Watching video twice at 2x can benefit learning better than once at normal speed

The top 2 voted comments including yours made me understand for the first time that some people hearing others talk about "2x benefits" perceive it as silly superhuman braggadocio. (Maybe the "2x" does sound like Patrick Bateman's American Psycho morning routine: https://www.youtube.com/watch?v=RjKNbfA64EE)

Yep, I understand what your humor is trying to do.

As counterpoint, the blind have been using screen readers TTS (text-to-speech) at 2x to 3x (~300 wpm) for a long time but the masses didn't notice. (E.g. A blind programmer using TTS at 450 wpm which is about ~4x normal speaking speed: https://www.vincit.fi/en/software-development-450-words-per-...)

However, when popular websites like Youtube added "2x" speedup function to videos, the general public suddenly noticed how it made many lethargic and dull videos more bearable to watch. (Or it turns 20-minute videos that are too long and stay unwatched into more manageable 10-minute videos that are easier to digest.)

It doesn't take superhuman ability to comprehend many videos at 2x. A lot of us can just do what the blind have been doing with playback technology.

(As for those productivity maximizers who looking beyond 2x and are interested in superhuman 4x+ abilities, there may be evidence from blind people that the brain can be trained to understand extremely fast speech: https://www.scientificamerican.com/article/why-can-some-blin...)

https://hitchhikers.fandom.com/wiki/Disaster_Area

While some stuff that gets posted here veers into productivity cult status, I personally don't think this deserves to be lumped in with that. In spite of speeding it up, I probably spend the same amount of time watching the lecture as I would at normal speed. I do it because in my experience any concept that's novel or hard in a lecture won't be understood by me on the first viewing. So will require several attempts and usually reading something.

For many listeners, they suffer from speakers talking too slowly which causes them to tune out and become disengaged. This handicaps learning instead of enhances it.

A lot of people can read text at ~200 to 300 wpm. Since many speakers talk at ~100 wpm, accelerating videos to 2x or more just gets the audio in the same ~200 wpm range.

This reminds me of something that happened to me 20 years ago, and cemented my view that teaching should be bottom-up: waiting for our computer architecture lecturer to arrive for the day's lecture -- Interrupts -- I quipped to a friend "I hope he doesn't just walk in and say 'There's something called interrupts'". In a comedic bit of prophecy, the lecturer just then walks in, walks over to the board and says, "There's something called interrupts".

Today, if I were him, I would never start by defining the solution. I wouldn't even name it at first. I would first state the problem: we know that a traditional CPU can process only one instruction at a time. And last time we learned that it looks at the program counter at the end of each instruction to see what instruction needs to be executed next. Now if the CPU keeps breathlessly executing the instructions from memory, it's never going to have time to check if I/O devices have any data. How do we solve this problem?

A good student will start by suggesting that we have the program poll the I/O devices regularly. Then someone will likely mention the overhead of repeated polling, especially when when there are no inputs most of the time. This should eventually lead to the idea that the CPU itself (rather than the program) should check some sort of input signal between the execution of every instruction. An I/O device would set this input to an active state when it has input for the CPU, and the CPU will catch it immediately after the current instruction. Now at that point, there is no way to tell the CPU which set of instructions to run in response. Instead, we'll have to store those instructions in some specific location that is known to the CPU. Now what do you call this scheme? Interrupts.

I think this search for faster learning is fueled by the realization that we only have so many lines of text, or minutes of video or audio content, that we can consume in life. I don't want to waste that on ads or someone's idea of a fancy video transition. It's the reason why I don't like podcasts or radio programs: too much of it is spent on introductions or random subjects. I have other shit to do, so if I can maximize the retention in a shorter timespan, that's much better.

There's an ever-present desire to learn more, and the more you learn the more new topics you find out about. Personally I am a bit bothered. I bought a bunch of books over the years and recently I've realized maybe I won't be able to go through all of them at my current speed. So I'm trying to learn how to read faster (with retention) and be more methodical about it.

For things you need to go in depth the learning is necessarily slow: a college course, a textbook, years of experience. But if I want to learn something small that has less relevance to my life, I don't want to waste that time.

The speed of consuming the content was never the limiting factor.

When I listen to an interesting podcast, I often pause the track to think about implications of what was just said or to formulate a rebuttal in my head.

I guess there are different kinds of learning and different levels of understanding something.

The material is the base, sometimes a bad quality material can even reinforce practice and attention.

I remember how I learned serious programming at 14: a badly translated manual in French and a casio scientific calculator, no internet. I had to try each command, see what it did, guesstimate what it meant. I plowed through variables, goto/labels, printing, conditionals, had lots of problems conceptualising loops but figured it out talking with friends at school who struggled too, and ended up doing decent things with the calculator.

When I started C at uni, nothing was new (except pointers, I admit lol): the very good material didnt help my fellow students much, in fact I remember helping them a lot gain a more intuitive understanding on what programs were and to love programming for the sake of it, me who learned from sheer frustration and trial and error with basically rope and wood.

So I dunno... maybe I could have optimized my time but I still believe the sheer will to learn how machines ticked is what mattered, more than anything and that s always how I teach someone new to it: love the result and the process, independently of the details of implementation and you ll be able to program anything or at least know you CAN program anything eventually given enough courage and commitment.

And at work, I m the multi hat guy who code on all the systems, on all the languages, always coming in as a humble idiot but slowly gnawing at the problems until I become expert and people at it for years start asking me questions, because I just never give up and never think it's too hard.

