The reason video is winning is because you can make a living on video advertising. It's not really possible in this day to make a living on writing, outside of specific niches. So people who are good at writing use that skill to make video scripts, not blogs or books.
Yup; I'd make the claim (as an internet commenter, not an expert) that audio / video is more aimed at passive entertainment, whereas reading and more importantly deep comprehension etc takes more effort and time, and it's harder to monetize.
Not impossible, mind - the author posted this on Substack which is a way that one can monetize writing (blogpost style articles anyway).
Amen. It's one real "downside" in this day and age is that it requires fairly undivided attention to be used... that aside, it's without question my favorite way to interact with information.
On that note, a big thank you to whoever added "read this page" to Safari on iOS! Being able to turn long form articles into ad-hoc podcasts has been a game changer for me.
It also fits in a handful of bytes or kilobytes what would take half a gigabyte to communicate in a video - sometimes making the difference if you have limited bandwidth or a cap on monthly traffic.
It's also ridiculously easy to cache (download a book in 9 seconds, board a transoceanic flight - no problem)
It also doesn't require the right sound and lighting conditions to see and understand a video (either those conditions, or good noise cancelling headphones - and now you're unaware of your surroundings)
It's also the only viable option on insanely low power devices which get months of battery life per charge.
It's also something you can read at an incredibly speedy pace if you are good at it and practice - though occasionally a decent audio/video player will be of use with this.
It's also something you can fall asleep while consuming, and tomorrow you won't have much trouble finding exactly where you left off.
> Text is searchable, skippable, scrollable, compact, transmissible, and accessible in a way that audio and video have never managed to be.
That's just a very long way of saying it's difficult to monetise; it's why audio and video are preferred by producers of content.
Few people are interested in disseminating an idea, a concept, anything... they are interested in levelling up their fame and followers. Text is typically no good for that.
Video keeps blowing up because people want to connect with humans, and life is making that harder than it needs to be, so people are settling for these weird parasocial echo chambers. With the rise of AI, all text is suspect, and authenticity is king.
Same thing if you swap "text" and "video". That's the point of different media - they differ along those dimensions. For example, "a picture is worth a thousand words" means that for some information it will be less compact to describe all the details of a video with words
While this is our course a good point, one extremely good part about text is that unless there given text is quite literally just plain text data, it's a lot easier to embed things like videos, pictures, audio, etc. into a textual medium especially when compared the other way around -- that is the fact that text in videos and pictures and so on tends to be quite limited when compared with the kind of "rich text" with the more audiovisual content added between blocks of text.
So one can use the thousand words of pictures while most content is textual, whereas the other way is significantly worse, since it of course lacks all the searchability et al.
I would do more video, but video editing is really difficult.
I think that today’s video influencers have gotten really good at “one take and done” recording.
I couldn’t do that. I’m way too much of a perfectionist. I always edit my text, and I’ve been writing all my life. I don’t think that I’ve ever written something perfectly, the first time (including HN comments. I tend to go back and edit for correctness and clarity).
A couple of weeks ago, I was interviewed for a podcast. The process was fascinating, and the woman that did it, obviously does a great deal of editing and refinement. I don’t know if I have that much patience.
Text is my favourite minimalistic medium. I keep a minimum eye on regular news through teletext and tech news via Slashdot and here because there are barely any distractions from the core content.
It's also very flexible in that I can immediately return to a previous sentence without needing intermediate steps like rewinding a video or audio format. I can copy parts into another document for reasons. It's easier to search.
This is also what makes learning from a book so much better than video (besides not needing batteries for it).
OCR exists, and the vast majority of new books are developed on computers and are available in a searchable and copyable format. Ebook software for research and collaboration is not as developed as software purely for linear reading, but there's no huge blockers.
Sure, LLMs can understand images and video, but when you make your program spit debug text you make it easier and faster for Claude Code to iterate on it and fix any problems.
See how much value does a text UI program like Claude Code provide, it really doesn't need anything else than cannot be done in a terminal.
> it really doesn't need anything else than cannot be done in a terminal
I strongly disagree with this.
