Shakespeare authorship question, books championing the Oxfordian view. The case cannot be made in less than a book, because it consists in countless small points, each a mere tidbit of information: here a fact about the Earl of Oxford that makes it more likely than you would have thought that he wrote something attributed to “Shakespeare,” there a fact about the man from Stratford-on-Avon that makes him less likely; plus an assemblage of minor remarks or actions by contemporary figures that slightly favor Oxfordism over Stratfordism. The case for Oxford consists in thousands of these tidbits, no one of which carries *much* weight. It requires a book.
(The main worry about such a book is that it may omit thousands of small points in the other direction.)
I think that is an excellent point. Many modern non-fiction books seem to have one or two points couched in a whole lot of fluff (and some evidence sometimes), but comparing that to Wealth of Nations or Theory of Moral Sentiments is like comparing an old coffee can with a single bean rattling around with a 50 lb sack of beans bursting at the seams. Not just many small points, but also many, many more insights supported by many small points. Granted, not all old books are like that; the incentives to pack fluff when you are paid by the word are immense. Still, I can't help but feel that the modern scholar feels the need to have a book with their name on their shelf, even if the book could really have been a 30 page pamphlet, or a long Substack article or two.
I had a rare experience recently, reading a book that had practically not a wasted word - each chapter was that pamphlet, in effect. In fact, the text was double-spaced, a relief to my old eyes. The author recounted some familiar things, in brief, but in a new manner/with a purpose that became evident as he built to his conclusion. So key were the earlier chapters as building blocks, that I went back and actually re-read several times when I got to the “As I described in Chapter 4”, etc.
The only passage at which I felt a little impatience was a digression about Dante, and that only 3 or four pages, easily forgiven especially since the author is Italian.
The subject was quantum gravity, which did not really exist and certainly wasn’t yet in the textbooks when I was in school.
The best part is the book was so persuasive it makes a whole lot of other books obsolete, for me, which is good because those books were much harder to understand. Supersymmetry, goodbye, it was good to know ya except I really didn’t and was never going to be able to.
The book is “Reality Is Not What It Seems” by Carlo Rovelli.
As for AI, I can honestly say that in the course of an indifferent public school education (not my fault), as well as a poorly conceived/chosen college curriculum (my fault) - I had my share of stupid teachers, but I would not choose cringe -inducing AI pablum over the least of them.
I really don’t know how people can stand it.
That's one hell of an endorsement for a book. Have you encountered the work of Joscha Bach? Reading the Amazon reviews, and then, yes, asking ChatGPT how Rovelli fits into an exploration I've been conducting with it regarding Bach, Wolfram and the views of computationalism and digital physics makes me think you would get a lot from engaging with their work.
Thinking it over, I believe in fact, I’m drawn to books that would resist summary without a good deal of loss.
As a result, I don’t read much of anything that is both current and about what is current.
And of course no current fiction.
At least for K-12 math, the combination of Socratic dialogue, gamification, and mastery learning with spaced repetition seems to make screen learning pretty damn effective for some kids. My anecdata on this is my son's positive experience with the Art of Problem Solving (and before that Beast Academy) self paced online courses, and my looking over his shoulder to get a sense of how they work. But it may not generalize to most kids or most subjects.
"Only a masochist would try to read a contemporary textbook straight through."
Depends on the textbook. For example, I read Sowell's Basic Economics straight through. I'm pretty sure many people have.
Yes, ok.
But in how many days?
I don't remember exactly but there were no long breaks. Something more than a couple days and less than two weeks.
You?
I read multiple parts in a row, but IIRC I didn’t go through it all in any one period.
I bought it for my daughter, who was reluctant to take an econ class at her very liberal arts college, but agreed to read this book when I “strongly” suggested she must do one or the other…
Re: screen learning. My company had excellent remote meeting capabilities well before COVID, but I insisted all meetings be in person for two reasons:
1. I found the folks who were just on a screen would be regularly and consistently not paying attention, whereas someone not paying attention in a room with other people gets noticed and easily called out.
