At the start of this week I read this post by Steven Sinofsky on tech going hardcore again in which he quotes an older article of his, from 2005, that refers to this 1989 book called Programmers At Work, whose subtitle reveals what’s inside: “Interviews With 19 Programmers Who Shaped the Computer Industry”. I had read Coders at Work many years ago, but didn’t know there was a spiritual predecessor to it.
So I immediately went and bought it used.
Then, a few days later, I re-discover a part of long-living file on my computer called “Mission Statement.md” that contains my Programming Principles. On line 237 I wrote the following: “The history of the software is as important as its future.”
Then, yesterday, the used copy of Programmers At Work arrives here. I open it and here’s the first paragraph I lay my eyes on, from the interview with Charles Simonyi:
INTERVIEWER: What was your first professional program?
SIMONYI: The first professional program that I wrote was a compiler for a very simple, FORTRAN-like, high-level language. I sold it to a state organization as an innovation and made a fair amount of money, none of which I ever spent, since I left Hungary soon after.
Way to start your career, eh?
I browse through the book and end up on this paragraph, from the interview with Andy Hertzfeld:
HERTZFELD: I found I had a talent for programming. A computer gives an amazing feeling of control and power to a kid. To think of something, and then get the computer to do what you thought of, was such a great feeling. It always has been. That's what attracted me to the field. Learning to program is like leaning to ride a bicycle; you can't read books about it. You have to do it.
For everything that’s changed since 1989, this hasn’t, has it? You can still think of something and get the computer to do what you thought of. And it’s still a great feeling.
From Sinofsky’s post: “Here we are in 2025, with all the companies having gone through layoffs, reduced benefits, and the vibe shift as some might say about focus on execution, delivering, and prioritization of important work. I think history will record that post-bubble era of perks and ‘Willy Wonka’ as the aberration and what we are seeing today as the best practice for innovation.”
Let’s start with the quote and you have to guess the name of the article: “Computer architecture isn’t telling a machine what to do. It’s establishing the possibility that it can be told anything at all. The work is superhuman, if not fully alien. Put it this way: If you found the exact place in a human being where matter becomes mind, where body becomes soul—a place that no scientist or philosopher or spiritual figure has found in 5,000 years of frantic searching—wouldn’t you tread carefully? One wrong move and everything goes silent.” No, sorry, your guess is wrong, the article is called Angelina Jolie Was Right About Computers and it’s about RISC-V. Before reading, I knew what RISC was and I had heard about RISC-V, but didn’t really have a clue. The article put some meat on those acronymical bones.
ThePrimeagen was on the Lex Fridman podcast and they spoke for at least 5h20m, which is quite a lot and also why I haven’t listened to the whole thing yet. But I do love Prime and it made me very happy to see that he’s been given such a platform. He’s one of the nicest, kindest people I’ve had the luck of meeting online. Now, that being said: I do love Lex’s view on programming with AI here, it’s exactly how I feel about it, and I’m prepared to also record a 5h20m podcast with Prime in which I make him see the beauty in all of this.
This is very, very good: A Field Guide to Rapidly Improving AI Products. Point two, about simple data viewers, is spot on in my experience. Best part is that it’s so easy to build them now.
As someone who once wanted to be a magazine writer just like this one, I loved Bryan Burrough’s review of Graydon Carter’s memoir about his 25-year run as Vanity Fair’s editor. It’s both a review of a memoir and a mini-memoir itself. It’s wonderful writing. While reading, I wrote in a note “urls!!” because this is one of the very, very, very rare articles that links to other articles it discusses. For example, it links to this 1999 article by Burrough’s about Bernard Arnault’s attempted take-over of Gucci with the great name “Gucci And Goliath”. If you’ve listened to the (fantastic) AcquiredFM episode on LVMH you know the backstory. But, back to Burrough’s article on Carter, isn’t this what working relationships should be about: “A connection of sorts was forged. After that, Graydon began calling me regularly; over the next fifteen years, probably three of every four stories I wrote were his ideas. I told myself, grandly, that I was ‘Graydon’s guy.’ I’m sure others thought something similar. His story ideas were simple. They often consisted of a single word. If, say, Rupert Murdoch was involved in something scandalous, he would call and say, ‘Wanna do Murdoch?’ We both knew what he meant, what I was to deliver.”
Surprisingly, the second Wired article I read this week that was also good and also about something I knew but not really: Inside arXiv — the Most Transformative Platform in All of Science. It has a very spicy paragraph in it that you need to read (you’ll know) and this line: “In 2021, the journal Nature declared arXiv one of the ‘10 computer codes that transformed science,’ praising its role in fostering scientific collaboration. (The article is behind a paywall—unlock it for $199 a year.)”
This article had its ten year anniversary this week, which is how I found it. Read it, please, I’m convinced you’ll be happy you did: I Played ‘The Boys Are Back in Town’ on a Bar Jukebox Until I Got Kicked Out.
