I will write code until the day I die. It is an art form which I embrace, alongside many other art forms I enjoy creating in like electronic music, writing, essays, travel, photography, and much more.
It is a terrible mistake that some people have come to believe that code is only as useful as its final executable result, and not the art of expressing logic and meaning within a computer system purely on its own merits.
Yes, same here. I've continued to write software all my life, and resisted nearly any kind of management position (except when design was also involved).
It's a creative process, and I've also occasionally focused on a personal project during holiday seasons and spent weeks in various cozy cafeteria corners with coffee (later lunch), a notebook and documents, writing code and staying in the flow for hours. Nothing is better for my brain than that (but of course not meaning _only_ doing that).
One point is that I can't see how, even with the most (not yet existing) AI to "write code" this can allow me to enter and stay in the flow, the mental state where nothing else exists than the creative flow in my mind, and hours fly by without me even noticing. If the AI is like a smart co-worker and my job is to sit there and explain.. one, there will be no flow, and two, there will be no fun. I'm not doing that.
You go! I don't think a lot of people believe that. No matter what, typing code will be around as a fantastic hobby and exercise. Just maybe not as a way to get paid.
There are developer jobs that amount to feature factories and there are reliability jobs about getting things done/keeping service. In this second camp, there is room.
> Chrome Extension Rewrite: Finally, I rewrote the element selector of the Chrome extension (the one that lets you select/pointer elements in the browser) the “traditional” way, by hand. Why? Because I realized that my struggles to explain it to AI were actually a problem of abstraction. Once I fixed that, the result was clean, solid, and perfectly ready for future iterations — this time with AI. I realized that getting these abstractions right was writing the spec. And that’s the kind of work architects do.
God, the fact that everyone and their mother think they need to have an opinion regarding "agentic" wankering is absurd.
There has been no discernible augmentation on a programmer's skills with these "tools". Zero, none, zilch.
Stop it.
You are making noise in a profession which a single mistake takes the "working" to "not working".
I swear the whole West's economic model is hyping bullshit to the uninitiated and make it seem plausible enough to either buy in whatever their selling or just plain self-satisfying "look at me I got a blog and writing code is challenging so here are some random ramblings about architecture springled with startupy vibes".
Petition to block "AI" stuff in this site for - at least - year.
> Today, AI types most of our code. We think, they type. Rather than diminishing our value, this shift amplifies it as thinkers — especially for those who love architecture
This way you get to experience the job of a senior principal solution architect: thinking about big ideas, and letting the engineering workforce build it and trying to make a square enter a hole…
Irony apart, been using on and off claude code for 3 months, tech is crazy already but… pretty sure there is no real acceleration (time spent dreaming and prompting count don’t get fooled), and the feeling of accomplishement to implement a feature is gone for me. So maybe i’d rather enjoy doing the tech myself and only use it as a very powerful stack overflow like q&a
While I see the lure of having an AI write the code completely, I feel like it is damaging those who come after us.
I'd much rather have junior developers follow my direction. That way I know they are learning, and can be more creative in how they approach their solution.
In the footnote, the author linkedly suggests to "Read That F* Code".
Indeed, it is good to be able to read F* code.
So not sure why the author does suggest to be Typer and Thinker at same time. Thinking in Types(and categories) composes well with traditional logical thinking imho.
Meh, sort of. Just because LLMs let you output reams of code, doesn’t mean you should use them to do that. As always, you should make the smallest diff that would accomplish your goal. Working this way, LLMs don’t really accelerate my workflow much except for work that’s truly boilerplate and for refactoring. But for the sort of small-ish changes that iterate towards product-market fit, I find I have to spend more time trying to get Claude to do what I want than just writing the code I need by hand.
Just today it gave me a bunch of deprecated MongoDB calls and completely botched some async Python code. But it's definitely gonna be writing all the code soon. Just six more months...
I’ve certainly found my bogus “5 hour limit” on the pro plan used up multiple times arguing with Claude about the simplest of concepts. So much in fact that I feel it’s by design to push users towards the Max plans… even if not true the fact that I think it at all is a loss for them.
We already know software companies are intentionally making their products shittier to drive profits. Google making their search worse to increase the number of times people have to search (so they see more ads) is a good example.
There is absolutely no reason to not think AI companies aren't doing the same. Dial in the accuracy so that each tier is only so useful, constantly and subtlety encouraging you to pay a little more for just a few more queries because "the next prompt will make it work, I'm sure this time!"
If you're not using Vim or Emacs, and not very proficient at touch typing, you're almost certainly limited by your typing speed.
Coding is an iterative process, regardless of whether you're handcrafting the code or using AI -- you need to move your thoughts / code / prompts from your head to the computer. You have to use the keyboard to do this. You have to do this over and over again, interleaving thinking with typing, and if you're fumbling for the mouse or smashing those arrow keys, your thinking is blocked.
I will write code until the day I die. It is an art form which I embrace, alongside many other art forms I enjoy creating in like electronic music, writing, essays, travel, photography, and much more.
It is a terrible mistake that some people have come to believe that code is only as useful as its final executable result, and not the art of expressing logic and meaning within a computer system purely on its own merits.
Yes, same here. I've continued to write software all my life, and resisted nearly any kind of management position (except when design was also involved). It's a creative process, and I've also occasionally focused on a personal project during holiday seasons and spent weeks in various cozy cafeteria corners with coffee (later lunch), a notebook and documents, writing code and staying in the flow for hours. Nothing is better for my brain than that (but of course not meaning _only_ doing that).