2. I found that the side discussions after the meeting were invaluable. Too many of my staff wouldn't seek each other out, but since there were already there together, there were plenty of "oh, I've been meaning to talk to you about..."
I've seen this with my high school aged kids as well. If they can skip paying attention because the lecture is on a screen, they will (even with their cameras on, it's easy to not pay attention). And when I ask them what they learned, it's usually something that involved their classmates.
So if I were overhauling college education, I'd try to get rid of as many of the multi-hundred person lectures classes as possible (where it is still possible to fall asleep in the back row without being chastised for it) and add as many 8-15 person in person discussion group classes as I could get away with.
Random Hypotheses:
1. Screen learning is a "thin" experience, leading one to be easily distracted, versus a "rich" (and respectful) classroom experience that has required resource expenditure to attend.
2. Serious reading requires an engaged brain or "flow" experience, versus the pablum of legacy media.
3. Vygotsky's zone of proximal development, which effectively promotes "flow" experiences, plays a role in here somewhere. For example, the older one gets, the more one seems to appreciate history, which happily tends to attract good writers/story-tellers.
Hi Dr. Kling. I suspect there are still live classes because tradition is powerful--students who go to college still expect to be in a room listening to a lecture. Also, having just written a book on modern approaches to learning and development, I have come to know that many college professors today employ a more interactive approach than in the past. They invite students to own their own learning far more than previous days.
I think it might be useful to differentiate between classes that focus on knowledge transfer and classes that use engagement and involvement tactics to optimize learning. The former can be recorded and replayed, as you suggest. The latter benefits from having a learned practitioner in the facilitator's seat.
Why do we read books? It strikes me that much depends on your purpose for reading. If all you want is the "key insights" then a 30-page summary might be all you need. However, the key insights I'm seeking might be different from yours, or from the conventional wisdom that an LLM will generate. If, instead, you're seeking to forge deep interconnections between the new material and what you already know, then 300 pages might not be enough. Similar considerations apply to gaining an understanding of historical nuance and detail, or to reading a classic text like The Wealth of Nations or Democracy in America in an exploratory fashion. With all that said, many newly-published books feel like padded articles to me, and I tend to read them using something like the Arnold Kling method: skip right to the conclusion, use the index to see if the book provides new insights on topics I'm already familiar with, etc.
I always learned most efficiently and completely by application of what I was reading. For me, it was a matter of simply doing every single problem a textbook included. This works very well for STEM textbooks of all kinds. Even in the age before the internet I didn't need the class lectures at all- not even a little bit. I suppose AI might have helped me in the study of subjects that are non-STEM such as literature, philosophy, history, and languages.
My prediction about the effect of AI on learning is that it will become a crutch rather than mechanized exo-skeleton for the vast majority of students.
Riffing on your oft-repeated "we choose what to believe by choosing who we believe", I would suggest books tend to be longer than absolutely necessary because the author has to convince the reader, who he knows is highly likely to be somebody unfamiliar with him, that he is somebody we should believe. That means a lot of otherwise extraneous information is included to demonstrate believability. A book, unless it is from an established author, doesn't have the institutional backing of a newspaper or other regularly published media that can provide a shortcut to establishing (or disproving, as the case may be) that believability. That doesn't mean authors can't also be disorganized, wordy, or poorly edited as well.
I don't think we'll get away from in-person face-to-face classes without super-immersible 3-D virtualization of ourselves as avatars. As you suggested, there is just too much information we communicate nonverbally from students to the instructor, among students, and of course from the teacher to make Zoom or any current chat platform a viable alternative.
I was going to suggest the same. The 10% we remember of a book is the conclusion. The 90% we forget is the evidence.
It's like a murder mystery. The conclusion and the clues leading to it are all that we remember. Well, there's the tone of the writing, whether we enjoyed the reading, but we don't remember that as the words themselves.