While we’re on the topic of “you’ll be happy you read this”, let’s go back eleven years further into the past and read this Onion article from 2004 that should be carved into stone because every sentence (I swear: every sentence) in it is worth it: Fuck Everything, We’re Doing Five Blades.
I wish I understood half of this post on Scarcity and Abundance in 2025. The parts that I do understand (or, at least, think I do) are insightful: “If you try to put your finger on what exactly distinguishes the products that feel magical from those that feel like slop, you’ll notice something interesting. The pointless incumbent products – Genmoji or Text Summaries from Apple, ‘contextual’ anything from Google and Microsoft, that kind of stuff – express an ethos of maximizing the value of a software asset of some kind. There’s a whiff of attitude where “the codebase is the capital”, and the point of all these AI tools is to keep drilling for undiscovered value in the asset. Contrast this to everywhere I see people ravenously using new tools, like Cursor for coding, OpenAI Operator, or even fairly ‘basic’ uses like lawyers using NotebookLM to summarize case documents in a way they can listen to in the car. There is no concept of an ‘asset’ here; the value of the product to the user does not really depend on rich context, network effects, or some other obvious software incumbency. The software is just doing work, and the work is tangibly value-additive, even if it requires some human supervision.” I’m not sure whether I can really, clearly point out the distinction, but I think I can feel it?
Thanks to the previous link I found out that Dwarkesh Patel wrote a book. It’s called The Scaling Era and I want you to turn sound on and go to this landing page. Then scroll and zoom and click & rotate and… yes. I love it, man. So, I bought the Kindle version (even though I’m still reading The Power Broker) and here’s the intro: "There’s a Sherlock Holmes story that captures our relationship with large language models. A new client comes to Baker Street. With a single glance, Holmes rattles off the man’s life story: that he lived in China, that he is a Freemason, that he writes a lot. The client, astonished, asks how Holmes knows all this. In great detail, Holmes explains the series of deductions that led him to his conclusions. The client responds, ‘I thought at first that you had done something clever, but I see that there was nothing in it, after all.’”
“We forgot how not to spy and steal attention" — good reminder, even if a bit shallow in places.
This article on v8 “leaving the Sea of Nodes” is one of the best articles I’ve ever read on compiler IRs and, believe me, I tried to find nearly everyone I could a few years ago. When I was building my optimizing compiler, I read Cliff Click’s PhD thesis (Cliff Click, btw., is probably the greatest name in computers and the person who came up with Sea of Nodes) and couldn’t make a lot of sense of it. I read the other paper and, I don’t know, kinda gave up on it and then went with a CFG in SSA-form. Now, I’m wondering, how much my decision was affected by not having articles that are this clear — look at the examples! the graphs! — and, in general, how many technical decisions are made like this.
“If you’re new to tech – say, less than 5 years in the field – you should take career advice from people who’ve been in the industry more than 10-15 years with enormous skepticism.”
Very good post by Michael Lynch on How to Write Blog Posts that Developers Read. I want to highlight one line here and disagree with half of it: “You can also use free stock photos and AI-generated images, as they’re better than nothing, but they’re worse than anything else, including terrible MS Paint drawings.” No, please, do not use stock photos in your blog. Never in the history of blogging has someone stopped and thought to themselves “huh, wow, that’s a good stock photo choice here” and no one ever will.
Ben Thompson interviewed Sam Altman. I caught myself thinking “that’s not as batshit crazy sounding as it did a few years ago” when reading this: “Where I think there’s strategic edges, there’s building the giant Internet company. I think that should be a combination of several different key services. There’s probably three or four things on the order of ChatGPT, and you’ll want to buy one bundled subscription of all of those. You’ll want to be able to sign in with your personal AI that’s gotten to know you over your life, over your years to other services and use it there. There will be, I think, amazing new kinds of devices that are optimized for how you use an AGI. There will be new kinds of web browsers, there’ll be that whole cluster, someone is just going to build the valuable products around AI.”
Craig Mod launched a “a new members-only social network called ‘The Good Place’”: “It’s no exaggeration to say that using Claude Code to build The Good Place (and also a bunch of other small tools and projects) is one of the most astonishing computing experiences of my life. It’s difficult to articulate how utterly empowering a tool like Claude Code (paired with malleable software, open software, open systems (i.e., not iOS/iPadOS)) is for someone like me.” Read on for his thoughts on social media and his social network. It’s very worth it.
I have to admit: I’m a sucker for the word agency and I’m a sucker for anecdotes. No surprise then that I fell really hard for this, yes, website: highagency.com. Yes, it’s a single page that you can scroll through. How wonderful is that? That old saying of “this book should’ve been a blog post”? Well, this website is the opposite: it’s a website that knows exactly what it is and isn’t.
Bonus: this is a page from Programmers At Work and shows the wonderfully silly IconBounce program by Andy Hertzfeld:
And, yes, the appendix does contain the program:
