The only reasonable approach is to adjust your reading speed based on the topic you're reading. Is it interesting or relevant? Read slower. Is it dense and complex? Read slower.
Are you just scanning or reading for hobby/light interest, maybe hoping to encounter something interesting? Read a bit faster and slow down when needed.
I find that there are so many weird misconceptions about "how you are supposed to read a book". Just read however you like, reading comprehension and regulating your speed based on the content is a skill, something you might want to practice. But you definitely don't have to speed-read or slow-read everything once you've learned how to do that.
> reading comprehension and regulating your speed based on the content is a skill
Knowing how to modulate is an important skill in programming and mathematics as well.
I've seen the two extremes of "failure to modulate" with programs, especially with new programmers (students/interns). Some will speed read through the program, never slowing down to pay attention to critical pieces that may help them discover how to fix a bug or implement a missing feature. Others will slowly and methodically read every line, wasting away days on what should take hours or minutes by performing a depth-first search through each stack at each program point.
I've also seen this failure to modulate in mathematics. Students who get to chapter 5 of a dense text before realizing they have no intuitive or even formal understanding of the definitions on page 1. And students who spend their entire week interrogating unimportant definitions or lemmas because they do not get a sense for the overall structure of the chapter before diving into details.
Reading is such an important skill, and we as a society do a fairly good job at teaching how to read non-technical material. But reading skills are probably the most important thing that's (not) taught in math and CS.
Assuming that the student is a complete beginner. Do you think the best way to read a challenging technical book is to speed-read at the start - to pinpoint the essentials and get an overall picture - then read again, this time slower?
I don't think "speed reading" is necessary, but being able to read not-slowly and knowing how to effectively skim are both useful skills. Truly speed reading anything remotely technical -- especially mathematics -- is probably an exercise in futility (at least for me).
Quickly reading from beginning to end is useful, but only if you know why you're doing that and what you should be looking for. Examples:
- "this lemma gets used over and over in the proofs";
- "this theorem seems to be the main result of the text";
- "this function is where all the action is";
- "the application is organized around this framework"; etc.)
The only really effective technique I know is guided practice. From a teaching perspective, you can spend a lecture on "technical reading skills" like this:
1. assign a reading and make it very clear that this reading is non-optional, that the whole point of the next lecture is to talk about technical reading skills, and explain how huge an advantage students have in life if they know how to read technical content well.
2. hold a quiz that checks for comprehension (e.g., a proof that's a trivial result of a key theorem; a program that's a trivial modification of a key algorithm; etc. Basically, something that's trivial if you did the reading and understood the "main point" but hard enough that you're not going to be able to do it otherwise).
You can peer grade the quiz on the spot, or not, but at some point make sure you grade the quiz so that students who did poorly know that they need to work on reading.
3. Walk through the text on the projector, pointing out "oh look that idea from the algorithm got used here and here and here" or "and we're using that lemma again and again", etc.
You do this once or a couple of times and then invite struggling students to office hours for individual practice.
Reinforce this the rest of the semester. Force students to practice reading by holding (fiar and easy) quizzes on the assigned reading prior to covering the material in lecture. Respect students by going beyond the assigned reading in lecture. Help students improve by doing a good job at grading the quizzes, or by doing little exercises on the board that focus on trivial extensions of the "main idea" of the assigned reading.
Good examples of CS topics this works for are the pumping lemma and dynamic programming. Simple enough that there's a relatively short reading on the topic, but complicated enough that you can really test technical reading skills.
Taking notes and making a book summary seems to be the most efficient way. I used to do it informally but after taking inspiration from Derek Sivers I do it more formally [0].
I personally feel there are multiple "levels" or "layers" to understanding a book. It's a bit loose, but I'd say technical books have "structure", "content", "detail" and "cohesion".
The structure of the book is how chapters, section, subsections and the index are related to each other.
You can learn about the structure by taking a look at the index, reading chapter intros/outros/summaries, and you will get a better feeling for it while reading the book as well.
The content is basically just everything in the book except any meta-content, such as headings, the index and references, etc... So graphs, text, chapters, blablalba
The details of the book are the most relevant, precise, thought-out content, you could alternatively call these "topics". Consider a given chapter of any book, it usually introduces a topic, gives some examples, counterexamples, might tell a story about it, it could add a debate, show a graph, and it might describe sub-topics. The whole chapter is mostly about that topic as a whole anyway. That's the detail.
Lastly, the "cohesion" is how the book's topics are interrelated on a detailed level. For example, what implications does the content introduced in chapter 5 have on chapter 6? And on chapter 4? Some concepts take multiple chapters (or books!) to comprehend as a whole. Sometimes a book needs to introduce some concepts before the bigger picture can be explained.
This cohesion, in my mind, is related to the "larger than the sum of its parts" idea.
I guess the idea of cohesion can also be extended to how the book's content and ideas relate to what you already know, or other books you may or may not have read. Again; it's how the topics interrelate.
Different reading styles, writing exercises and re-readings of a book will give expand your understanding in those 4 layers.
For example, skimming (30 min, skip through entire book) will give you a vague but helpful foundational feeling for the structure and cohesion of the book. Likewise, speed-reading will give a reasonable foundational feeling for the content and detail of the book. Writing a mind-map of a chapter, asking questions about paragraphs, summarizing chapters will improve your understanding of the content and topics.
I personally feel that I often don't completely "understand" a book, I believe that if I take the time to work on those 4 layers of understanding on every level of the book, I would.
I strongly strongly strongly recommend taking a look at the book "how to read a book", the concepts I just stated are my own, but the book helped me completely rethink about how I approach reading in general. I used to be afraid of tacking huge books, I feel like I can actually read them with some confidence
See, I’ve been focusing on reading this year, more specifically to find out what my style and interests are, and how I can develop them further. For me it’s unrealistic to “just read away” - there are many different types of books and courtesy to our digital world/ genes/ whatever, my brain is definitely re-wired so that it’s very difficult to maintain focus for long. So naturally I’m interested in reading techniques (and other types of acquisition) and what others’ experiences are like.
But the deeper I go into this project, I am also beginning to wonder that perhaps, even self-proclaimed bookworms may not actually be reading as effectively as they think they are. By “effective”, I mean getting the same-ish level of clarity that the author had when writing it. And so I think you’re right about books having layers of meaning. Which this alone has powerful implications: it means a reading list should not be a ticking box exercise, as so many of us think, but a reference library - you have the freedom to read and revisit however you like, depending on your current intellectual needs. And it also means that books are actually very dynamic and so you shouldn’t really take them too seriously - it’s OK to get an understanding to a limited level only.
It’s interesting, let’s see how this reading project goes.
Anyway back to chomping dense technical books. This is definitely a big weakness of mine - no matter how interesting a book is, I just cannot stay motivated for long. But it may be because I’m approaching it the wrong way. I generally like to dip in and out, and be very selective about what to savour in. So I am probably treating technical books wrongly - taking it too lightly when each should be an individual project, really. I think another factor is confidence - it’s easy to get intimidated but I find that the more I become familiar with the concepts and basics, the tome gets easier. Which means it’s necessary to have a “learning pathway” of books e.g. Sal Khan’s circuit videos and “Electronica for Dummies” before the granddaddy “The Art of Electronics”.
Are you just scanning or reading for hobby/light interest, maybe hoping to encounter something interesting? Read a bit faster and slow down when needed.
I find that there are so many weird misconceptions about "how you are supposed to read a book". Just read however you like, reading comprehension and regulating your speed based on the content is a skill, something you might want to practice. But you definitely don't have to speed-read or slow-read everything once you've learned how to do that.