Claude-code would be super-powered if it had a better grasp of running processes without logging output. Imagine if it could somehow directly trace running programs, spotting exceptions and gauging performance in real-time.
It would be super-powered if it could actually navigate around a code-base and refactor through language servers without having to edit files through search & replace.
Imagine if instead of code, the program was first compiled to an Abstract Syntax Tree and claude worked directly on that AST instead of code.
Never a misplaced semi-colon* or forgotten import directive.
It needs a fundamentally different model to an LLM to operate it, but I'm convinced that thinking that Text is the endgame is a form of blub.
It's where we are now, and it's working very well, but it shouldn't be considered the long term goal. We can do better.
* To be fair, this one hasn't been an issue for a while now.
For small games I work on I make sure claude (well, codex cli) can produce screenshots of whatever screen it's working on and evaluate them. It has some instructions on using codex exec (claude -p) to use a clean instance for evaluation, so it can pass a screenshot and description of expectation and get a pass/fail and description of the failure. The main agent can also just view the image but for things with a clear pass/fail I prefer it invoke a clean context.
Not all reading is the same. In other words, I wish this article had differentiated between different types of reading. For example, I read that many young adults have picked up reading "new adult" genre books. They enjoy the physical experience of an analog medium and consume one edition after another of popular series. This sounds fine at first, but the content is problematic. These books are not literature, and they may convey problematic views of behavior. For example, they may perpetuate outdated views of relationships between men and women, portraying them as unequal and reproducing clichéd stereotypes from the last millennium.
In short, the article focuses only on the amount of reading, but the content is also important. This should be part of the equation.
I see no reference to this in the article. Nor have you explained why these books are "not literature". This sounds like someone looking at a piece of art, and saying "that's not art".
As we're referencing young adults here, they already have a degree of understanding of the world today. Reading of the past, gives historical context to how the world is today, to why the world is as it is today. I'd have hoped they'd been well exposed to such things in school, and you can be absolutely sure they've been exposed to such things in movies, or music (have you heard some rap music?), or.. you know, this thing called the Internet.
In 12 seconds I can find more untoward content on the Internet, than I could in an entire library or book store.
Another advantage of text over the long-term: it is accessible for discussion.
Let us say that you want to analyze, say, drinking culture in Ireland. You could write documentary on it, or do a fictional character study. However, those require actors, camera equipment, editing tools and time, and it generally extremely expensive and time consuming. A quick TikTok video may be a bit cheaper than a full-scale film, but still needs some of that equipment and cinematography skills.
Music is not much better. You need skills in singing, translating ideas of rhythmic lyrics, as well as supplies for instruments.
Writing, however, is simple. At minimum, all you need is paper and skill in articulating ideas. Almost anyone worthy to rationally ponder a topic already has the skills to put it to paper (assuming that they have gone through a proper First-World education and know reading and writing).
Text is also one of the easiest to share. A picture is worth a thousand words, but that poses problems in sending all that information. Plain text, however (or even most rich-text formats) can be transferred to anyone over almost any protocol, even rudimentary ones such as word-of-mouth. Ideas shared through text can be sent at an unrivaled pace.
This is why I think all video content should have auto generated transcripts for various reasons; subtitles, auto translations, but more importantly index- and searchability.
You can't expect anyone to view 20 million videos a day to find trends in current day video discourse. In theory machines could do it, but it costs a fortune. But 20 million text transcripts? That's doable on someone's local machine.
As I counter claim to the one that today is more recorded than ever, one could suggest that these recordings are not guaranteed to last long, not even the span of one lifetime.
>> Books are disappearing from our culture, and so are our capacities for complex and rational thought.
are they? maybe it's a cultural thing or maybe the author's perspective is from 1st world countries. here where I live ppl can't stand reading books on digital devices (not counting tech bros in my N)
The reason video is winning is because you can make a living on video advertising. It's not really possible in this day to make a living on writing, outside of specific niches. So people who are good at writing use that skill to make video scripts, not blogs or books.