Two points on your wise questions raised in the essay:
“If we were to try to read ten 30-page pamphlets instead of one 300-page book, our minds would be overwhelmed and we would learn nothing.”
For analogy, scrolling and reading various essays on Substack is similar to reading the ten 30-page pamphlets. And perhaps the learning results might be alike. Or maybe not since a lot depends on reader comprehension and focus.
This brings me to second point about screen learning only, apart from human interaction. A small class, say less than ten humans as opposed to the large auditorium, seems more conducive and aligned with our evolutionary background for learning. A learned tutor (mentor) might suffice. But either of these likely will not pass financial muster in the current educational milieu.
"This raises the meta-learning question of why books are so long."
Depends on the subject and the time period.
My personal experience is that it's easy to be concise with people who are more or less friendly and open-minded and can be expected to have a sufficient intellectual foundation such that the "inferential distance" from their starting point to my claim is fairly low.
But it is impossible to be concise when facing a "hostile peer review" of critical adversaries always interpreting everything in as malicious a manner as possible, posturing as willfully obtuse, and looking to exploit every little mistake or poke any possible hole in every possible argument or point out any omission of detail. A lot of books - especially on controversial topics likely to get one in hot water unless phrased and supported just right - end up spending more time and effort on throat-clearing disclaimers and forestalling disagreement and larding things up with countless endnotes than they do on actually preparing and presenting the core of their content.
This is what happens in a modern adversarial trial, which is part of why trials are now so extremely burdensome and the records of cases are now so extremely long. The total length of text in all the concurrences and dissents in SCOTUS and appellate court opinions has got to be two orders of magnitude larger than it was a century ago. There is also the typical kind of too-clever word-twisting legal argument that has to be long by necessity because it is, at heart, completely incoherent nonsense. In the law it's easy to be coherent and brief and usually unnecessary to be coherent and crushingly comprehensive. You can often get away with incoherence buried under a mountain of confusing phrases, but it's hard to get away with the curtly incoherent. So, given that the law has been a total mess for a long time, you get a lot of LOOOONNNNNGGGG incoherent opinions.
Another reason books used to be long was because arguments were presented verbally instead of graphically, and a picture's worth a thousand words. Speaking of "The Wealth of Nations", the late P.J. O'Rourke read through the whole thing and in his attempt at summary made that exact observation, that to the modern impatient reader it was obvious that several pages of Smith's text could have been captured in a simple cartesian diagram or chart, but that was not easy given the publishing tech and costs of the time, and also, readers of that era were experienced in and used to the verbal, not so much the graphical, especially regarding topics distant from geometry and pure mathematics. NB: Internet comment tech also doesn't let me make arguments in pictures, so I'm forced to give you a thousand words, sorry.
For History, it's because history is really complicated and full of both mysteries and important details and and interactions, and adequately "setting the stage" for anything requires a lot of previous history. It is hard to narrow things down in a way that doesn't lose sight of the big picture and how all the puzzle pieces fit together, and it likewise hard for authors of history to avoid the procrustean exercise of forcing the messy facts to serve a narrative and fit into the presentation of a particular ideologically-flavored theme or lesson. You can read 50 huge and high quality books with mostly independent content about WWII and still get important new insights from the 51st.
On the other hand, some forms of communication have gotten lots shorter. It used to be something of a joke to point out the extreme duration of speeches given by current or aspiring Communist Dictators, but not too long before the Lenin-to-Castro era, comparably long political speeches and rounds of debates were not uncommon in the free western countries, where they were often well attended and with every word clung to by members of the general public in a way that Hartley's phrase "the past is a foreign country" does not even begin to express with adequate intensity. Most people probably think that change may have been for the best. On the other hand, those of us who watched it from the beginning have a feel for what the tendency toward more short text-messages and tweets did for the discourse, and in that regard I suspect that the average judgement about it is much less positive.