Strangely, your duds section has some of my favorite articles. I am fully braced and looking forward to more of them.
I'm also a big fan of Scott Alexander but somehow I could never finish Meditations on Moloch, even though I know it's probably his most famous and popular essay. Go figure.
_Meditations_ is incredibly long, I agree. I think perhaps partly because he also has a bit of "art" in there, which is not typical of his essays. There's a real feeling to it.
There's a way in which someone could in theory just have said to me, "Systems predestine outcomes." But reading a long essay about it, being walked through the reasoning and the examples, is helpful in landing the concept more permanently in my brain.
Funny enough, this is why I quit Blinkist. At first, I felt so "productive" because I blew through 6-page summaries of all the currently-talked-about nonfiction. But after a few months, I realized none of it influenced my life the way that, say, reading "Four Thousand Weeks" or "Die With Zero" did. Both of those books could essentially be summed up in a few paragraphs. But they've influenced my long-term thinking because of the way the concepts were landed more thoroughly.
Philip you definitely should not be writing to please an audience. Please just be yourself no matter who that is, I promise you it will be great, and also not great, because that's what being human is.
Philip, you don't have to prove anything to anyone. This is your publication, after all. So I would say, just keep writing about stuff you find interesting and want to share with the world. Or with your words: do not be captured by your audience to only produce what's popular.
While I usually enjoy reading about your thoughts on how the future of computer science (careers?) look like, I really liked your article and reason about the totally unrelated topic of the Fermi paradox. Or that it's better to watch series when they were continued after a few years. I think the mix so far is great.
> It’s a selfish act to foist writings about random unpopular things on people…
I disagree. We are not your captive audience. You do you, and a smaller or larger readership will be pulled, or not, by that specific post title/mail client blurb. The first part of the "impacting a reader" funnel, your thinking while writing, is worth it even if the funnel for a specific post ends up being extremely narrow..
Thanks for your writing, from an uncommitted reader that, more often than not, clicks/skips based on random stuff in my life than on the post's title, and that, when clicking, usually reaches the end of the post happy of having clicked.
I try to write at the intersection of what I care about and what people will find helpful. Fortunately neither category is precisely defined nor static. You can't escape the power law, though.
I really enjoy your writing, Philip and hope you keep it up. Many times I've found myself musing about a topic you wrote about, especially the broader ideas about how we'll adapt to technology both personally and societally.
I love reading your random thoughts Philip! I'm a big fan of people doing things just because they enjoy them. There's a certain kind of magic in that. I love to write too - I just don't publish anything. You've inspired me to post at least one article though. P.S. Nice use of the em-dash 😊
You've got to let me know when you start posting! Would love to follow.
The em-dash is something I've been doing for decades, but now feels shunned because of its association with AI. I still use them, though, because they're great! There should be a new type of em-dash that means "this is an em-dash that a human typed."
I’m reading Emerson’s The American Scholar at the moment, and his speech is full of em dashes. Yet I don’t feel that it makes his writing any less valuable.
What an irony. Nowadays, the moment people see an em dash, many immediately associate it with AI. I intentionally try to use em dashes less, not because I dislike them, but because I don’t want people to devalue what I’m trying to say simply because of a punctuation mark.
Can I ask you a quick question? Why is it that when you ask an LLM to generate an article, it almost always writes line by line instead of paragraph by paragraph? I’ve always found that interesting.
I think it's line-by-line (token by token) because the model depends on all previous tokens in order to generate the next one (based on how it was trained).
It appears line-by-line because, at least when I was at OpenAI, they made the UI change so it looks a bit less jagged to readers. But behind the scenes, over websocket/etc, it's really coming down token by token.
It completely makes sense! Thank you so much for explaining it to me. Keep up the great work! I genuinely enjoy reading your posts.
I subscribe to many publishers, but you're the only one I've come across who can write about such a wide variety of topics and write every one of them so well!
Strangely, your duds section has some of my favorite articles. I am fully braced and looking forward to more of them.