Yup; I'd make the claim (as an internet commenter, not an expert) that audio / video is more aimed at passive entertainment, whereas reading and more importantly deep comprehension etc takes more effort and time, and it's harder to monetize.
Not impossible, mind - the author posted this on Substack which is a way that one can monetize writing (blogpost style articles anyway).
Text is searchable, skippable, scrollable, compact, transmissible, and accessible in a way that audio and video have never managed to be.
Amen. It's one real "downside" in this day and age is that it requires fairly undivided attention to be used... that aside, it's without question my favorite way to interact with information.
On that note, a big thank you to whoever added "read this page" to Safari on iOS! Being able to turn long form articles into ad-hoc podcasts has been a game changer for me.
It also fits in a handful of bytes or kilobytes what would take half a gigabyte to communicate in a video - sometimes making the difference if you have limited bandwidth or a cap on monthly traffic.
It's also ridiculously easy to cache (download a book in 9 seconds, board a transoceanic flight - no problem)
It also doesn't require the right sound and lighting conditions to see and understand a video (either those conditions, or good noise cancelling headphones - and now you're unaware of your surroundings)
It's also the only viable option on insanely low power devices which get months of battery life per charge.
It's also something you can read at an incredibly speedy pace if you are good at it and practice - though occasionally a decent audio/video player will be of use with this.
It's also something you can fall asleep while consuming, and tomorrow you won't have much trouble finding exactly where you left off.
I could continue..
> Text is searchable, skippable, scrollable, compact, transmissible, and accessible in a way that audio and video have never managed to be.
That's just a very long way of saying it's difficult to monetise; it's why audio and video are preferred by producers of content.
Few people are interested in disseminating an idea, a concept, anything... they are interested in levelling up their fame and followers. Text is typically no good for that.
Video keeps blowing up because people want to connect with humans, and life is making that harder than it needs to be, so people are settling for these weird parasocial echo chambers. With the rise of AI, all text is suspect, and authenticity is king.
1 reply →
Same thing if you swap "text" and "video". That's the point of different media - they differ along those dimensions. For example, "a picture is worth a thousand words" means that for some information it will be less compact to describe all the details of a video with words
While this is our course a good point, one extremely good part about text is that unless there given text is quite literally just plain text data, it's a lot easier to embed things like videos, pictures, audio, etc. into a textual medium especially when compared the other way around -- that is the fact that text in videos and pictures and so on tends to be quite limited when compared with the kind of "rich text" with the more audiovisual content added between blocks of text.
So one can use the thousand words of pictures while most content is textual, whereas the other way is significantly worse, since it of course lacks all the searchability et al.
It’s also runnable (scripts), clickable (urls), and context-dependent, which makes it a nice UI.
I would do more video, but video editing is really difficult.
I think that today’s video influencers have gotten really good at “one take and done” recording.
I couldn’t do that. I’m way too much of a perfectionist. I always edit my text, and I’ve been writing all my life. I don’t think that I’ve ever written something perfectly, the first time (including HN comments. I tend to go back and edit for correctness and clarity).
A couple of weeks ago, I was interviewed for a podcast. The process was fascinating, and the woman that did it, obviously does a great deal of editing and refinement. I don’t know if I have that much patience.
Text is my favourite minimalistic medium. I keep a minimum eye on regular news through teletext and tech news via Slashdot and here because there are barely any distractions from the core content.
It's also very flexible in that I can immediately return to a previous sentence without needing intermediate steps like rewinding a video or audio format. I can copy parts into another document for reasons. It's easier to search. This is also what makes learning from a book so much better than video (besides not needing batteries for it).
But a book is neither searchable, nor easily copyable. :/
OCR exists, and the vast majority of new books are developed on computers and are available in a searchable and copyable format. Ebook software for research and collaboration is not as developed as software purely for linear reading, but there's no huge blockers.
1 reply →
I guess the latter depends on your standards for "ease" and the former your ability to find an "index"
Ebooks are
In an LLM world text will also be is king.
Sure, LLMs can understand images and video, but when you make your program spit debug text you make it easier and faster for Claude Code to iterate on it and fix any problems.