One point is that I can't see how, even with the most (not yet existing) AI to "write code" this can allow me to enter and stay in the flow, the mental state where nothing else exists than the creative flow in my mind, and hours fly by without me even noticing. If the AI is like a smart co-worker and my job is to sit there and explain.. one, there will be no flow, and two, there will be no fun. I'm not doing that.
You go! I don't think a lot of people believe that. No matter what, typing code will be around as a fantastic hobby and exercise. Just maybe not as a way to get paid.
There are developer jobs that amount to feature factories and there are reliability jobs about getting things done/keeping service. In this second camp, there is room.
I feel exactly the same way!
I love love love coding. When I'm done with work, and I have time between family stuff, I code to relax!
If I have a fun project, I could easily code from when I wake until I go to sleep, and have before having a kid.
It's a pure joy second only to family
I am reading the book "flow" at the moment and this echoes the authors sentiments.
Infact it feels like one of the few books to change my perspective on life.
Something tells me the author doesn't seem to find joy when writing code, and how writing the code can make the architecture more sound.
The author kind of gets there eventually:
> Chrome Extension Rewrite: Finally, I rewrote the element selector of the Chrome extension (the one that lets you select/pointer elements in the browser) the “traditional” way, by hand. Why? Because I realized that my struggles to explain it to AI were actually a problem of abstraction. Once I fixed that, the result was clean, solid, and perfectly ready for future iterations — this time with AI. I realized that getting these abstractions right was writing the spec. And that’s the kind of work architects do.
God, the fact that everyone and their mother think they need to have an opinion regarding "agentic" wankering is absurd.
There has been no discernible augmentation on a programmer's skills with these "tools". Zero, none, zilch.
Stop it.
You are making noise in a profession which a single mistake takes the "working" to "not working".
I swear the whole West's economic model is hyping bullshit to the uninitiated and make it seem plausible enough to either buy in whatever their selling or just plain self-satisfying "look at me I got a blog and writing code is challenging so here are some random ramblings about architecture springled with startupy vibes".
Petition to block "AI" stuff in this site for - at least - year.
> God, the fact that everyone and their mother think they need to have an opinion regarding "agentic" wankering is absurd
This would have been a great place to stop if you wanted to avoid accusations of hypocrisy.
If there's any real truth to be found, it's somewhere between your opinion and OP's, but your righteous dogmatic opposition is just as suspect.
Upvote posts you like, don't whine that posts you don't like shouldn't exist.
I agree but... I do want people to contribute incase folks who don't see what they are seeing, are missing something.
The problem is, said people, are not coming to the table with stuff of substance. Its just bold claims with nothing to back it up.
> Today, AI types most of our code. We think, they type. Rather than diminishing our value, this shift amplifies it as thinkers — especially for those who love architecture
This way you get to experience the job of a senior principal solution architect: thinking about big ideas, and letting the engineering workforce build it and trying to make a square enter a hole…
Irony apart, been using on and off claude code for 3 months, tech is crazy already but… pretty sure there is no real acceleration (time spent dreaming and prompting count don’t get fooled), and the feeling of accomplishement to implement a feature is gone for me. So maybe i’d rather enjoy doing the tech myself and only use it as a very powerful stack overflow like q&a
While I see the lure of having an AI write the code completely, I feel like it is damaging those who come after us.
I'd much rather have junior developers follow my direction. That way I know they are learning, and can be more creative in how they approach their solution.
In the footnote, the author linkedly suggests to "Read That F* Code".
Indeed, it is good to be able to read F* code.
So not sure why the author does suggest to be Typer and Thinker at same time. Thinking in Types(and categories) composes well with traditional logical thinking imho.
Meh, sort of. Just because LLMs let you output reams of code, doesn’t mean you should use them to do that. As always, you should make the smallest diff that would accomplish your goal. Working this way, LLMs don’t really accelerate my workflow much except for work that’s truly boilerplate and for refactoring. But for the sort of small-ish changes that iterate towards product-market fit, I find I have to spend more time trying to get Claude to do what I want than just writing the code I need by hand.
Just today it gave me a bunch of deprecated MongoDB calls and completely botched some async Python code. But it's definitely gonna be writing all the code soon. Just six more months...
I’ve certainly found my bogus “5 hour limit” on the pro plan used up multiple times arguing with Claude about the simplest of concepts. So much in fact that I feel it’s by design to push users towards the Max plans… even if not true the fact that I think it at all is a loss for them.
We already know software companies are intentionally making their products shittier to drive profits. Google making their search worse to increase the number of times people have to search (so they see more ads) is a good example.
There is absolutely no reason to not think AI companies aren't doing the same. Dial in the accuracy so that each tier is only so useful, constantly and subtlety encouraging you to pay a little more for just a few more queries because "the next prompt will make it work, I'm sure this time!"
If you're not using Vim or Emacs, and not very proficient at touch typing, you're almost certainly limited by your typing speed.
Coding is an iterative process, regardless of whether you're handcrafting the code or using AI -- you need to move your thoughts / code / prompts from your head to the computer. You have to use the keyboard to do this. You have to do this over and over again, interleaving thinking with typing, and if you're fumbling for the mouse or smashing those arrow keys, your thinking is blocked.
When I code, typing is about 10(?) ercent of the time spent.
That's true for most people
[dead]