I'm also a big fan of Scott Alexander but somehow I could never finish Meditations on Moloch, even though I know it's probably his most famous and popular essay. Go figure.
_Meditations_ is incredibly long, I agree. I think perhaps partly because he also has a bit of "art" in there, which is not typical of his essays. There's a real feeling to it.
There's a way in which someone could in theory just have said to me, "Systems predestine outcomes." But reading a long essay about it, being walked through the reasoning and the examples, is helpful in landing the concept more permanently in my brain.
Funny enough, this is why I quit Blinkist. At first, I felt so "productive" because I blew through 6-page summaries of all the currently-talked-about nonfiction. But after a few months, I realized none of it influenced my life the way that, say, reading "Four Thousand Weeks" or "Die With Zero" did. Both of those books could essentially be summed up in a few paragraphs. But they've influenced my long-term thinking because of the way the concepts were landed more thoroughly.
Philip you definitely should not be writing to please an audience. Please just be yourself no matter who that is, I promise you it will be great, and also not great, because that's what being human is.
Philip, you don't have to prove anything to anyone. This is your publication, after all. So I would say, just keep writing about stuff you find interesting and want to share with the world. Or with your words: do not be captured by your audience to only produce what's popular.
While I usually enjoy reading about your thoughts on how the future of computer science (careers?) look like, I really liked your article and reason about the totally unrelated topic of the Fermi paradox. Or that it's better to watch series when they were continued after a few years. I think the mix so far is great.
> It’s a selfish act to foist writings about random unpopular things on people…
I disagree. We are not your captive audience. You do you, and a smaller or larger readership will be pulled, or not, by that specific post title/mail client blurb. The first part of the "impacting a reader" funnel, your thinking while writing, is worth it even if the funnel for a specific post ends up being extremely narrow..
Thanks for your writing, from an uncommitted reader that, more often than not, clicks/skips based on random stuff in my life than on the post's title, and that, when clicking, usually reaches the end of the post happy of having clicked.
I try to write at the intersection of what I care about and what people will find helpful. Fortunately neither category is precisely defined nor static. You can't escape the power law, though.
I really enjoy your writing, Philip and hope you keep it up. Many times I've found myself musing about a topic you wrote about, especially the broader ideas about how we'll adapt to technology both personally and societally.
I once dived deep into your lore at Facebook (I work there) and you’ve had such an interesting trajectory. Keep writing please!
I love reading your random thoughts Philip! I'm a big fan of people doing things just because they enjoy them. There's a certain kind of magic in that. I love to write too - I just don't publish anything. You've inspired me to post at least one article though. P.S. Nice use of the em-dash 😊
You've got to let me know when you start posting! Would love to follow.
The em-dash is something I've been doing for decades, but now feels shunned because of its association with AI. I still use them, though, because they're great! There should be a new type of em-dash that means "this is an em-dash that a human typed."
I’m reading Emerson’s The American Scholar at the moment, and his speech is full of em dashes. Yet I don’t feel that it makes his writing any less valuable.
What an irony. Nowadays, the moment people see an em dash, many immediately associate it with AI. I intentionally try to use em dashes less, not because I dislike them, but because I don’t want people to devalue what I’m trying to say simply because of a punctuation mark.
Can I ask you a quick question? Why is it that when you ask an LLM to generate an article, it almost always writes line by line instead of paragraph by paragraph? I’ve always found that interesting.
I think it's line-by-line (token by token) because the model depends on all previous tokens in order to generate the next one (based on how it was trained).
It appears line-by-line because, at least when I was at OpenAI, they made the UI change so it looks a bit less jagged to readers. But behind the scenes, over websocket/etc, it's really coming down token by token.
It completely makes sense! Thank you so much for explaining it to me. Keep up the great work! I genuinely enjoy reading your posts.
I subscribe to many publishers, but you're the only one I've come across who can write about such a wide variety of topics and write every one of them so well!