See how much value does a text UI program like Claude Code provide, it really doesn't need anything else than cannot be done in a terminal.
> it really doesn't need anything else than cannot be done in a terminal
I strongly disagree with this.
Claude-code would be super-powered if it had a better grasp of running processes without logging output. Imagine if it could somehow directly trace running programs, spotting exceptions and gauging performance in real-time.
It would be super-powered if it could actually navigate around a code-base and refactor through language servers without having to edit files through search & replace.
Imagine if instead of code, the program was first compiled to an Abstract Syntax Tree and claude worked directly on that AST instead of code.
Never a misplaced semi-colon* or forgotten import directive.
It needs a fundamentally different model to an LLM to operate it, but I'm convinced that thinking that Text is the endgame is a form of blub.
It's where we are now, and it's working very well, but it shouldn't be considered the long term goal. We can do better.
* To be fair, this one hasn't been an issue for a while now.
For small games I work on I make sure claude (well, codex cli) can produce screenshots of whatever screen it's working on and evaluate them. It has some instructions on using codex exec (claude -p) to use a clean instance for evaluation, so it can pass a screenshot and description of expectation and get a pass/fail and description of the failure. The main agent can also just view the image but for things with a clear pass/fail I prefer it invoke a clean context.
Not all reading is the same. In other words, I wish this article had differentiated between different types of reading. For example, I read that many young adults have picked up reading "new adult" genre books. They enjoy the physical experience of an analog medium and consume one edition after another of popular series. This sounds fine at first, but the content is problematic. These books are not literature, and they may convey problematic views of behavior. For example, they may perpetuate outdated views of relationships between men and women, portraying them as unequal and reproducing clichéd stereotypes from the last millennium.
In short, the article focuses only on the amount of reading, but the content is also important. This should be part of the equation.
I see no reference to this in the article. Nor have you explained why these books are "not literature". This sounds like someone looking at a piece of art, and saying "that's not art".
As we're referencing young adults here, they already have a degree of understanding of the world today. Reading of the past, gives historical context to how the world is today, to why the world is as it is today. I'd have hoped they'd been well exposed to such things in school, and you can be absolutely sure they've been exposed to such things in movies, or music (have you heard some rap music?), or.. you know, this thing called the Internet.
In 12 seconds I can find more untoward content on the Internet, than I could in an entire library or book store.
Another advantage of text over the long-term: it is accessible for discussion.
Let us say that you want to analyze, say, drinking culture in Ireland. You could write documentary on it, or do a fictional character study. However, those require actors, camera equipment, editing tools and time, and it generally extremely expensive and time consuming. A quick TikTok video may be a bit cheaper than a full-scale film, but still needs some of that equipment and cinematography skills.
Music is not much better. You need skills in singing, translating ideas of rhythmic lyrics, as well as supplies for instruments.
Writing, however, is simple. At minimum, all you need is paper and skill in articulating ideas. Almost anyone worthy to rationally ponder a topic already has the skills to put it to paper (assuming that they have gone through a proper First-World education and know reading and writing).
Text is also one of the easiest to share. A picture is worth a thousand words, but that poses problems in sending all that information. Plain text, however (or even most rich-text formats) can be transferred to anyone over almost any protocol, even rudimentary ones such as word-of-mouth. Ideas shared through text can be sent at an unrivaled pace.
This is why I think all video content should have auto generated transcripts for various reasons; subtitles, auto translations, but more importantly index- and searchability.
You can't expect anyone to view 20 million videos a day to find trends in current day video discourse. In theory machines could do it, but it costs a fortune. But 20 million text transcripts? That's doable on someone's local machine.
As I counter claim to the one that today is more recorded than ever, one could suggest that these recordings are not guaranteed to last long, not even the span of one lifetime.
>> Books are disappearing from our culture, and so are our capacities for complex and rational thought.
are they? maybe it's a cultural thing or maybe the author's perspective is from 1st world countries. here where I live ppl can't stand reading books on digital devices (not counting tech bros in my N)
It's just the introduction that highlights the "reading is doomed!" narrative, the rest of the article says it's not actually so bad.