But another point is that the presentation is not the main place where people learn, at least in math, which is the field I teach in. The main place people learn is in doing the exercises. I will pull a number out of my posterior and say it is 80% exercises (a matter of opinion, I concede, but what else can you do with opinions besides spread them around?). Obviously twiddling with video speeds doesn't apply at all to that.

I have to disagree here. There has never been a time in history where it has been easier to access high quality educational materials. See, for instance, MIT OpenCourseWare. Many other great universities have similar programs (Stanford, Berkeley). That, combined with the countless people putting out educational content on platforms like YouTube and Coursera, means that it’s basically possible for anyone with an internet connection to learn from some of the world’s best teachers.

Imo, speakers in educational videos tend talk very slowly. They do not talk in normal talking speed, they are slowed and speeding it up is basically moving them to normal speed.

In a more radical way I almost think playing a video in any nonlinear, noncontinuous fashion would have learning advantages because what matters isn't what you passively remember from the lecture but the amount of work you spend actively thinking about the material.

I often am thinking about the applications of knowledge during a lecture. I am currently listening to an economic history course and the last lecture I compared the foolish inefficiencies of 16th century mercantilism tarrifs to the new modern ones of the last few years.

Issues: first of all, transcript quality: those of TED (Chris Anderson's "Technology Entertainment Design" conferences) are curated, those of YT are oftentimes not. Then, discriminating significant and less significant keyframes, but if one only needs "decent" instead of "perfect", good compromises for heuristic algorithms (more "parametrized procedures" or "recipes") can surely be found.

You are tempting me... In an evening one should get something already usable. Better than a presentation, I see the transcript text with aside thumbnails linking to the full-sized images. As someone who sometimes studies material from video, when I am listening or watching I am not in the strict need for text analysis, and when I am working on the text itself I do not need the video. On the other hand, if I think of the MIT OpenCourseware material and similar (Yale etc.), the "blackboard" shots can be precious to have alongside the text.

--

  ¹  ffmpeg  -i video.mkv  -vf "select=eq(pict_type\,I)"  -vsync vfr  -frame_pts true  keyframes-%02d.jpg

the general rule is that you can read about twice as fast as you can understand spoken word - 300 vs 150 words per minute.

I remember reading this my first day in a public speaking class in college. Just one of those factoids that’s stuck with me, and why I’ve always preferred reading articles versus seeking videos.

Of course, speed of comprehension is just one consideration. I’m mindful that there are some contexts where a visual is necessary or particularly helpful, and that some folks resonate with visual/spoken word more than written.

But this is why I love HN. It’s almost always a link to a written article, and I can pop open comments (which I often do first) for more written text on the subject.

But sometimes the more efficient reading text instead of listening to a video is negated by not being able to multitask. E.g. The "dead" time of driving, walking a treadmill, raking leaves in the yard, etc can be filled by listening to text. Can't do that with a book.

That's why productivity can be helped with both complementary technologies:

- audio-to-text: auto-generated transcripts from video for max speed and random seeking

- text-to-audio: auto-generated TTS (text-to-speech) from text to multitask while performing a mindless physical activity. This helps get through backlog of books and articles without having to block out dedicated reading time while nothing else is getting done.

Have you watched many educational YouTube videos lately?

I used to also prefer reading for most things, but YouTube and advances in technology have just made educational/edutainment videos so much better than they used to be. There are some incredible videos out there.

Of course this depends on the subject, if I want to deep-dive on e.g. maths, of course only a Textbook will suffice.

In a simple world you'd just slow it back down to 1x when you get to this point, and you wouldn't need 2x, you could just skip forwards 10 seconds at a time like most people until you find it.

Unfortunately there tends to be fillers throughout the video, even inside of the sentences people speak. The real lifehack is to keep the high speed and then pause when the video said something important, so you can digest what you just heard, and resume - rinse and repeat.

A variant solution I prefer is to always be regulating the speed according to the load on your attention span. Videos change so much in info density from one minute to the next, it's nothing short of strange to watch at one set speed throughout. I think the fact it's so common to "give up" and "accept" a given video speed is an artifact of the fact that people have never had the experience of letting their subconscious finely control the speed with muscle-memory. Doing it keeps your attention at 100% usage all the time. Low info density, high speed. High info density, low speed. Use a browser addon that gives you hotkey control over the speed, or use ff2mpv to exploit mpv's native speed controls.

To be honest, I agree with you. I like to read. I learn better when I read. Text is much more convenient in so many ways that I honestly don't understand why would anyone opt for a video instead. But the reality is that the average attention span and the kinds of concentration people are capable of changes with time, and recently started changing a lot in relatively short amounts of time (1 generation). That's reality, and we have to learn to cope with it - even if it's painful - or we won't be able to teach anyone anything!

Wouldn't "epistemically toxic" be characterized by false, unsubstantiated or misleading content?

A documentary is video, as are recorded lectures, or random experts explaining anything.

A video allows us to learn and share knowledge without needing to be in the same room - the overhead of which detracts from the volume of material possible to teach or learn. The 1.5x is helpful when the speaker talks slowly, and makes sense that it's more effective, since your attention is more focused to keep up.

> They were told to watch the videos in full screen mode and not to pause them or take any notes.

It's basically another symptom of our culture's ever-increasing rush for things going faster/easier.

Speaking slowly is the norm in education in general. My parents are both teachers, I am not but I received basic training in education too, and "speak slowly" is always among the top advice. Maybe we need to review the rules for video you can control at will.